Being an African born & bred in Nigeria, I particularly find this appalling. Captcha usually asks users to identify fire hydrants, crosswalks & other objects with design patterns that are not common here, in everyday living. There's a big question on whether these approaches work well for onboarding the next billion users to the internet, especially since the demographics are much different from existing internet users.
Fire hydrants are not at all common in European cityscape but you will learn about them from Hollywood movies or from music videos.
But in the part of Europe I am from, if I did not go to the daycare in a larger city, I would not have seen any crosswalks or traffic lights. I could have easily recognized a harvester or a railroad crossing though.
In the part of Europe I live in (London), if I did not go to any sites using reCaptcha, I would not know what a 'crosswalk' is.
Even the things I have seen (traffic lights, buses, 'fire hydrants') don't always ('fire hydrants' never) look the same as their American counterparts.
It's just a totally ill-conceived system for any global website.
London has plenty of pedestrian crossings, but they're unmarked, at least not in the same way - a stop line for cars before the lights, sure - except for 'zebra crossings', which look different, have different rules (no traffic lights and cars must give way to pedestrians) and would never be called a 'crosswalk'.
I suppose they just don't need to be, because not using one isn't 'jaywalking'. If there's too much traffic you identify a safe crossing by zebra stripes (or flashing beacon) or traffic lights or island.
When reCaptcha first asked me to identify the 'crosswalks', sure, I guessed it must be the road markings. But I just (still, to this day) select all the road markings that aren't obviously just centre direction seperating lines. I might get loads wrong, no idea.
Same here. They are usually marked with a yellow "H" sign about a foot tall, and the numbers on it tell the fire brigade where to find the hydrant if its not immediately below the sign.
>>There's a big question on whether these approaches work well for onboarding the next billion users to the internet
That isn't what captchas are used for. There are two side to every captcha, what the website is using them for and what the captcha creator or provider is using them for. In google's case, captchas are being used to hone AI. So it is no wonder that items like hydrants, store fronts, crosswalks and traffic lights dominate. Those are the objects that any AI-controlled vehicle will have to recognize. Given that the first market for AI-driven vehicles seems to be western nations, specifically the US, it is no surprise that google gears its captcha program to objects from that area.
If google were developing robots to pick fruit captchas would ask us to identify bananas and oranges. I worry about the day that captchas start asking me to differentiate people from other objects. I don't want to help train an AI to tell people from trees, soldiers from bushes.
> Given that the first market for AI-driven vehicles seems to be western nations, specifically the US, it is no surprise that google gears its captcha program to objects from that area.
This is a roundabout way of just saying the most obvious explanation (sites with US-centric captchas are presumably mostly interested in US users anyway) but wrapped up in an unnecessary conspiracy layer to add creepiness.
https://www.motorbiscuit.com/how-security-captchas-crowdsour...
(Not the best source perhaps, but I just picked one among dozens)
It's not a conspiracy theory, it's actually true that captchas are used as data for self-driving car algorithms, and that the data would be useless unless it was relevant to things those cars are likely to encounter. The reason they're so common is they are distributed for free, for the same reason: more data.
I often feel like I ought to get paid when presented with an annoyingly long queue of them.
Given Google tracks everything about me, there's no reason why they can't find-and-replace "crosswalk" with "zebra crossing" to match my locale. Its not a conspiracy, its just laziness on Google's part
That is one way of running an captcha: set a task that is not possible for a human. Anyone who answers correctly must be a bot.
"How many flowers in this field?" "Solve for X: 3567/134 = X" Anyone getting those correct in a reasonable time is probably not a human and so has failed the captcha.
That doesn't work, because either you need to randomize which way you ask (and then all a bot needs to do is try a couple of times to get the question where it needs to answer correctly), or you always need to ask one way or the other, which also doesn't work, because then the bot will answer incorrectly on purpose.
It would be different if the captcha would look at the incorrectness. I.e. the bot would now have to make the same mistakes that humans would make, which might be a workable path for captchas, however its unclear what benefit this would bring compared to answering correctly.
> So it is no wonder that items like hydrants, store fronts, crosswalks and traffic lights dominate. Those are the objects that any AI-controlled vehicle will have to recognize.
The point of the GP was, it's training these AIs on specifically _American_ hydrants, store fronts, crosswalks and traffic lights, which sucks not only for those of us unpaid mechanical Turks who aren't Yanks, but also makes you wonder if nobody cares that these coming AI-controlled vehicles won't be so great at discriminating NON-American hydrants, store fronts, crosswalks and traffic lights from _non-American pedestrians._
And no, America does not equal “the West”; most of your street furniture looks slightly“ off” compared to, say, that of most European countries. Anyway, the part of the “West” that will probably be at the forefront of this wave will probably, as usual, be the most hyper-advanced: Japan, Hong Kong, Singapore, Taiwan...
My prediction: This will lead to new users being turned away from established products from the anglosphere and develop their own alternatives.
Maybe after a while the big companies will notice and adapt, which might already be too late by then, or maybe they just buy the smaller ones directly, who knows.
I think this is already happening. Hopefully local entrepreneurs in their own countries can continue to build for their countrymen before people like Zuckerberg and Bezos get their hands in the metaphorical pot, and suck all the wealth right back to the US like economic hitmen of the last century.
Facebook is actually doing much better in this regard than Google/Amazon/etc..
For example, in certain developing countries "social media" is included for free with your phone plan, often there is no net neutrality and competition between carriers has forced all of them to offer this. As a result, local businesses are forced to have Facebook and Instagram pages, because their audience cannot visit a regular website on most days. Which ensures most people need the "real" internet less and less, and are less likely to pay for it, locking them into Facebook/Instagram/Whatsapp for a lot of their communication.
The alternative is that these people won’t have Internet access. Poor people have difficulties procuring even food, not to mention internet. Of course, as the economical conditions improve, the tradeoffs change and net neutrality becomes worth it.
Hold up, what does this mean, even? I thought "net neutrality" meant classifying ISPs in such a way that they are regulated by the FCC instead of the FTC. Obviously this is an American framing. How does that map to other countries?
Also, I was assured by users on this website and many others that if the ISPs were moved to be regulated by the FTC by evil Ajit Pai the internet as we know it would end. Those actions were taken by the administration three years ago, the media fell silent, and the internet is largely unchanged.
So what do you mean, in an international context, by net neutrality?
> Network neutrality, most commonly called net neutrality, is the principle that Internet service providers (ISPs) must treat all Internet communications equally, and not discriminate or charge differently based on user, content, website, platform, application, type of equipment, source address, destination address, or method of communication.
Think of it as more like Comcast shouldn't be allowed to charge you 100x per GB (in addition to being your only choice in internet) if your traffic is to/from any streaming provider while providing their own streaming service for free, or throttle your usage of "other streaming traffic" to sub-144p quality. [OK, in reality, every mobile carrier does this in the US]
In an international, developing country context, outside of some places like India with Jio's cheap plans, it is incredibly frequent that you are paying per MB or similar at rates that would be unaffordable otherwise. Facebook funds these things so that your usage of Facebook's apps, Whatsapp, Instagram do not count against quota and are free to use. Thus local competitors get no visits and don't run their own sites because users do not want to pay a day's wages to browse your menu when they could browse your menu for free on their FB page.
It has pros and cons, one thing being that they don't have to pay for data and they get access to some portions of the internet, which is good? But it also makes it completely impossible for a local player funded locally without the billions acquired elsewhere to start, because people quite literally cannot afford to use your service unless you can afford to pay for all of them to; you can't compete. And since it's their only choice, now they monopolize and capture an entire country's worth of communications from start to finish. You'll see whatsapp numbers in many countries in place of a phone number or website.
I use Lyft instead of Uber because Uber forced me to do so many traffic-related captchas to install(presumably with the aim of eliminating employees eventually), I felt like they were squeezing free labor from me. It's getting there.
Its not even the anglosphere, because I'm from English England and had no clue what an American crosswalk was until recently. This is something that affects non-americans, not just non-angles.
hCaptcha engineer here: this is an issue with Google's approach, not all visual challenges.
Our challenges have average solve accuracy within 1% when comparing users in major African countries vs US users in our user testing. Nigeria does slightly better than the US, in fact.
American here. I had to say a mailbox was a parking meter yesterday to satisfy one of these terrible things. Your measurement of success may be biased by user acceptance of incorrect information in the process.
And as a non-American, I neither know nor care what American parking meters look like in the first place. They're training their model on my model of their model
Considering how horrifically broken hCaptcha is, slapping your hand on the keyboard would have an equivalent solve rate. I've had hCaptchas where one of the images was virtually identical to the example and it still didn't count as one.
It's trickier than most people think. The term "crosswalk" is American English -- it's known as a pedestrian crossing or zebra crossing in other countries. (another oft confused word is "pavement", which means sidewalk in British English, but refers to the hard surface of the road in American English)
The markings also cause confusion. To the non-American mind, does "crosswalk" refer to the crossing with the two parallel lines or the one with the stripes? Whereas with "zebra crossing", there's no ambiguity.
You don't really have to know which squares are actual crosswalks or what a crosswalk really is, reCAPTCHA doesn't care and it thinks that any street with white striped lines is a crosswalk even if the lines are for indicating something else (which is also annoying because it'll ask you to verify you are human again if you don't check those non-crosswalks). So do as I do and select any street that has white striped lines (same goes for its more rare bicycle/motorbike, small truck/car or truck/bus mix-ups).
Yes, crosswalks are a marked place on a road where pedestrians are supposed to cross. They often have dedicated signaling to let you know when you have the right-of-way to cross. The signaling usually looks like this [1], though typically the other fingers on the hand aren't blacked out, so it's a full hand, rather than one showing a middle finger (I couldn't quickly find a better example).
The question of what a crosswalk is can be surprisingly contentious despite attempts at federal standards! In my city there is a situation where the police department and the transportation department currently define "crosswalk" differently, leading to some rather bizarre situations with police reports and attribution of fault (e.g. pedestrian crossing in something the transportation department considers to be a crosswalk but the police department does not, such as a marked bicycle crossing, because their definition is stricter).
Off topic: imgur showed me the image, then gave me a cookie popup and when I clicked decline, the image was replaced by the text “could not find image”
The GP poster knows what a pedestrian crossing is, what it's for, and how it looks in most countries that can afford to build them; they were just pointing out that the word “crosswalk” is only found in American English.
I assumed it was an enclosed bridge between two buildings, a skybridge or skywalk, but apparently they are black and white striped crossings that I know as "zebra crossings".
I wonder how many people in the U.S. would recognize what is instantly familiar to us in Uruguay, the Yerba Mate gourd (10 out of 10 Uruguayans would recognize)
Fire hydrants is the first thing I've remembered too. This is very much US thing, don't remember ever seeing it in other countries (maybe they exist, but not where I lived) - at least not in the form Google wants them.
My favourite is how it blocks me from doing anything because I use various cookie blockers. I hate recaptcha so much, because all I want is some goddamn privacy.
The USA is the home of the internet, just keep building yours up and hey, start a captcha of your own. Don't let google dominate.
Edit: This is serious, if you let them take everything over, they will.
What counts as a truck? Is a fire engine a truck, is a 4x4 (like a Landrover), a HGV (ie lorry), a box van?
What counts as a bus? Is a coach a bus, a minibus?
What counts as a bicycle? Is a pedal-and-pop (a pedal cycle with an engine)? The image shows part of a headset that might be on a bike, is that enough to call it a bike, might be a lightweight motorbike?
Is a hovercraft a boat? Is a helicopter an aeroplane (according to the people setting the question)? Do pedestrian facing lights count as traffic lights?
Then every now and again they add in an image that is, I assume, purposefully obscured (uniform noise) as if painted by an impressionist and titled "boring city with smog as viewed by someone with cataracts, there might be half a bicycle wheel in somewhere".
One question that I apparently answered wrong for years as a German are the traffic lights. In English, traffic light seems to refer only to the assembly of two or three lights used to control the flow of traffic. The equivalent German word most commonly refers to the entire installation, including the pole, button (for pedestrians and visually impaired), speaker (for the visually impaired) etc.
So I selected much more than what Google expected me to select, I selected all the cables and poles as well.
I'm pretty glad that hCaptcha at least gives me an example of what it's looking for
This always throws me off, if it says 'Select all cars' for example and there is a car going into the next part for 10% but the main thing of that square is a tree or whatever, do I mark that? But I also always select the whole pole, no idea if it's right
My goal is not to optimize google's AI training, it's to get access to the damn website. I think part of what it does is compare your answer with other people's, if the AI is not sure, so it's more like a Keynesian beauty contest where I'm trying to guess which tiles the average person would pick.
(And fail and then have to go through 5 more screens of this.)
I live in US (though not born here) and truck and bike ones give me a bit of a pause too. The fact that bike is both a bicycle and a motorcycle doesn't exactly help either. And btw is scooter a bike?
My heuristic (which seems to be pretty accurate) is to accept anything that's vaguely related to the object that they ask for. eg. if they ask for "bicycles" and the picture has a motorcycle I select it.
Are these examples of captchas that have failed you or just questions that come to mind? I ask myself the same sorts of questions but then haven't actually had captchas fail me for it much.
Sometimes the turbine housing, or a section of a stabiliser aerofoil looks quite aeroplaney. But yes, I assume I've resolved that one correctly -- just getting in to the flow of my rant!
Being an African born & bred in Nigeria, who comments on Hacker News, the assumption may be correct—but they're hardly representative of the average Nigerian.
I get them wrong more than I get them right. Half the time I don’t bother anymore and just think “well, another website I won’t be visiting”. I really do believe that a bot would be better at solving them than I am.
It would be racist and anti-Afro to suggest that either you are a bot or a scammer for failing all your captchas. Also it would be offensive and inappropriate. And it would be outside the scope of this topic, and it would be a decidedly ad hominem comment when you should be celebrated for standing up to such narrowly contrived captchas.
As an aside, how do you like American security questions? "What was the make and model of the car in which you had your first kiss?"
Security questions are completely pointless anyway and you should just always treat them as an extra set of passwords and store them in your password manager. I have changed from using randomly generated items to vaguely plausible sounding nonsense instead since sometimes you have to actually read these things back to a human.
As a blind person with english as not native language, the audio captchas are pure hell. I can't imagine other visually impaired people who don't know english at all and have to deal with a lazy website which dumps them the default english audio captcha if there is audio at all.
Hahaha, I love the fact that someone made a Captcha solver. Proof that if Google truly wanted to test for "humanity" they would design something different.
After so many, you're still going to get the "you're sending automated requests" message just like you do if you keep answering the quizzes correctly by hand.
ReCaptcha doesn't care at all if you're human or not, but instead has some vague aim of "preventing abuse."
Hi, I'm the author of the extension. You can find more installation options in the GitHub description. Make sure to also install the client app to simulate user interactions. Open the extension's options to get started and download the client app, then follow the guide linked below to finish the installation. I'm happy to chat on GitHub if you need further help.
If you're into anime, you might also enjoy Search by Image, it includes support for a handful of search engines tailored to anime and manga. Visit the extension's options to enable them.
Not sure if that's still a problem, but when I tried this addon some time ago, audio captchas were quickly disabled for me because of irregular activity from my network. It might not be a good idea to use this if you rely on audio captchas and can't easily change your IP address.
This thing with the network is an excuse. If I need to pass by most of the captchas, I just need to login in a GMail account. Somehow, this makes me a human according to Google.
At one point a vendor tried to claim that ReCAPTCHA was accessible because of the audio fallback. No-one in my team or at the vendor could actually pass it, so we blocked them from implementing it.
As someone with almost 100% visual impairment, I'm having a hard time dealing with CAPTCHAS. I find it more difficult to slove hCAPTCHA than Google reCaptcha.
Most of the times I can easily solve Google reCaptcha. However, If I don't know the meaning of the object being displayed / any difficulties seeing it, Google won't let me switch to audio mode ( It'll display something like "This feature is currently unavailable" or "We have detected unusual traffic from your network"
The error you're seeing is a false-positive, an it may be illegal for reCAPTCHA to deny you access.
I've rised the issue last year with the W3C [1] and now your experience is publicly documented [2], though the next step would be to file complaints and take legal action for your rights to be respected.
There was a CTF a while ago where someone beat Google Captcha by simply inputting the Audio captcha into Google speech recognition. It worked ~80% of the time.
I'm sure that this can't scale at least against Google. They can fingerprint their own files. Might be interesting to check with MS or IBM services as a backup.
reCAPTCHA’s audio CAPTCHAs (at least in English) are much saner these days, it’s a voice snippet from a YouTube video their AI had issues transcribing.
I understand the purpose of CAPTCHAs when used as a security measure, but recently it seems like there has been a huge proliferation of them in all sorts of random places and it’s becoming very annoying.
Even Google has taken to occasionally popping up CAPTCHAs just so I can see search results! (“suspicious activity on your network” - I’m using a reputable, top-tier UK mobile provider)
Is there grounds to legally challenge CAPTCHAs on the basis that they’re discriminatory? They must be a nightmare for those using assistive technologies, or who have forms of cognitive impairment.
> I’m using a reputable, top-tier UK mobile provider
I'm in the UK too and have found that If you use any mobile internet provider, and have a browser configuration that dumps cookies etc on close, you will get a google captcha every - single - fucking - time... I just end up tunneling everything through sshuttle to one of my servers to save time, but these days you also get a bunch of websites blocking IP address blocks assigned to datacenters, so you can't win :(
As far as I can tell all mobile internet IP addresses are black listed for google captcha - I assume because they are frequently rotated between users.
I've found this to be the case for Three, EE, and O2 so far (been testing them all out to find the fastest because fixed line internet always sucks around here). I presume it's not the same on a smart phone mostly because people don't have such aggressive browser cache/cookie etc purging configured.
Pretty sure this is at least somewhat related to Carrier-Grade NAT[1]. Influx of connections from a single public IP shared by multiple public users = 'security concern'. Not to forgive the big tech giant of its inflexibility :P
> Even Google has taken to occasionally popping up CAPTCHAs just so I can see search results!
The most annoying one for me recently has been stackoverflow. The most basic use of stackoverflow (other than arriving through a web search) is to use its own search bar to search for something. Every time I do so, it asks for a captcha. It would make some sense to ask for a captcha when asking or answering a question, but why ask for a captcha when searching? And it's always the first search after opening the browser (the browser is set up to clear cookies on close), so it's not some sort of rate limiting.
Search is restricted because it's an expensive operation and prone to abuse. This practice is not just found at Stack Exchange, now GitHub doesn't even display the full code search result to visitors. The only option is to log in (if you are already an established SE/GitHub user and doesn't mind the web tracking)...
That would be because they're trying to leverage browser fingerprinting, and you're screwing with the fingerprint.
The sad thing is, if you're privacy conscious and don't play ball with the rest of the world, and actively mitigate these surveillance/data leaking mechanisms, you're assumed to be a malicious actor by corporate risk departments the world over.
Welcome to the web. Give us your data, and enjoy being tracked.
In a world where bot herders try to blend in with the crowd, and historically lots of them aren't great at it, you have to be careful about your choices. It's very easy to look like a bot by accident, and odds are pretty good that anyone who sticks out is likely to be a bot. Just like odds are quite good that any HTTP request coming through Tor is quite likely to be malicious.
This is a pretty shitty scenario all around. It shouldn't have happened. It should never have come to this! That said, I've yet to see a better solution on offer for general-purpose low-effort anti-abuse purposes.
People here on HN generally seem to either hate CAPTCHAs with the passion of a million blazing suns (but have nothing better to offer) or want everyone to use something strictly worse (like argon2 hashing).
Well, exactly. And of course the IP addresses are dynamic. So it seems a bit silly to treat a whole network as suspicious because someone on it (amongst thousands or millions) is doing something suspicious...
I get the same thing on Wikipedia sometimes because someone, who at some point in the past had my randomly assigned IP address, once made some dodgy edits. (Solution: reboot router to get new IP address!)
I suppose if we had static IPv6 addresses (each account always gets assigned the same IP) then this would be less of an issue.
Companies need to stop treating IPs or groups of IPs like identifiers. When I'm in my office, I can't access yelp. They blocked that ip range for some reason, maybe because of something we did. We block entire countries at work because we were getting a lot of DDOS from those countries and we don't do business their but it ends up blocking a lot of legit use cases and wasting a lot of time.
I worked at a company once that decided to add an entire range of IP addresses to their firewall’s deny list. Seems an executive had their personal home network pwned, and an unknown figure told him it was from a certain part of the world.
Said executive made enough noise and put enough pressure on the network engineers to get it done without question or delay or...most critically..a reviewed and authorized change control ticket that would have caught what was about to happen
Following Monday help desk was underwater triaging tickets from the entire office full of team members we had just denied from the corporate network, and customers who now couldn’t use our product at all.
That's the point though, it's not just someone doing it that has triggered their defences, it's many - most probably unwittlingly because of malware or even some of those cheap shitty games that mysteriously steal your battery and/or data allocation with unknown background processes.
The point is thus: they are receiving enough bad traffic from that range to have determined they need to filter it somehow.
These companies are so determined to keep people from legally scraping their sites that they don't care that they're walking all over the ADA in doing so.
That is because >50% of internet traffic is bots. Mostly BS marketing tools, ad conversion scams and quack-security 'scans' from my experience. Pretty sure the entire captcha proliferation is to reduce fake clicks in google ads.
I've been working on ways to block bots from some websites I manage the hosting for. It is nearly impossible. I rate limit the hell out of most of the world and block a long list of fake user agents. Still like 40% of out traffic is bots.
>'legally challenge CAPTCHAs'
Probably not, depending on where you live etc. Here, in the netherlands only gov. websites are required to be accessible.
> “suspicious activity on your network” - I’m using a reputable, top-tier UK mobile provider
Same in Canada. WiFi is fine. Using the web when connected to cell data (from the second biggest mobile operator in the country) brings up CAPTCHA almost 100% of the time.
This reminds me of account security questions. They ask things like "what's your childhood nickname", "what's your first pet's name". To my Chinese mother-in-law, who was born in the middle of the Cultural Revolution, when food was scarce, everyone was poor, education was not easy to get, government was non-functional due to anarchy, and when her first priority was to survive, all those questions are just weird. They reek of first-world assumptions.
Then why are they framed as realistic questions? Sure, the most secure way of handling them might be what you suggested, but that is not how they designed to be answered. Security questions are created to verify your identity when password does not work (e.g. when they are forgotten). Answering them in unrelated ways just create another password that will also fail under the circumstances that security questions are needed.
The problem is that security questions are fundamentally flawed. Most of them are easily guessable with a little bit of research, and because they can often be used to bypass your password they're effectively a backdoor into your account. You're generally better off using them as either a backup password (that is, not guessable even given knowledge about you), or simply not using them at all. If you forgot your password then reset it via your e-mail account. In short, don't use security questions, they're fundamentally broken.
you are not supposed to type realistic answers to security questions
Assuming that you are allowed to type an answer.
United Airlines makes you pick from a list of pre-selected answers. And some of the questions are easily worthless.
For example, one of the United "security" questions is "What is your favorite ice cream?"
How is someone who is lactose intolerant supposed to answer that? Or someone who has never had ice cream? Or someone who doesn't like any of the flavors on the list? My favorite ice cream today isn't the same as it was 20 years ago when I signed up for my frequent flyer card.
United is probably the most idiotic of those. What's my favorite sport? Well, I'm not too much into sports but I'd say... Oh, it's not on the list. How about... Also not on the list. No wonder, there's hundreds of sports out there and United didn't pay their contractors enough to make a PhD level survey of sports. OK, baseball it is.
Good luck teaching my mother in law how to use a password manager.
Heck, I tried teaching her to write down her passwords in a notebook, and she either keeps forgetting to write down, or she loses the notebook completely.
I actually did that once. The rest of the security practices were so ridiculous, I set all five security questions to the same nonsensical answer. I had to go through a security flow with a support rep. at some later point, and he's asked me all five questions, each time I'm giving him exactly the same answer, and by the fifth thinking, "Really? Do you think I somehow won't get this one?"
Ha! That's a fair point. I'll award you with another security question story, then:
I called into support. This was not my first call, but it was the first call where I was prompted for the security question by the rep. And, there was no mention that it was a security question, just "So deathanatos, what color is the dress?" (Not the actual question, of course.) I was like, "uh, what?" at which point the rep. mentioned it was the security question. I'm already logged in, so I dug around the account settings, and found that under there, the security questions and their answers were visible, so I rattled off the answer at that point.
That particular method of support contact required you to log in, too, so anyone in contact could "answer" the security question by just looking it up…
I use a password generator for the answers and save them in my password manager. One time I had to read 20 random characters to someone on the phone as a response to my favorite food
If food was scarce like it was in the western world in the 30s people are more likely to remember a favourite food. My mother ate boiled onions everyday the one time she had a piece of bread would be extremely special and a memory to cherish.
Those of us living in the western world now would have a harded time picking a fav food because there is so much choice.
Your comment has a lot of first world assumptions about third world countries through the len of time without understanding your own history.
I hate captchas with a passion so I'll add another anecdote to the mix:
I'm currently staying in SEA and I love gaming cafe culture here. The only problem is that every time I go to one it takes me good 20 minutes to solve all of the captchas to connect to discord, spotify etc. before I can actually enjoy the experience. So often when I only have 2 hours to spare I really don't feel like spending 15% of that time doing slave labour for google for free instead of enjoying the social gaming experience I went there for.
Sure the cafe could be setup better with more IP addresses or something, it's a small niche scenario and there are probably some hacks around it but it shouldn't be this way - it's just so disgusting how the web got hijacked by this nasty invasion.
Unfortunately minority affected don't have big enough voice in this to bring any change.
not trying to stir up a shitstorm, but since this article mentions IQ tests - "what's a nickel?"...
Check out the history (1971) of Larry P. and California's use of IQ testing in schools.
> As a group, African Americans across the country scored lower on IQ tests. The lawsuit alleged that was because the tests were biased toward Eurocentric culture. Questions like, ”Who wrote Romeo and Juliet,” they argued, didn’t assess a student’s innate capacity to learn. It tested knowledge that some – and not others — had acquired at home or school.
Those aren't anything like the sort of questions involved in IQ tests. IQ tests rarely involve words at all, usually a series of geometric shapes where you're supposed to pick the next one.
Depends on the test. There's been a move away from tests that require cultural or language specific knowledge and which instead rely on "which shape in the sequence comes next" type questions but they have their own set of problems in that they don't capture linguistic ability which is a part of intelligence, and are also useless for blind people. Earlier IQ tests required a lot of culturally specific knowledge, and were often quite up front about it because they considered the possession of such knowledge to be a marker of intelligence. Generally that viewpoint is out of fashion now so test makers try to come up with tests that measure "pure" Intelligence, whatever that means.
Culturally specific knowledge obviously correlates with an ability to learn by the very definition of the ability to learn.
I doubt this viewpoint is out of fashion. The problem is that it is hard to compare people from different environments if you use knowledge that depends on the environment...
So does this mean Einstein's IQ was subjective? Was he supposedly smart because he had encountered things on the test before? Would he score lower on one today?
Not in 1971 when the lawsuit was filed. Now they use picture-based tests because those are supposedly less dependent on learned knowledge. (Though I find that many Raven's progressive matrices-type questions are a lot easier if you know about symmetry groups and XOR.)
> Though I find that many Raven's progressive matrices-type questions are a lot easier if you know about symmetry groups and XOR.
Yes, at that point people who have already developed certain mindware have an advantage, but arguably that's the point of an IQ test. You're not testing the potential of a person to eventually be really smart, you're testing for the current problem-solving potential.
Well, it is problem, because the assumption is that you're measuring a proxy for g that supposedly doesn't change. So if it can change, by something as simple as learning some boolean algebra, the IQ test isn't measuring what it is assumed to be measuring.
That being said, you're right that it would still be useful for measuring problem solving potential, but that's explicitly not what IQ tests are supposed to measure.
I personally think it's not something that's really true, but it's how I learned it in university and seems to be the default position of experts in the field.
This is of course assuming no traumatic brain injuries or the use of psychoactive substances.
This is very wrong. Most complete IQ tests contain a lot more than just progressive matrices, because there is such a thing as verbal IQ that can't always be measured well by progressive matrices.
> Questions like [..] didn’t assess a student’s innate capacity to learn
Back in the day, I did a boat-load of tests and passed a boat-load of exams at school and then university. I have a drawer full of certificates from (apparently) respected institutions to prove it.
None of which are any use to me right now.
I can't say that any of the tests or exams I sat actually assessed anything close to my "innate capacity to learn".
For immigration purposes, I've had to take (and pass) a German language test. I've taken so many tests in my life, over 25 years of "school", and I've learned how tests work and how to pass them, that I've made the joke that I could have passed the the written part of the test in any language where I could read the alphabet.
I don’t think reCAPTCHA was designed to reinforce American culture. I think it was simply designed to make ML training sets for their other projects, and they neglected to make it accessible.
No. Look up to so called "literacy tests" used to disenfranchise voters. The people who designed those tests, and the "IQ" tests knew exactly what they were doing.
Similarly, https://arxiv.org/abs/1110.1556 - the "Jewish Problems" that were extra-hard maths questions used to find an excuse to prevent certain students from getting into university.
I'm not American, and I agree there is a lot of information there, but it seems reasonably sane if you read through it.
I don't see any obvious contradictions. You can't park 11-1 tuesday due to street cleaning. You can't stop there mon-fri 7:30-16:00 with the exception of 7:30-8 for school business (dropping the kids off). You can park mon-fri between 4 and 6, once per day, per district (to pick up your kids after school), district permits (people who live there, not sure what the 4 means) are exempt.
Oh I absolutely agree. I'm not defending it, just pointing out that it's not totally insane. Here in the UK I've never seen anything that nuts however.
It doesn't matter if it seems sane that way. How many seconds does a driver have to read through that all when he's driving near the speed limit? Do we expect drivers to still keep an eye on the road while trying to decipher it? Do we expect every driver to slam on the brakes, and stand still until they've parsed it all?
I think the idea is you stop, read the sign (probably need to get out of the car), then either stay or go after parsing. Or if you're a regular you learn the rules that apply to you.
I assume the 4 means parking district four. They give people some limited room to hunt for street parking near their home by breaking the city into small districts.
I am American, and it confuses me, which perhaps proves the point.
In particular, why is it necessary to point out that parking is prohibited from 11 to 1 on Tuesday, when stopping of any kind is already prohibited on a superset of that time (7:30-4 M-F)?
Look at the bottom bolt on the "No Parking" sign: It's partially covered by the later-added "No Stopping" sign.
Also, because this is apparently in a school zone, you have to think about how stopping is different from parking: You don't have to pull into a parking zone to drop off your kid, you can just stop momentarily in the middle of the street. During the non-overlap hours, you can pull over and park, just not stop in the street.
That said, it’s not too bad. It’s long and the organization is bad but the policy is reasonableish.
There’s street cleaning one day a week (year round) 11am-1pm, and you can’t ever park then.
It’s a school zone as well; when school is in session you can park for 15 minutes in the afternoon to drop a kid off or for 1 hour to pick the kid up. When school is out you just need to avoid the street cleaning.
I rented a car there and I was amazed at how much text there was, instead of symbols. Eg they have signs in turning lanes that say "Right lane must turn right". By the time you've read that, you've turned right.
As an American cyclist who frequented SF… this gave me a good chuckle, thank you.
True story: Had a driver who swerved from the left-hand lane, across a lane of traffic, across the bike lane, and into a drive way, after I yelled at him (while rapidly braking) tell me "I didn't see you." I've got 3 reflectors (from his PoV), I'm wearing long sleeves reflective highlighter safety orange, and it's daylight…
Cyclists ride. They do not drive. Drive implies you aren't the prime mover in the mechanism (i.e., an engine is doing the work, and you're merely performing the task of executive signaling and control to that set of movers and systems carrying you along do so. The inflection point is actually legal in nature; in that once you slap a motor on it, you require licensure, because you're driving now. This is also why I disagree with driving being a privilege granted by the State; I'm not big on large edifi telling me I can only utilize mechanical devices at their whim, but such is life, yes, I've had the argument before, no I'm not going into it now).
A cyclist, however is the mechanism. You can make the argument that it's silly, but I'll bet you more people will look at you funny for saying cyclists drive than for saying the operator of a motor vehicle rides especially given the legal baggage that comes with driving.
Aside from that, they're quite easy to identify. They're usually the ones either breaking traffic laws, or being ignored/cursed at because their mere existance increases the volatility of motor vehicle traffic patterns in urban environments.
First of all, they've self identified as non-American and presumably not a native English speaker. You could have presented the difference between drive and ride with less pedantry.
Secondly, cities with large numbers of cyclists tend to have better, not worse traffic. It's obvious why: fewer people driving plus infrastructure for bikes is developed and separated from car infrastructure. You should wish for more cyclists, not fewer.
Unfortunately, you can't escape the need for pedantry as a native speaker, and as a non-native (which I sssumed they were), I generally find they appreciate th clarification.
Arguably, what makes motor traffic "flow better" is increased predictability, and decreased demand on shared infrastructure. I'm not condemning all cyclists or saying it isn't worthwhil; merely that the arrangements we've got with them being sometimes pedestrians/sometimes road users leads to a cultural clash that leaves everyone frustrated.
The entire urban network architecture leaves quite a lot to be desired.
I'm not one of those who ignores them. Quite to the contrary. I've been known to roll down the window and ask someone to please pull ahead of my A-pillar.
"Identify which country calls them vasistas because Germans used to ask "was ist das?" from a smaller aperture on the door before opening the actual door"
Depends on the part of the USA. We have tons of roundabouts in the part of the USA I live in. Everything from tiny ones on neighborhoods up to huge multi lane ones (though nothing as crazy as some of the ones I’ve seen in Britain).
Have driven all over the US, and while roundabouts (aka traffic circles) are more common in the northeast, they’re common enough all over that I doubt anyone would struggle with that captcha. /Maybe/ some people would trip over the wording (expecting traffic circle and unfamiliar with “roundabout”).
Tangentially, we've been having fun with this recently. "Why haven't you completed Active Shooter Training?" "I'm busy, prioritising my time, and this training is irrelevant to me." "But what if there's an Active Shooter in the office?" "I've been working at home since April, and I've already been told to expect 6 months more." "It could still happen at home!" "What country do you think we're in?"
Head Office seems to be really struggling to grasp that this training might be less relevant to other countries, other offices, let alone remote workers.
"Although training is currently optional, businesses and organizations are beginning to face citations due to non-compliance with OSHA's guidelines regarding Workplace Violence.[3][4]"
I'm supposed to take it every 6 months. Though they've got it whittled down to 15 depressing minutes of "don't help anyone and try not to be seen by the cops"
I used to have an english boss who used this retort frequently.
That was until I asked him how he pronounced the name of his quintessential made-in-England roadster. You know, his Caterham.
For those not in the know, it is pronounced Kate-rum. The second syllable is normally pronounced as if it should not exist, you move as quickly from the t sound to the m sound as you can. Same thing as Leicester (phonetically leh-str)
To be fair, the “h” was silent in UK English when the English colonized America, and then the English began pronouncing the “h” at a later point. UK English deviates from the common ancestor surprisingly frequently.
I remember a flame war between US-English and UK-English speakers, where one of the former said “if it weren't for us, you'd be speaking German now” and one of the latter ones said “if it weren't for the French, you'd still speak English now”.
Like random3821 commented, it's about American English not being “proper” English (while supposedly if the US of A were a British colony, it would be “proper”.)
Apparently Europeans are bad at this too. Or do you think the French “socialist” party whose members have done things like lead the IMF is really socialist?
Depending on context, the word "socialism" is polysemic in practice in a lot of places in Europe. As long as people are aware that the S for "Socialista" in Spanish governing party PSOE is used in the sense of "social-democrat", and not in the same sense as the 2nd S in USSR standing for "Socialist", it's fine.
Indeed, you’re right. But isn’t this just restating my point?
Americans don’t reliably distinguish between various left-of-center ideologies in their language, but that’s not uniquely American, because neither do Europeans.
I've heard lots of Trump supporters saying Biden is Communist; he seems just right of center to me. But it's about "othering", and USA have had a decades long programme making Communists into evil bogey men, so it's understandable from that perspective.
The fraction of people in the US who genuinely believe that Biden is a communist must be in the single digit percent. The fraction that understand he isn’t but use that as an exaggeration for rhetorical effect is higher, but still not the majority. However, jokes like “let’s ask ‘Americans’ to distinguish these” implies that conflating Biden-ism and communism is a mainstream consensus opinion, which is absolutely wrong.
I see a lot of stuff about Reichsbürger and similar movements in German media; should we conclude anything from that about “what Germans believe” ?
Captchas have gone mad. The other day a major service gave me less than 10 seconds to solve a puzzle with a mouse in a maze and some cheese, and subsequently locked out of my account.
Actually before I got locked out, I thought I would stand more chance with the alternative for the visually impaired. It jumped straight in to a fuzzy voice reading 20+ numbers at a rate of more than 1 per second. I was already behind on typing them in before it started, and I failed that too.
Its as if solving an unfamiliar problem with fuzzy images/audio (that are increasingly fuzzed beyond the absurd) wasn't mad enough. But now I'm expected to be faster than a computer as well.
Imagine being faced with this problem. It's not even visually clear what's going on in the picture. That's before you even get to solving the puzzle. Just "identify what we mean by a mouse" would be enough to be the captcha.
And then -- and I don't know how everyone else's brain works -- but, being the simpleton that I am, it's a linear search checking each one against the condition. There's enough time to look at about 3 of them before the clock ran out. No, I really don't want more time, either.
I'm comforted by the comments in that Reddit post. I haven't gone crazy.
The second part of that actually seems to be more valid than you'd expect. Most users I've seen who actually use voiceover et al have TTS at a frankly astonishing rate.
Although I agree in general. Plenty of capatchas have failed me. I refuse to say I've failed the capatcha. Given I'm human, and its role is to detect humans; I can't fail to be human - it can only fail to detect it.
Yes, the American focus of the internet is absolutely appalling. Assumptions of timezones - PST, CST and the others whatever they are. Northern hemisphere assumptions of seasons - fall, etc - in the southern hemisphere the seasons are the opposite and we don't use the word fall.
Even Apple, known for its "sensitivity" to cultural matters has a big miss on this one.
In US and Canada, the time zones are named after geographical regions. From east to west, they go Atlantic (UTC-4 in standard time; services Nova Scotia, New Brunswick, Labrador), Eastern (US east cost, Canada's Windsor-Quebec main population center), Central (Chicago, St Louis, Texas), Mountain (Arizona, Wyoming, Colorado), Pacific (US/Canada west coast), Alaska, Hawaii.
Most US residents are going to be familiar with these time zones because you are more likely going to have to deal with multiple time zones in this set than any other international time zones. Television shows that announce to the broader country usually announce their times as, say, 11 Eastern/10 Central, because they'll be broadcast in both Eastern and Central time zones.
If I'm talking to someone in the US, I'm not going to bother prefixing the timezone with "US" because it's implied. But if I were talking to someone international, I'd be more likely to prefix "US", as in "this meeting is scheduled for 11:00 US Pacific time". This formulation has the added benefit of also describing whose daylight savings rule take effect in international settings.
It’s the western time zone where Denver, Salt Lake City, Santa Fe, and El Paso are. The region is extremely mountainous and high elevation, and hence is often called the “mountain west,” as opposed to the Pacific Coast (which has a lower elevation and mostly smaller mountains by comparison).
I think he already knows the answer, since googling this exact expression gives a very clear response. It looks more like an exclamation of "this doesn't mean anything to me", which I understand and partially share as well.
I agree nobody should have to know this, but on the other hand what harm does it do to learn it? In the US in elementary school I studied world capitals, major cities, history, etc. and came away with a pretty decent primer on things overseas, which I really value. Later in life I’ve come to know a fair bit about European football, about daily life in Taiwan, about the systems of government in Australia and New Zealand. The list goes on.
I’m not offended to learn about that stuff, I think it’s really interesting.
I agree that knowing the names of time zones shouldn’t be required, but I’ve only ever seen them displayed with their UTC offset beside them. When I get calendar invites for “AEST” I just look for that offset.
Just seems to me there are far worse things in the world to get worked up about /shrug.
I don't think it's about whether it does harm learning it: I think the problem is the relative barrier to entry that it creates for people who don't already know this information. Friction like this reduces social mobility and access to services for many.
As for your last point: that would apply to anything. There is _always_ something worse to get worked up about.
> but on the other hand what harm does it do to learn it?
None. In fact, I personally thought this was an interesting little fact to learn. I didn't get worked up about it, all I said is that you shouldn't have to know it.
However, because Americans aren't very culturally sensitive with respect to foreigners, there's lots of little things that we do have to learn to access the so-called global internet.
In middle school I often had to reset my plugged-in 12-hour alarm clock at night because I would switch plugs for some reason.
I noticed a pattern that sometimes my alarm would refuse to ring in the morning but would ring at 8 at night.
I eventually noticed that it only happened when I stayed up late and reset the time/alarm at midnight and figured out that my clock was broken because it thought that 11 PM is followed by 12 AM.
It was more than a decade until I found out that it was correct.
Even worse is the continued lack of UTF-8 use yet offering services outside of the US. In fact even Apple had non ASCII character issues on their dev website not so long ago.
Hey, as someone who does Quality Assurance, I want to point out that 99% of the time, those issues are brought up. They never get dealt with out the gate because business evolves via iterative value delivery, and most systems capable of creating value can work just fine with massively simplified assumptions to keep the lights on.
Also, if you think timezones/calendars/seasons are bad, just understand that many people specifically have a difficult time with shifting their mind into a frame of reference where you can actually talk about time in the sense of "there is a moment in time $now, where I am, the clock displays X because formatting, the system clock is actually Y, and I want people in P to get experience Q which had relationship R to X and S to Y. I have libraries T, U,
and V to work through the abstractive layers of. These same people are all inclined never to admit this difficulty or the complexity inherent thereto.
Throw in the fact most folks in the U.S. don't even understand the written script of more than a handful of places, and you have a recipe for !quality. Again though, there's only so much political capital you have as a Quality person, and prioritization tends to be the Project Manager's shtick.
Also throw in that the real optimization function in the market has been "the crappiest software that can still maintain usership" and it should be no surprise that the software packages that do localization right are far fewer in number than otherwise.
It's a rather sad state of affairs, but it is what it is.
Just... Wanted to point out there has been at least one person who brought up your points in design meetings everywhere they've been, and never ceases to end up receiving platitudes.
I just had a (probably stupid, but never mind) idea. Abolish time zones. Universal time everywhere in the world. 6 "hours" called red, green, blue, yellow, white and black. 100 minutes in an "hour", 144 seconds in a "minute". Night and day are different colours in different countries but so what? Someone would have to write a brand new "Falsehoods programmers believe about time" though.
What’s more fun is being American and intentionally misinterpreting the captcha. Select all street lights? Then think of how that question could be translated and misread in another language. Then select everything that would qualify as a light on the street: lamps, car headlights, crossing signals, et al.
Sometimes I select a square that I think the machine placed in the set to get more certainty but doesn’t match. For example, “select all chimneys”. If it presents me with an image of a house in the distance that has an object infront/behind it that is not a chimney but due to positioning and image quality it appears to be a chimney I will select it.
Captcha is a game.
- For Alphabet/Google the game is: mine the public for image recognition data so we can automate a car/drone and sell it to said public.
- For me the game is: can I play captcha chess with Google’s AI?
And if Google flags me as a bot, so what? At any point I know that I can quit playing captcha chess by giving it what it wants for a round or two and I’m trusted again.
To me it's not a game and not fun at all (I use Tor so I've seen more edge cases than most people). I found CAPTCHA is a school quiz. Some questions are ambiguous and open to interpretation, if it happens, it's not important to select the "correct" answer, but the "expected" answer.
- Select all squares with $object.
If $object only occupies 1/2 or 1/3 of a square, should I select it or not? The "correct" answer is selecting them, but Google marks it as a wrong answer. The "expected" answer is ignoring them.
When using Tor, Google switches the CAPTCHA to the more strict "9 pictures" (or even the "endless pictures with delay on each") version. The rule of "If there is none, click skip" no longer applies. In this version, when Google misidentifies something as the object and you are not selecting it, the answer is wrong because Google thinks you are not selecting all answers. You must guess which one is the misidentified object and select it.
- Select all squares with a traffic light.
If there's a road sign with a picture of traffic light, should I select it or not... Also, sometimes both the big traffic light and a smaller one in the next crosswalk are visible, should I select the smaller one or not? The expected answer is "no", but there are exceptions... Even worse, sometimes there will be 3+ traffic lights spread across a crosswalk, if I select them all, Google marks it as "wrong", sometimes it's unclear which one is the expected answer at all.
> When using Tor, Google switches the CAPTCHA to the more strict "9 pictures" (or even the "endless pictures with delay on each") version. The rule of "If there is none, click skip" no longer applies. In this version, when Google misidentifies something as the object and you are not selecting it, the answer is wrong because Google thinks you are not selecting all answers. You must guess which one is the misidentified object and select it.
Even using Firefox with an adblocker is enough to trigger this one. Worse still is when you then go on to fail whatever the rate limited action is (maybe a forgot password form) and you get sent back to the captcha on the same site in a short interval. Then you get the one with the images fading out and being replaced soo.... slow...ly.....
Literally, it takes 30s for an image to fade out and 30s for the next to fade back in and sometimes there can be chains of 3-4 images in the same slot taking up 2-3 minutes for the captcha process. Arguably at that point successfully completion is more likely to indicate ill intent as most actual users will give up and go away.
It's worse for Tor users. Even if I solve all puzzles carefully and correctly, Google still fails me, gives me another round of new endless CAPTCHAs, and fails me again, another round, and finally grants access if it wants to. It takes literally at least 5 minutes to pass it, 10-20 minutes is possible. And even if you pass Google, there are still two failure modes: (1) The website may still think you fail the CAPTCHA even when Google said you pass, seems like a bug or an expiration timer, then you'll have to solve another three rounds of nightmarish CAPTCHAs. (2) Google may ultimately fail you and also disqualify you from solving any CAPTCHAs because of "suspicious traffic from your network", and this denial is often only triggered AFTER you've already solved all the CAPTCHAs. I have no problem if Google blocks Tor completely, but now Google is effectively playing a bait-and-switch game on the users, I'd call it a form of human right abuse.
Solving Google reCAPTCHA via Tor is a prefect game for masochists like myself... Well, to be honest, the only reason I'm doing it, is to get true pseudonymous online accounts.
You will probably want to use a captcha-solver extension.
What you’re describing is what led me exactly to intentionally give bad results. When I give bad results, I'm not permanently denied entry. Instead, the data is sent back to Google and it notes that decision and I get to try again. If enough people were to always select all 3 squares, or an entire road that you can walk across rather than a painted crossing marks, or whatever false positive, then Google Captcha’s idea of a correct answer is more broad.
I can’t get rid of captcha. However, since it’s a data training game, I can try to eliminate the issues as much as possible by feeding it extra data so that people don’t have the 1/3 tiles question.
That question is difficult to answer because captcha has no rules. This is because it’s not actually used to identify if you’re human or a bot, that’s a side effect.
Don't get me started on how infuriating Google's ReCaptchas are on Tor Browser.
Buster captcha solver fails immediately. And when you do actually manually solve the captcha correctly, sometimes Google still thinks you're a robot.
And sometimes it asks me to select all "bicycles" in the images and there are NO GODDAMN BICYCLES in any of the images whatsover. So I hit Skip and eventually google thinks I'm a robot.
Everybody please either switch to hcaptcha or..., although I've not seen a website use this yet, upgrade to ReCaptcha v3 maybe?
I read somewhere that ReCaptcha v3 is far less annoying than v2? Is that true?
It's just that the classification gap between human and computer is closing, so now computers are better than a good percentage of humans, which means that there's a chunk of humans that now cannot conclusively prove they aren't bots.
Funny you should say that. reCAPTCHA v3 is meant to be invisible, watching the behavior of a user in the background, and then returning a score between 0.0 and 1.0 for the website to do with as they please.
so if you don't pass, say, by using TOR or even just a VPN, you don't even get a chance to prove you're not a bot? Cool, really cool, I can't see that being illegal at all.
EDIT: This comment is stupidly worded, please see clarification below.
My comment was poorly phrased. My point is that this can quickly lead to legal problems, when such a captcha restricts your access to something you do have a legal right to.
Say I buy a new phone and it stops working after 3 weeks. I want to contact the seller, but they only have a web contact form and I fail to get past the captcha.
This is already getting very close to being a huge legal problem, and the only reason why it's probably not that bad is that technically that website is supposed to give at least a contact address so I could just send them a letter which, while hugely inconvenient, is legally still a good enough way for me to contact them.
I'm sure if you spend a while longer you could find a few cases where the problem is even worse (also considering more than just one legal system).
So what I should have said: Using this technology to block users from the wrong endpoints could end up being illegal.
It's definitely a problem regarding government sites - the UK even has a page advising their own government agencies to not rely on them unless absolutely necessary[0]. US government sites have simply resorted to putting everything behind a login wall and verifying identity, such as is the case for ssa.gov and healthcare.gov, or just not caring and letting spammers do whatever they'd like (typical for local-level governments).
GDPR could be a concern if you are not able to opt out of automated processing. Deciding if you're a spambot or not is a huge decision for a robot to make.
To be fair, any captcha would have this trouble, but an entirely invisible one would not be able to be questioned
I'm pretty sure recatcha has a punitive/tarpit mode where it had already decided to reject you but just gives a series of slow tests which it will never pass you on anyway.
It seems like Google punishing you for taking actions to increase your privacy, which not-so-coincidentally reduces the data they're able to hoover up from you.
That's just silly; it's well known that Tor nodes are widely abused by spam bots. Google is "punishing" known sources of spam. When I ran a forum many Tor nodes ended up being outright blocked, since there was so much spam coming from them.
I'm not sure how to feel about this. I run Tor relays to support the network, but at the same time having moderated a forum that received tens of thousands of signups per day over Tor and NordVPN and PIA, all of which were able to verify email via various domains,, all of which were for spam...
It's not like I intentionally targetted Tor relays or anything, we just targetted source of spam, which ended up being a lot of Tor relays. Because we were (and actually, still are) running a rather outdated version of vBulletin (upgrading is difficult) things were getting pretty dire if we didn't do ... something.
ReCaptcha v3 is basically a means for Google to let the site owner take the blame for bad UX patterns while still getting browsing data from much of the web.
I'm not American and I often see captchas that ask about "cars" or "trucks". And they use similar images. I sometimes don't pass through those. My question is, when I see a truck and I'm being asked to select all cars, should I select the truck or not? For me a truck is a car. Is it not one?
That might depend on what 'truck' means to you! Is it equivalent to the British 'lorry', or does it include pickup trucks? I'm Australian, and for me 'truck' means 'lorry', pickups are a kind of ute, utes are a kind of car, and so trucks and cars are clearly distinct. (QED!) There seems to be endless room for cross-cultural ambiguity here though.
A side quest to yours - when I'm told to select traffic lights - should I include the pieces with just the poles or it's fine to select only the ones with actual lights? I never get through them.
The way I learned it, a "car" is any roofed 4+ wheeled vehicle up to van and SUV size (yeah, more Americanisms). Pickup trucks and larger are "trucks".
Even this exemplifies bad translations between two dialects of English. To me a van is a Ford Transit, but it sounds like to you a van is a Ford Galaxy (which I'd see as a "people carrier")
I think trucks are cars. I would pick any automobile, really even anything with 4+ wheels and a motor. I generally pass those.
The ones that get me are the street sign / street light ones. Do the poles count? What about a tiny edge of the sign that is barely visible in a neighboring square? Ugh.
The thing is that "probably" is the entire issue. You might just be training an AI that "truck" is a specific thing in the USA and something else where you happen to live ... or you might also be trying to guess what Google already thinks a "truck" is.
The stakes might be low (you get presented with another captcha) or astronomical (you get your account locked or something). There is literally no way to know with Google, it's a complete black box.
right but im sure whatever purpose it's using it for is going to be used against the commoners at some level -- weather it be to sell a new service at a higher price point, help further their monopoly, or to better create dark patterns (against us).
they have no "benefit of the doubt" -- it's pretty blatantly clear they are not on "our" side.
In my country, the Ford F150 truck would be referred to as a car or a "half-truck", while a semi-truck is what would be referred to as a truck. That's a piece of cultural difference that will dirty your training.
It might be called a car or a pickup truck but never just a truck by someone here. While "pickup truck" is an American inherited phrase which has truck in it, they're considered trucks about as much as a catfish is considered a cat.
The crosswalks Captcha provides the most difficult cultural discrepancy for me. Some US crosswalks look like European "forbidden for all traffic" road markings. Pretty much the opposite of a crosswalk. I always have to do a double take on those.
Zebrastreifen ("zebra stripes") is the German colloquial name. You will hardly find anybody (at least in my parts) calling them the proper official "Fußgängerüberweg" (pedestrian overpath/pedestrian crossing). Even most police will call them Zebrastreifen except sometimes when writing reports or in court.
I failed a few recaptchas like that too, tho not due to language barrier, but because I didn't pick some things that didn't look like crossings at all. I am convinced that some of those supposed markings were not crossings at all but things that somehow got into google's labeling db as outright false positives.
If you use a browser which blocks fingerprinting, CAPTCHA may fail you 3-4 times even if you give correct answers. CAPTCHA was giving me anger management issues last few weeks until I realized I can just open same URL in another browser and then most of the time it doesn't even ask for CAPTCHA because the fingerprinting is so accurate it knows who I am (or it lets me in after the first round).
Or just leave that website. Why should you be subjected to torment for the privilege of boosting some website's visitor numbers?
Imagine if, every time you went to the supermarket or the library, there was some goon on the door who wouldn't let you in unless you could do ten press-ups and answer half a dozen general knowledge questions. They'd soon go out of business --as all websites which treat their visitors with such contempt deserve to do.
Sadly, while it is extremely rare that a supermarket offers a product differentiated on anything but location, the same is not true of websites: not going through the CAPTCHA means I don't get to see the content I wanted or needed to see; I can't just drive five minutes further to get the same content from someone without a CAPTCHA. This thereby isn't a functional market and so market economics can't help us: this requires regulation.
There's very little content on the Internet that isn't available from a variety of different sources. It's very rare to find something completely original and unique to only one site.
I mean... add to that the fact that my ISP literally uses a single public IP address for all its subscribers, and then it's no wonder I hate this stuff. Googling anything in a private window is a gamble. This and Cloudflare. Cloudflare is the worst.
Check if you are logged into google. (pretty much unavoidable in chrome). If you have a google account then they know your phone number is verified and getting a phone number usually requires government mandated identification.
Otherwise they use fingerprinting to identify you and correlate it with activity on google, youtube, google ads, etc.
If those two things fail then Google will give you an endless stream of captchas even if you answer all of them correctly. I often had to complete 20 captchas only for it to tell me "sorry". There was maybe a 1/3 chance that completing all 20 captchas let me through.
Meanwhile on a different browser where I'm logged into google I get through on first attempt. Google captchas are just a data mining operation. They have nothing to do with proving whether you are human. After all, if they have enough data about you they often give you the option to skip the captcha altogether. The captcha is a psychological attack on you to convince you to give Google more data.
That's the absolute best outcome for Google. Not that they get a couple more data points for self-driving cars, or make the website owner happy they protected the site from bots, but that you switched from Firefox to Chrome.
My favourite is the Toucan crossing. They are a lot wider, with two sets of buttons at pedestrian height, and a cyclist height. Named as such because "two (types of road user) can" cross it.
As listed by others, popular crossing types in this group (the UK has a set of rules for how this should work and then local government can do paperwork to get something else appropriate if none of the existing designs are suitable)
Zebra has striped road markings, with flashing yellow marker lights so that vehicle users have plenty of advance warning of the crossing. Pedestrians have right of way on these crossings at all times, other road users should slow until they can discern whether any pedestrians are trying to cross and if necessary stop.
Pelican is an older design, though still used in new crossings in London and some other places. The pedestrians controls are in a box at hand height but the signals are on the far side of the crossing. These trigger road signals in a pattern with an extra flashing amber phase meaning "road vehicles may cross only if there are no pedestrians using the crossing".
Puffin is a replacement for Pelican with two innovations. Firstly the pedestrian signals are with your controls on your side of the road and they're placed so that to face towards them you're also facing oncoming traffic (this will most often be the right side of the crossing since the British drive on the left, but not in every situation). Secondly there is no flashing amber phase, the Puffin has infrared detectors so it can discern whether pedestrians are still on the crossing and extend the red phase slightly if they are.
The Toucan and Pegasus mentioned elsewhere are variants of Puffin rather than Pelican.
At least in Scandinavia, all road traffic (including bicycles) must give way to pedestrians at crosswalks, unless the crossing is controlled by traffic lights
The presence of a crosswalk generally gives pedestrians the right of way. At traffic lights, there are normally the usual car traffic lights (red, yellow and green), and then "pedestrian traffic lights" that face the crosswalks. The pedestrian traffic lights usually light up white in the shape of a person when you have the right of way to cross the crosswalk, or they show a red hand if you do not have the right of way.
For crosswalks not at an intersection (typically when there is several hundred feet of road without an intersection for crosswalks), I believe pedestrians always have the right of way. Sometimes they have a button you can push that will activate yellow lights overhead so drivers know that you're crossing (primarily useful at night). I think you have the right of way regardless, but given how little attention many people pay, I tend to treat them like I'm jaywalking anyways.
Sorry I assumed you were in the US, our (I'm also in the UK) zebra crossings do yes, but I was under the impression that they just marked any crossing (even at traffic lights) in the US.
But I'm basically only familiar with them from bloody reCaptchas, so I don't know!
In Colorado, pedestrians have right of way regardless of where they cross. Bikes and horses are treated as vehicles and entitled to the entire lane if they want. If the road is one lane, then uphill has right of way.
In Ireland we have zebra and pelican crossings. The general term used in the driver training literature is "pedestrian crossing." Most people would recognise the term "crosswalk" from American media, though.
unless they live in a country were American media is dubbed. Even then I'm not sure if you really recognize it, how often do I actually hear cross-walk in a movie or show?
Even in the US things are different from state to state, in California every uncontrolled intersection has 'crosswalks' where pedestrians have right of way, even if they are not painted in. Pedestrians are not supposed to cross in the middle of a block, unless there's a marked crossing (which is the only place you're likely to see an actual zebra crossing)
When I was young in NZ, they were zebra crossings. But then I noticed that they started more and more to be a pair of lines across the road than the zebra patterns, and the name seemed to gradually fall out of use.
In Australia (or at least in NSW) only crossings at traffic lights are marked with a pair of lines across the road. The majority of crossings not at traffic lights are zebra crossings.
So basically it seems that in almost every country in the world they're referred to as 'Zebra Crossings' except in the US. But, of course, we've all got to 'adopt' their terminology.
It's a bit like that idiotic back to front MM-DD-YYYY date format that no-one else uses outside America, but even non-US gadget manufacturers insist on making the default. I nearly sent my Xiaomi MiBand back when I realised it was hard-wired to display the date back to front. Luckily, with my middle-aged eyesight and the minuscule 3pt text it uses to display the date, I can't read the fecker anyway, so the annoyance is slightly diminished.
I am not sure what to make of this argument. We adopted English, the language from the British. The standard colonial oppression arguments already start from why use a foreign Language.
They used zebra crossing (even though there are no Zebras in the UK either, I think), so we used it. Now the new argument is that the cultural imperialism of the US is forcing us to change the term again. Sure, the term "cross-walk" sort of makes more sense than zebra crossing, but not enough that you don't need it defined for you the first time you encounter it. So, it is not such a precise and great term, as you almost say, that one can ignore the cultural imperialism argument.
Both arguments (colonialism and cultural imperialism) are distinct and can be treated separately.
We do have Zebra, have you never heard of a Zoo? (I jest; you are of course correct).
As for cross-walk making more sense; Zebra crossing are so named because of the black and white stripes they comprise of, they require you to have seen one, yes, but, I'd argue that without context "cross-walk" is no more descriptive.
Collectively, we call our crossings (Toucan, Pelican, Zebra, Pegasus), "road crossings". They are so named due to their properties; Toucan because "two (types of user) can" cross there, Pelican (formerly pelicon) because it stands for PEdestrian LIght C[O]Ntrolled crossing), Zebra for its stripes, and Pegasus due to the buttons being accessible at heights suitable for those riding horses.
Another one that I've seen on reCraptchas is "Boardwalk". If it had not been for the song "Under the boardwalk" I'd have had no idea what that was referring to. We call it a "Pier" round these parts. Taking Crosswalk as a reference, Boardwalk sounds like it should therefore refer to a place where you walk on some boards... such as where there are roadworks... or a wooden bridge... or outside the saloon in a Western.
In fairness, I don't think this is well known in the US. My impression was always that a boardwalk is just a pier that no longer has any use for naval stuff, so they built a carnival on it.
Pier is a more familiar term to me. I think I've only heard the term boardwalk in relation to some place in New Jersey.
Hmmm... "Pier" doesn't really have any naval connotations in Uk English. Somewhere ships [naval or otherwise] tie up would be a "Jetty". A pier would usually suggest what you're describing as a 'boardwalk' -- found in a seaside town, with cafés, amusements and other entertainments on it and lots of people strolling about eating ice cream.
A zebra crossing is a specific type of pedestrian crossing with specific rules different to those of other types. It needs a non-generic name (and look) so people understand how it works. Sure you could change your suggestion to something like "non-light controlled pedestrian right of way crossing" but I'm glad we say Zebra instead!
Yes, American fire hydrants are very different from NZ ones too. Ours are under a yellow steel panel flush with the ground, so not something self-driving cars need to identify.
U.K. firefighters connect to something similar, but they are unpainted metal panels on the ground, and while there are ways to distinguish them from the other ground-level utility access covers (and nearby signs) nobody bothered to teach them to me either in school or during driving lessons.
The two I've had trouble with are Traffic Lights and Parking meters.
Some of pictures are the kinda that hang suspended from a cable. We just don't have those in Australia so I tend to miss them.
Not sure I've ever seen a real parking meter. (Just ones in cartoons like the Simpsons). I kept getting mixed up with what was a parking meter and what was too ambiguous to tell. (Like could be intercom or a letter box, they pictures get blurry)
At least for humans. It's fairly easy to write a script with current technology even for an average "hacker" to solve them. But on the other hand it's extremely hard, nearly impossible for a person with special needs to complete them. Even for average person solving some of CAPTCHAs is a hassle.
Most of the popular CAPTCHAs services are "robot-friendly" and providers don't care who solves them, they just need data, they don't need to prove you're human.
CAPTCHAs being american-centric, ableist, etc are all valid criticisms here, but I think people in tech fail to understand the value of them.
Being "robot-friendly" still doesn't mean it's a walk in the park, and it's a hurdle that spammers will have to account for. If your site is a low value target or the spamming in general is low-value, it's often effective. Running any sort of small time open to the internet blog or forum will make the value of CAPTCHAs abundantly clear. The issue here is being inclusive to people, not making them "work" against robots.
In those situations, captchas don't offer any improvement over a text field that always asks you to put a 3 in it. It's just different enough from other systems that it requires custom programming, and that's just not worth it for most small websites.
There are user friendly and secure CAPTCHAs out there if implemented correctly.
A particular gaming forum uses game covers and asks users to select the correct title. This is quick enough to decipher for a human, while any OCR bots would have a difficult time with it. They could build a database of all covers, but it would be difficult to keep it up to date, and site owners could simply add more variation, so I doubt any bot authors bother to do so.
Google's CAPTCHA is a cancer of the internet. We're all training their AI without any renumeration. I genuinely hope that some government figures out how to sue them. I've sat down for minutes, repeating CAPTCHAs over and over again, just to log in to an account or download something.
It is not without remuneration. We use CAPTCHA to accessing a service. Imagine the "worst case scenario": you don't use Google services but a paid service has Google's CAPTCHA. It may seem like you are paying for training Google's AIs and you get nothing in return... but:
- Bots cause trouble to the service
- The service has to find some mitigation
- Mitigations cost money, and that cost will be passed down to the customer in some way
- Google offers a free solution, maybe not the best, but it will not cost the service anything (or close)
- So you are training Google's AI in exchange for cheaper service
Of course, you may disagree with it, you may think that a paid service shouldn't use Google's CAPTCHA, that you already pay too much, etc... But you are free to go elsewhere, and the rest is just market considerations, not something governments generally mess with, at least not governments that support free markets.
You could do lawsuits based on specific terms and conditions, but I guess large companies, Google first, have lawyers who know they stuff and get the company covered.
Unfortunately the developers of those particular things (often like ad developers) use obscurity of things like div names to make the task less trivial.
>Not anymore, firefox has it by default now, just gotta turn a setting on.
I don't use Firefox as my main browser. I use Yandex. I do sometimes wonder if HTTPS everywhere is becoming a bit redundant these days anyway. Most sites seem to have moved to HTTPS these days.
>Sadly this [uMatrix] is dead
Yes, that's a shame. It still works though. Hopefully whoever takes on the task of keeping it alive via a fork will make it a bit more mobile friendly. [The same could be said about uBlock Origin]
No, the owner of the site is getting a free service from Google to try and prevent bots from using their site. If you don't like CAPTCHAs then your problem is with the site owner, not google.
Google's CAPTCHA is specifically engineered to gaslight you though.
Often it will ask you to identify fire hydrants or bicycles, and despite selecting all of them (and it being easy to recognise them as they are fairly unique objects), it will still give you an error, and give you a new set of images to try again. If it is feeling really malevolent it will even give you a third round.
Hell, it happens regularly on the second and especially the third try it will give you excruciatingly slowly crossfading images, and the last image routinely seems to need 2-4 clicks to finally make the object it wants you to identify go away.
Hcaptcha (Cloudflare's alternative) is an absolutely breeze by comparison, and it even allows you to pre-generate a bunch of tokens via an extension (PrivacyPass).
Ok, apart of the fact hackernews just made me fill in a captcha to register this one-time account (neverthelss, cudos HN, you are still better than the rest):
I do not understand Hcaptcha hype here. Since CloudFlare switched to Hcaptcha, as a VPN user, every time I run into a site that has CloudFlare / Hcaptcha, I can forget to pass that test. No matter how many times I try, CloudFlare still shows it all the time. It is so bad, so that now every time I see now Hcaptcha shown, I just close the site without even trying anymore. PrivacyPass :) - the biggest joke I seen on privacy - they that endorse it are either naive or have some other agenda.
These "free" zero-cost captchas for site owners are now in all places where a captca is not really needed. If CloudFlare, supper-dupper DOS solution relies on captchas, or if Google super-dupper search relies on captchas to get results, that demostrated how bad these companties really are at the main thing they do, and only want more data.
Even pirate stream sites ask now for captchas. It is so easy to integrate them, why not.
captchas, followed by 2-factor authentication with phone SMSs that we are forced to accept in all main life services slowly, combined with and laws to register phone SIM cards so they really know who we are, killed the web we know - people are being identified with hight quality.
We all rant about this here. It comes same in waves in HN all the time, but what can you do?
At least, if you own or manage a site or blog do not use: CloudFlare, Google, etc - no CDNs, no shared fonts.
There might be something really particular about your setup that trips up Hcaptcha.
I run Firefox with a stringent profile config, Ublocker in medium mode (3rd party default deny), I don’t care about cookies, Cookiebro, LocalCDN, Canvasblocker, Smart Referrer, on PIA VPN, and I never have issues with Hcaptcha.
As far as Privacy Pass goes: their code is on GitHub and you can verify the checksums. Doesn’t give me a 100% guarantee but it’s good enough for me.
I hate CAPTCHAs as much as the next person, just pointing out the it is a service that Google provides and no site is required to implement it. So if you don't like CAPTCHAs then your beef is with site owners.
The only good thing about hcaptcha - what's better than one page of images? - is it doesn't seem to care what you click, but then neither does Google if you're using Chrome.
I absolutely don't aim to defend Googles captcha here, but the other day I was setting up an Outlook account for my son, and I ran into Arkose Labs bot detection, and I actually got heatedly angry, which is extremely rare for me. I wanted to punch anyone related to that abomination.
Upon researching it seems to be used by the Epic launcher and Roblox (among others), which might explain why I've never encountered it before.
Someone else's screengrab, which looks larger than what I was presented on my laptop: https://imgur.com/a/jF1HxbN
So they:
* are very small (or I'm old)
* use faux 3d walls which further complicates the image
* have to be solved 10 in a row correctly
* have an unspecified time limit (which in my untempered rage felt like maybe 3 seconds per image tops, no promises)
* don't tell you you've failed by answer or time until you're through all 10.
I as a full grown human with ~25-30 years on the internet, as well as video games and puzzles for fun, could not get through it in less than 5 (*10) tries. I accept I might be occasionally slow, but this should not be an issue.
TL:DR; Can someone at Arkose Labs please just do an rm -rf /
Adding insult to injury, it seems pretty easy to write a quick image filter + path finding algo to solve these... as apparently all the walls have solid borders, while none of the walkable paths have them. So a targeted bot should have a much easier time solving these than a human.
Absolutely, they've produced something fairly consistent making it easy for bots, yet by design made it harder for humans to see (small, image noise), and solve (10 consecutive, short time limit, no user feedback).
I recognize that by looking at just the screengrab it's an extremely simple concept to solve, it's just that at every implementation turn they made the worst choices, and it just infuriates me.
Wow what the hell is that. Thank god I've never had to see that. It's better to load a broken image and ask to enter the numbers (that has happened to me though).
Oh, that actually reminds me of an entire other dimension of the whole thing. I blame rage induced fugue state.
What I described was just the procedure of 1 "level" of captcha. I had to complete either 2 or 3, the delineation is kind of blurred at this point.
The one I had before the above was audio based, but it failed to load a bunch of times, and failed my answers a few times as well, inexplicably.
It read out not 4-5, but 10-12 numbers, which honestly was manageable, but there was no audio spacing between the numbers that anyone who has to look at their keyboard to type would have to re-listen to it a few times to keep up. This one would also be entirely solvable by a bot, but problematic for a significant portion of humans.
I just don't understand how they make money, nor why Microsoft specifically would pay them for their services. I find the LEAST outrageous explanation that they're bribing someone in Microsoft's COTS purchasing.
But Google CAPTCHA has been asking the same questions now pretty much since its inception. Are we really still training it? Or is it just running on auto-pilot at this point? I'm guessing it's likely the latter.
I'm old enough to remember when reCAPTCHA was first introduced to help with deciphering text from OCR'd books. At that time, it didn't feel so bad answering those, as we were using our intelligence to genuinely help preserve our cultural history.
It then switched to being numbers on buildings and street signs, and it immediately felt worse - we were now doing a job for Google, and an annoying one. Mechanical Turk from Amazon was invented to do this kind of chore.
It's now creating training datasets for whatever else Google wants them to - it appears to mostly be for self-driving cars now, to identify landmarks and road signs.
It's definitely not on autopilot, and it's definitely a real problem.
There's no difference between helping google solve book scanning vs other datasets for whatever else they decide.
The fact that captcha exists on a website is the website owner's intension to cause friction for their users. It has nothing to do with google's use of reCapture directly. The only vote you have is to not use said website - sometimes harder said than done but that's the only option you have.
Every time I get a captcha with traffic lights I imagine a Google self driving car stopped at an intersection waiting for me to complete the captcha so it can figure out what it sees and move along :)
I used to have fun filling in those reCAPTCHAs with incorrect answers. They would show you two words, and it was always obvious which word was the actual CAPTCHA test and which was the OCR input, because the former would be warped into a funny shape while the latter was always a rectangular block of normal text. So I'd type in the correct answer for the warped word, and something like "fuckface" or "cocksucker" for the regular text, and it would be accepted.
I like to think that somewhere out there on Google Books, my efforts have resulted in an innocuous word being replaced with something offensive.
Before I discovered the Buster plugin, I used to try to be as unhelpful as possible in my forced Google AI training sessions.
I'd use the audio option in the reCraptcha and then see how little of what I heard I could get away with actually entering into the form. Often it sufficed to just type one word out of the entire audio clip or even enter a word which sounded similar [eg. audio clip says "tranced", I write "transit"]. The most satisfying ones of all where when, after listening to a complete sentence containing either word, I could pass the reCraptcha by simply typing "the" or "an" into the form.
One of the things that I find slightly disappointing about Buster is that it types in the entire sentence, when solving the reCraptcha. I hate to think Google are under the impression I've suddenly started trying harder!
>>I'm old enough to remember when reCAPTCHA was first introduced to help with deciphering text from OCR'd books
Am I the only one who always, on purpose, put in the wrong answer for the clearly scanned word? For me it was kind of a rebelion for being used this way, but it was always super easy to tell which word is generated and which one is scanned - and the algorithm only required the generated word to be correct, so I always put in absolute nonsense for the scanned word to break their OCR detection.
They do not ask the same questions since the beginning. At first, CAPCTHA has been used to make old books digitally accessible, which meant that you solved those potato quality scans of old books. When bots became advanced enough, reCAPTCHA was used to improve Google Maps by reading house numbers on poor quality cropped street view images.
Then the zeroCAPTCHA became a thing, which turned out to be even more annoying than others if you slightly cared about your privacy (meaning you at least installed an abblocken) and faced those multi-stages challenges. By looking at those challenges I think it is pretty obvious what CAPTCHA is used for nowadays, for training a recognition model for a self driving vehicle.
It would be interesting to see why someone would downvote it.
Since when are captchas a bigger issue than spam bots and fraud users?
and yes i also think that those tasks are actually helping our society. Adding models for self driving cars, for security/emergency breaking systems etc. So whats the issue?
I did not downvote you, but I assume it's because you're wrong. Depending on your browser settings, addons, etc. CAPTCHAs may be nearly impossible to solve. That is to say, you ARE solving them correctly, but Google thinks otherwise, and punishes you with ever more pictures that fade in and out ever more slowly. So the fact that they're easy to solve or work well for you means nothing. Other people are having real issues with it, and not because they are unable to solve them.
> It would be interesting to see why someone would downvote it.
If I had to guess:
> "solving one is not a real hurdle to you"
My experience is that it can be a real hurdle.
Sometimes reCaptcha forces me to complete 2 or 3 captchas (or more), some of them with an annoying artificial delay when loading new images. It seems to be worse when using a VPN or a browser like Firefox (or some privacy extension) that blocks Google tracking.
I understand why people use reCaptcha and that it works well for some users, but it can be a pain in the butt if you don't use Chrome, aren't logged in to your Google account, or if other devices in the network did something that Google doesn't like. I sometimes get captchas on my phone when using mobile data (on a major provider here in the UK) and Google search because of "unusual traffic from your computer network"...
I always used to wonder when I see a 'Pie' or other items specific to U.S./Western countries on re-CAPTCHA if people there would get it right if a Masala Dosai/ Idli (Common South Indian food) or Moimoi (Common Nigerian food) is shown.
Also it makes me wonder the success rate of people working in CAPTCHA farms(unfortunately), which I'm positive are not located in Western countries. Perhaps a large print outs of Pie, side walks are hanging on their walls and they just have to get it right couple of times to understand what it is.
P.S. I recently received an CAPTCHA on LinkedIn to identify a 'Spiral Galaxy', although that made me happy, the questions raised about CAPTCHAs seems validated with it.
And frankly, the ones in TFA aren't the worst. I'm familiar with yellow taxis from American films. I'm not familiar with 'crosswalks' (full stop) or American 'fire hydrants'.
Your buses also look somehow different often, and those can be hard. Traffic lights even more so, some are familiar, but some styles I would just never see in the UK, so I actually have to look at each square to check, I can't rely on recognising them.
Then, even once you get past all that, it'll frequently complain you haven't selected them all, and you have to guess which square that doesn't have a bicycle it thinks has a bicycle.
I don't actually find taxis as easy as most seem to. Here, that large sideways tub would usually be a pizza delivery. Taxis have a very specific roof adornment and that isn't it. I'd usually need some hint to mentally context-shift into TVland.
I love searching for very American-looking fire hydrants or pedestrian crossings. Great case of US cultural imperialism - I suspect that the developers didn't even think it might not be the same thing abroad.
Took a lot of years for American developers to figure out that string inputs are not limited to ASCII. And even that is more due to emojis (bless them). It's a slow process.
It always amazes me that we as a society insist on using ambiguous words ("crosswalks" or otherwise) when there is an option to use an unambiguous option.
It often happens with things designed and made in the USA. It is assumed that the terminology or customs of the USA apply to the rest of the world, when it's often not the case.
For example, their bizarre date system. Widespread use of state abbreviations and timezone abbreviations that are only known in the USA. "Zip code" on forms for customers in New Zealand or India or France.
Dates are the worst. 11/12/2020, is that November 12th or is that December 11th? I don't know and your shitty system makes no attempt to let me know whether they're using the regional format or the world-wide one.
Good point overall. I disagree with the abbreviations being exclusively a US thing. Two letter country abbreviations (UK, DE, IE, FR, DK, etc.) are popular in Europe, including on the back of food packaging which lists ingredients in multiple languages.
Yes, but, when taken out of context, you have no idea if you're looking at a state abbreviation or a country abbreviation.
AL can be Alabama or Albania, AZ can be Arizona or Azerbaijan, DE could be Germany or Delaware, etc. Just under half of US state abbreviations are "unique" (as in, not also used for some country in the ISO 3166-1 alpha-2 standard).
Personally I'm mostly annoyed by the concept of states when filling out delivery address. When I enter my country (that has no concept of states), I often find myself having to make something up because the state field is a required field.
Even worse: sometimes they do ask "state" and have a dropdown list of American states. If you live in another country that does have states, you're SOL.
Why would your dropdown still have those when I selected a different country?
Also why's the country picker at the bottom? I've learned to fill it out first and scroll back up the hard way. Otherwise you never know when fields will completely change when you select a country, or just discard everything you've entered previously.
NZ has postcodes, and they're 4 digits, which confuses some websites.
Yes, the dates are ridiculous. I've had to use an online assessment system with hard-coded backwards American due date formatting that I have to warn students about. The vendor just shrugged.
As another poster pointed out, elsewhere "postal code" is usually used. The format differs. In some places it might be 4121, somewhere else it might be SW1A 1AA. In the ROI, there are no post codes.
Driven primarily by frustration with foreign companies expecting post codes - the postal service was late to computerising which paid off in having better OCR/address lookup when they finally did so actually said they were pretty indifferent to the implementation of postcodes.
I have to say my experience with posties in Ireland has surpassed my experience with postcodes. I feel like the most valid "post code" would just be the name of the postman they should ask if they're not sure.
But I have to call out Ryanair in this context. For many years, before eircode, they demanded a postcode on their payment page. An Irish company, serving Irish customers, demanded postcodes when most the country didn't have them.
(And I paypal still believes my postcode is null. I don't recall ever entering that, so I wonder where/when it derived that.)
When I got a (remote) Irish jobs a few years ago I had to fill in a post code at my tax service and a few other things, and I also had to fill in an address in the format of "street number". This makes perfect sense in the Netherlands, and I can't really blame Dutch folk for not knowing that Ireland doesn't have postal codes and that not all houses have house numbers (but rather house names).
To be fair, I have seen non Americans making basically same assumptions.
I have even once seen a thread where Easter Europeans who never been in America argued with Americans about meaning of English word and about American culture. It was literally "are you sure you know better then American how Americans would interpret that word/situation in their day to day life" situation.
Here is an interesting idea, they might know it from their English lessons in school. Which might have taught British English and not American English. But who knows.
What's the unambiguous alternative for crosswalk? The only alternative that I can think of is "pedestrian crossing" and I'm not sure if that would be clearer.
Stairs too. Do they include the hand rails? If a step ends three pixels into a new square, do I select that square? What about the side face of the steps?
The irony of Captcha is in the very statement we're being asked to confirm: "I Am Not a Robot".
The image sets are supposedly to be used for training an AI, which is ... a robot, sophisticated, yes, but just as programmed.
It's often said that "It's human to err". Well, instead, the Captcha seems to be eager to sample some fine "specimens" of Robot-candidates, indeed giving us poor fallibles the early lessons of artificial programming.
"You'll be assimilated. Resistance is Futile!" Borgs, eh?
Yeah this is an interesting point. What constitutes a "traffic light"? Just the bulb? The housing around it? Or the pole as well? What about the control unit? They won't function as "traffic lights" without that. Remove the control box, and they become just dumb street furniture with no purpose.
Do they need to be on a road to be traffic lights? If I took the lights, the pole, and the control unit and installed it in a forest, would it still be "traffic lights"?
I used to close the browser tab when being forced to train Google's image recognition AI, now it's impossible as even some services from public administration of my country embedded this cancer.
"Many children with ONH have difficulty processing sensory information through
some or all of their senses--referred to as sensory processing disorder.
They may show sensory defensiveness and delays in learning fine and gross motor
skills, which can make reading braille and learning basic daily living skills,
like dressing, grooming and personal hygiene,
difficult or impossible without additional supports."
I find audio captchas infuriating at times, because I struggle to
separate the words from the noise, though they have really improved over the
last 3 or 4 years. I remember one night during 2012 or 2013,
wasting 2 or 3 hours of my life, just listening to audio captchas. All I wanted
was to delete my Facebook account and my girlfriend's Facebook account.
Here's a different but similar whine. Probably worth its own blog post except I don't blog.
Dear Apple, can you test your products outside California?
I just left my laptop for 10 minutes on the balcony because I had to get a delivery. Because I'm not in California, the temperature on said balcony is 3 Celsius. Yes, I'm not american and I don't understand Fahrenheit.
Guess what, when I came back the laptop was shutdown and wouldn't start until i brought it indoors. After only 10 minutes in the cold.
Please test your famous benchmark breaking M1 laptops outside your climate controlled clean offices. Use them outside at the north border, near the great lakes. See if the keyboard still works after a month and if they turn on if you try to use them on your front porch.
Example from NYC: my iPhone used to constantly give me nag messages to turn on “driving mode” when in my car. I don’t have a car; it assumed I was driving while I was actually on the subway. Apparently, Californians think if you are moving a certain speed, you must be driving.
A prime example of arrogant engineering - they were wrong both about what was going on, and for writing dickish software.
Don't write nagware. Not only are you not as smart as you think you are, software is still very, very stupid and the world is much bigger than you think, even when you think you are taking that in to account.
I really doubt that had anything to do with arrogance or thinking they were smart. They knew that everyone would just turn that annoying "feature" off, but presumably some lawyer thought it was a good idea to ensure people didn't blame car crashes on Apple, particularly in a court of law.
I used to get that in Sweden when riding a bicycle.
Not to mention that maps at that point (and still) doesn't understand cycling directions. It could often steal focus from google maps. Very annoying and I nearly forgot about it until you mentioned it here.
Yeah, this is a pain with Apple - instead of trying to understand the business and consumer culture of the particular country, Apple tries to impose US business model everywhere.
Some of the things I have observed in India:
- Apple tried to entice and sell iPhones through the carrier, where as Indians largely have a cultural apathy to buying expensive phones on loan / credit and locked to a network.
- Apple still doesn't allow you to update your phone on cellular data even if the indian user wants to - most of the cellular data plans in India offer anywhere from 1GB / day to 5GB / day on 4G. So even if we have the speed and bandwidth, Apple insists you can only update on Wifi. (Obviously nobody wants forced, silent updates through cellular data - but Apple should atleast allow update on cellular data if the user initiates it. It's irritating that you can watch large HD videos, but not update your phone.)
- Apple still doesn't seem to have any data centre in India, and sends an expensive international SMS to the US everytime iMessage or Facetime has to be activated. (And when anybody complains about this online every "helpful" suggestion blames the carrier for this rather than Apple - because apparently some carriers in the US don't charge for this "activation SMS". Obviously since US carriers are sending a local SMS, they can afford to absorb the cost of an iPhone they have sold and profited from. Why would indian carriers absorb the cost of an international SMS for Apple for no obvious profit!? If it was local, it would be free too in India as many indian carriers offer free SMS in their plan. See https://apple.stackexchange.com/questions/390275/how-to-avoi... for more details on this SMS activation.)
- My iMessage was blocked many times because the amount of messages I sent because Apple didn't recognize that social networking in India had moved from Orkut and FB to mobile messaging (that's why FB bought WhatsApp, and it is the number one messaging App in India even on iPhone.)
- Apple still insists on a credit / debit card for use on App store in India, despite India having better online payment solutions that protect our privacy better. Also, cards of India's payment processor (RuPay) is not accepted, and you have to invariably use a Mastercard or Visa card.
> Why would indian carriers absorb the cost of an international SMS
Because SMS pricing is a scam and in practice is a rounding error to handle all of those activation messages. Both countries should just exclude that number from billing.
This is similar to how European carriers suddenly started competing on free international calls/data once EU enforced better rules. We've been told for decades it's really expensive, but now? Nah.
Everything corporates do can be called a "scam". To me, it is a "scam" that a trillion dollar company like Apple avoids putting up activation servers in India because it is more profitable to not do so and just make the indian user pay an international SMS everytime.
It is also a "scam" to me that Apple insists that I can only install everything from the App Store, where as apps would actually be cheaper for all of us if Apple didn't take a cut from the developer. (And I can go on ... but we have to live with the reality of the world don't we - Apple isn't going to give up control over their App Store just as carriers will keep trying to monetise on SMS or RCS.)
Yes, even Android has this issue. It's definitely a legacy carried over from the US - data is still expensive there and carriers probably discourage phone makers from sending large data over their network.
Plenty of places in the world charge for SMS. Ask all you want, doesn't mean things will change.
MY wife just applied for a mortgage in her home country, they won't approve it until they have a picture of her on file.. some places just do things differently hence the whole 'please test outside of the US' point of this post which appears to have gone over your head.
Because SMS is essentially free. Companies exist to make money, but international phone fees ought to be a relic of the past. That they are not is a problem that we would all be better off getting to the bottom of and solving.
What a ridiculous argument trying to blame another corporate to hide Apple's folly. By that logic, the same argument applies to Apple too - Installing apps used to be free before Apple decided to charge developers for it, and make all apps costliers for all us.
I agree working in a wider range of ambient temperatures would be nice, but they do have a support article stating the temperature should be over 10°C, so I wouldn't say they've not tested it.
> I agree working in a wider range of ambient temperatures would be nice, but they do have a support article stating the temperature should be over 10°C, so I wouldn't say they've not tested it.
Yeah, the point is 10 C is sufficient ... for California. Not necessarily for the rest of the world. I'm sure they specify some temperature range, it's just not quite enough. Besides it was left outside for 10 minutes, not hours.
To add insult to the injury, all my older Apple laptops were able to survive bouts on the balcony, and if anything, the winters were colder 10 years ago.
Discharge rate of batteries falls with temperature. 10C is a very common lower limit for high performance lion cells. There's a complex set of tradeoffs vs fundamental physical limits here. It's not at all a case of "aw shucks we forgot places other than CA exist" which is something you're just projecting onto the situation.
I had to buy a new battery for my car because despite a battery shop and the dealer saying there was nothing wrong with the battery, short trips in the car below 10C would kill it in a week.
As someone else pointed out, the dew point is not something you want to explore with personal electronics.
I don't know where you live but I've spent enough winters right next to Cupertino, 20 minutes walking distance from 1 Infinite Loop to be precise, and it's not warm at all, in fact it's cold as hell without heating (not as cold as some places apparently, but still cold enough to freeze your balls off, or cause your laptop to shutdown). So no, it's chemistry at work, not "Cupertino is warm enough to not have this problem."
I don’t think you’re being entirely reasonable here. If you need a laptop that works at 3°C, don’t buy one that states it needs 10°C. If you just didn’t know, you’re presumably still in the laptop’s return window.
Fwiw, I’m in New York, and I really do think this is a reasonable operating temperature for a laptop.
So you need a laptop whose operating range includes 3°C. How long it is exposed to 3°C doesn't matter because components can already take damage from the very moment they are cooled below the operating range.
That the laptop shuts down when below minimum operating temperature is a Good Thing™. For example, the battery performance changes greatly (reduced capacity and discharge rate), and the battery may even take permanent damage when charged at low temperatures.
Also it is a very bad idea to bring a laptop indoors from the cold and immediately turning it on. Condensing water can cause short circuits (people who wear glasses can probably relate). Always let it warm up to room temperature first.
This is not a MIL-STD 810G rugged outdoor device. If you want that, buy a Thinkpad.
I am in Montreal, Everyday I got to school during winters I had to subject my laptop to -20c for 45min before coming back to a warm indoor. It's not like we have any alternatives. It's true for our phones as well.
OP's points still stands. I would add that if your product can't stand this constant hot-cold cycle, don't get in the Canadian market. If you are selling here, we can assume you have accepted the reality of climate and have factored it in your warranty costs.
You obviously never lived in cold climates. Everything outside my winters clothes is at outside temperature (inside is often not terribly hot either..). I am not about to carry my laptop in my coat, which are usually a somewhat tight fit.
Anyone who has bought hot food on a cold day has an opportunity to learn about how some containers do nothing and others do quite a bit. Even thin paper versus heavy brown paper makes a pretty big difference in insulation.
Many bags designed for laptops have open cell foam for shock resistance, and thermally conductive foam is a specialty item. You practically can’t make a laptop sleeve without an R value. So it’ll depend a lot on whether your laptop was room temperature when it went in or is warmer.
If you’re looking for a Christmas idea for yourself, I don’t think you find a better laptop bag than Tom Bihn. They have a laptop sleeve that’s practically an insulated lunch bag: https://www.tombihn.com/products/brain-cell with an air gap at the bottom that prevents damage if you drop your bag. Which I have.
Their stuff is expensive, but they’re bombproof and not Apple-expensive. They’re waterproof, including the zippers, the shoulder straps are butter-smooth, and the factory workers are paid a living wage.
If a device requires some special enclosure to work in the climate it's intended to be used it should come with it. People would find their MacBook Air a lot less appealing if it came with a mandatory carrying case. This is not on the consumer to fix.
As others pointed out, that was likely intentional. Apple designed this safety mechanism into the laptop, so they would not have to deal with such warranty claims.
Also the temperature change is not a problem while the laptop is non-operating. The alternative therefore is to wait for 10-15 min before turning it on.
This is not entirely accurate. Electronic devices are fogging up just like my glasses everytime I come inside. This is known to trip the water sensitive stickers used to deny warranty, even if the problem is obviously unrelated. Those are triggered regardless of if the device is powered on or off. To me this is strong evidence that electronic devices are not tested properly for our climate.
That moisture indicators are used as a way to weasel out of warranty claims is indeed common. But that was not what I was referring to.
What I mean with having fewer warranty claims is that the Macbook doesn't break in the first place. Condensing water is not a problem while the device is powered off, and will evaporate once it reaches room temperature.
That is not similar at all. 3°C is not a normal use condition, silicon normal working temperature is around 50°C (sustained by its own heating in normal condition) and there is probably a notice where the minimal temperature is stated in celcius.
The OC is about assumption on user culture. Assuming that a macbook won’t be used in extreme condition is just normal engineering scoping.
You aren’t chilling on the balcony in underwear, are you? Well of course not because your body, even raised in a cold climate, needs to keep itself at 37°C. Your macbook too need to keep warm, and it cannot use its muscle when the CPU is not heating enough.
But, hey, maybe you just got a good business plan ! Design a macbook parka to sell to your fellow countrymen.
I do wonder about that, could it be battery operation? I know when I'm out using my GoPro on the slopes the battery will die very quickly when it's a super cold day. I blogged a while back about a possible problem of using Macs at altitude when it comes to battery swelling: http://leejo.github.io/2019/07/27/altitude_sickness/
Anyway, on the original topic here's a talk on the [mis]use of CAPTCHAs which is definitely worth your time: https://www.youtube.com/watch?v=3gKRTg-RXsc # note the first 10s or so are cut off.
3c isn't very cold. it's an ambient temperature any phone would have to be able to withstand for an indefinite period to be marketable even in many US states.
Losing capacity is one thing, shutting down and refusing to start after 10 min outside is another. What use is a phone if i can't call 112 because it's cold?
Edit: Also remember that I've had Apple laptops before, and only this newer model deems it necessary to shut down.
That's a battery technology limitation, you wanted a lightweight high capacity and low memory effect battery? the cost is bad temperature range. Otherwise pull out an old phone with NiMH batteries for your 911 calls
Try to use an eInk device in cold weather, its response time becomes terrible down to almost unresponsive. Again that's a technology limitation.
How do other laptop manufacturers fare in this regard? For ThinkPads (T490) the environmental temperature is listed as "5°C to 43°C", and your 3°C would be too cold.
My X1 Carbon 3rd gen got quite some use in the last 5 years, some of it also in ~ 0 °C weather (for at least half an hour), worked fine.
That is naturally just one personal experience, so I do not want to assert that all X1s or even ThinkPads fare better in colder weather. Just that at least some do.
My hands have historically never been happy doing fine motor work at those temperatures.
Even though I’ve acclimated to doing outdoor work in colder and wetter conditions, I’m having trouble believing people are actually working on their laptop while most of us would be wearing gloves.
That sounds close enough that it would probably work just fine. I used my laptops on my terrace in winter, but as i never had any of them 'overcool' i never took note of the temperatire.
I can’t be the only one offended by the unpaid labor involved in CAPTCHAs (training self-driving AIs or whatever).
Sometimes an annoying CAPTCHA has literally prevented me from giving money to a site when I was going to buy something but instead ended up just closing it in frustration.
I wish Apple would offer a way for sites and services to verify that a client is indeed human via Touch ID/Face ID.
Also the frequency with which they appear, and the number of iterations required to pass appears to have a strong inverse correlation with whether you're signed into Google with Chrome or not (we use a wide range of browsers for testing, all of which get reset to baseline).
In other words, it's much easier to just use Chrome and stay signed into Google.
<shrug> We know (pretty surely) that no company's autonomous vehicle of any claimed level is going to survive for 30 mins on Indian urban roads and traffic conditions. Afaik, none of these companies are even trying to get a vehicle to work in such conditions. This gives me confidence that the AI apocalypse is nowhere near.
Many of the criticism of the article focuses upon people being able to figure out Captcha's even if the aren't American, but there are a few things to consider:
* A lot of native English speakers are going to have at least some exposure to American culture. For example: they may recognize that taxi's are yellow from American television and movies. Not everyone is going to have that exposure.
* It assumes that there is nothing from their experience that contradicts the question. For example: there is a business in my area that uses yellow cars with signage. (They can do that because taxi's in my area are not yellow.)
* It places a higher barrier to people who are not familiar with the cultural reference. Sure the author figured it out, but it would take them more thought and time to figure it out than someone who is American.
This ignores an important point that this is a game of elimination and contrasts.
For example in that taxi captcha, there are 4 pictures of trees and 5 pictures of vehicles.
1 vehicle is not identifiable, 1 is an utility van in the distance, and the remaining 3 are yellow cars.
These yellow cars have phone numbers on the sides and signs on the roof (where visible).
My bet is that the 3 yellow cars are taxis.
(And as it happens yellow actually seems to be the most common colour for taxis throughout the world where taxis have a specific colour.)
It usually works out fine. On rare occasions it's tricky but then it's possible to request another captcha.
I think people are trying too hard to find problems. I find hard to believe the comments that blow this out of proportion.
While you might guess that Yellow means Taxi in most of the world taxis aren't actually yellow. Around here they are mostly big dark Mercedes (many of them vans), maybe with some ads on them. And no big white sign on top either.
You just ignored the part where they show how you can figure out yellow means taxi in this set without having known it before. People are smarter than monkeys memorizing flash cards. They have the ability to spot patterns not previously recognized.
The most important one was mentioned earlier. It offers an implicit advantage to Americans. In the olden days we would call this American-centric. These days we would call it prejudiced. Claiming that it is prejudiced is valid since it is effectively asking different people different questions. To the American, it is asking them to identify taxis. To a non-American, it is asking them to solve a riddle.
To highlight that point: most of your criteria relied upon cultural factors. Those taxis have no resemblance to taxis in my part of the world. Taxis in my part of the world are private vehicles with a small magnetically mounted "taxi" sign on the roof. There is no consistency in colour, little consistency in type of vehicle, and they most certainly aren't covered with advertising. I also wouldn't be surprised if there are a handful of utility van used as taxis in my city, since taxi companies offer accessible services.
The game approach also implies some degree of understanding of captcha's. The approach you outlined assumed that the unidentifiable vehicle would not be treated as a valid answer and that more than one image would have to be selected, so you can look for patterns. Knowing the rules of the game may be fine if you are playing a game. It certainly isn't appropriate in the case of operating a business or offering a public service.
I've come to the conclusion that most bots are not that sophisticated. Basic CSRF-protection seems to eliminate most automated requests and the need for captchas.
That is true for crawling bots which hit simple web forms. Also using some JavaScript kills most of them.
Although such methods won't work with targeted bot attack.
I recently suspended my Ebay account. One of the reasons was the fact that, although I used 2-factor login, Ebay forced me to solve captchas every second login.
What am I? A trained monkey jumping through hoops?
Ebay even sells FIDO2 sticks, but is to dumb to actually implement FIDO login?
If you can’t identify a yellow taxi cab I’m willing to loose the sale vs. not having the captcha at all. The fact is the number of people who can do it who are also your prospective customer is still very high.
It’s impossible to find something that is 100% culturally universal. To use the Internet at all it requires it’s own kind of special knowledge and culture so I don’t think the argument is correct.
But the author, presumably non-american, was able to solve the challenge, so obviously it didn't prove him american. But it did prove him human because humans have that innate flexibility (if they choose to!) to recognize stuff and understand the intent even when it's not completely unambiguous or immediate. And that's the point.
Taxis aren't all yellow in the US, either? We have orange and green ones in the Seattle area, not including lyft/uber private vehicles. Those CAPTCHA vehicles are recognizably taxis because of the sign on the top / writing on the side; not the color.
Wouldn’t geo-locating the client request, and selecting the CAPTCHA challenge accordingly, be a way to mitigate this risk? i.e. for client requests from the US, it’s “Please select all the images of crosswalks” with photos of American crosswalks, but for Commonwealth countries it’s “Please select all the images of zebra crossings”. Admittedly there are flaws in this plan, such as when someone uses a VPN to obfuscate their IP address, but surely it would be a net gain?
I hate these things. I failed a few because I was skipping ones that asked me to identify "crosswalks", only to discover about a month later that I was misunderstanding it to mean a skywalk bridge between two buildings. What actually reinforced this mistaken belief was that recaptcha actually called them "zebra crossings" until very recently.
Bonus points to "captcha" where they show you a picture from a perspective of someone wealthy and privileged enough to have seen it from that perspective - like the inside of a private aeroplane or boat, and then expect everyone to guess it.
Or just about anything to do with passenger trains in a country where most of that service was cut decades ago.
As an Iranian, I did not have problems with CAPTCHAs being “American.” Perhaps rural people might have trouble with them, but I don’t think it’s a bottleneck. Adding an example picture of what is desired will help everyone though (Cloudflare does this.).
Honestly, I'm not that bothered. I think most of these users are high-pain users so dropping them is okay if ultimately it lets the website operator offer a service to me cheaper.
Sure, some are unfairly blocked, but that's a sacrifice I'm willing to make.
As I commented elsewhere, sure, when I was first asked I guessed it must be the road markings. But to this day I just select all the road markings that aren't clearly separating lanes.
I don't know if I get loads wrong as a result, but it certainly could catch me out by testing identifying the difference between a 'crosswalk' and whatever other markings you might put across your roads.
To TFA's point, good job reCaptcha, you identified my unfamiliarity with everyday American things.
I've decided long ago that, if I ever need some form of flood-prevention for a website, I will have a deep look into using an automatic proof-of-work system and essentially give the users computer a captcha.
When I think of how America centric the Internet is, CAPTCHAs are usually not at the top of my list. There is so much else wrong with the internet when it comes to being basically designed for North Americans
Just annoyed that everytime a big SV CEO is interviewed no journalist asks questions about Captcha/blind-support. They seem to get lost in asking about quantum computing or future of AI...
All the taxis in America aren't Yellow. In fact, I'd hazard to say most of them are Black. Only the ones in NYC are, and maybe in some other major cities.
I have always wondered how much server costs are these websites saving by using captchas instead of more respectful anti-spam techniques? 30%? 50%? 90%?
I got so frustrated with CAPTCHAs that started using my computer's visual and natural language recognition software to solve them.
It works much better than me at solving those stupid puzzles. It is great because if the software fails, it does not get frustrated like I did.
There are CAPTCHAs that my software solves that I know are wrong, but google says it is ok. Like "pic the pictures with a signal on it" and there is a picture with a signal visible from behind. My software does not detect that, neither google.
They have Street View photos from most regions and they already set the language of their web apps based on IP, but hey - I learned to recognize an American hydrant from 100 yards (not meters, heh).
Doing that tends to make stuff worse. I've lived in countries where I don't speak the local language and despite everything being configured to use English, via IP detection some sites decide to just show me content in the local language when I'm trying to log in. Paypal is notorious for this.
ReCaptcha actually does this too. Everything on my system is configured for English, but it asks me to find the "Fußgangerweg" or whatever. Have to use Google Translate just to know what I'm supposed to be looking for.
Google is bad at this as well, even if I go to google.nl it insists on offering me the damn Indonesian version. I have to add hl=nl to get the Dutch version I want. While I can deal with an Indonesian UI (I do for Google Maps, can't be bothered to change it all the time), dealing with Indonesian search results in harder.
The Nintendo Switch store also. I can't convince it to give me game descriptions in the system language rather than the language of the country where my account is registered.
Like they are not familiar with overcomplicating things that should be simple, just for their own weird large scale performance benefit that applies to them only. Maybe a few other huge companies in the world.
same for self driving cars. Some West Coast CEO affirmed that he was training his autonomous cars in SF rather than AZ because he wanted to do "the hard thing". He'll have a surprise when he sells his system to India. with his system that struggles to turn left in SF and not kill bikes in a bike lanes.
I think captchas are weighting in your natural human responses, which may include your mistakes (which are human), mouse/pointer movements and actions and timing of your answer.
You're partially right. Yes, the captchas do include mouse movements, screen res, IP and a WHOLE LOT of information to try to decide if you're a human... but it was sure based on just that, it wouldn't show the images anyway. It would silently accept me.
It shows the picture selection test ONLY if it is not sure (based on the rest of the data) that I am indeed a human.
When it shows those many different images, it already knows with high certainty (based on majority of user input) that some of them have cars (or whatever) and some that don't. It also shows a few for which it doesn't know because they're using us for training in the first place.
As long as you get the ones it knows with high certainty right, it doesn't matter what you answer on the other ones.
Anyone with a half functioning brain can figure out which ones are the taxis. Taxis are not yellow in my country and I would have no issue picking those out. They have the very distinctive sign on the roof that no normal car has.
Not even OP was confused by this, they are just complaining on the behalf of some theoretical person who might get confused.
I think the blog is complaining about a more general thing - that Americans often assume that everone is American and make decisions based on this assumption. This can lead to a range of frustration from the annoying-but-whatever (taxis aren't always yellow) to WTF (dimes, nickels oh-my).
They're currently the dominant culture so it's not an awful assumption - people generally can accommodate and work around these issues. It's just, I dunno, a bit insensitive or frustrating?
I just got one "mark everything that shows bikes" with pictures of cyclists, motorbikes and cars. The translation used the norwegian word "sykler" which to me is the human powered one. But it looked like it wanted me to also select motorized ones as well. That's something that's lost in the translation.
Testing some more here https://www.google.com/recaptcha/api2/demo I also got "mark all fire hydrants". I've never seen one in real life, only those red in American cartoons. I got "parking meters", but none of them looked like they do in my country (and I've never heard anyone call them that word they used for translation). I got mark "mountains", but to quote Dundee: You call that a mountain?
> Anyone with a half functioning brain can figure out which ones are the taxis.
Unfortunately, there are tons of people with less than half functional brain... well I'm slightly joking here, but the point is you can't assume your user's cognitive ability. The CAPTCHA as it designed should be obvious to any human that would use your website (including those with impaired cognition) while being hard to robots. The OP is claiming that Google CAPTCHA is an international service and yet can fail to be obvious to a significant portion of non-Americans that would use your website.
That's just a US thing, and if you aren't familiar with that, then you won't recognise it as a taxi.
Here in the UK, taxis are normally just normal cars with vinyl stickers on the side. I think many people would associate the sign on the top with being a driving instructor's car
I would have no problems identifying a minibus either. They typically have writing / phone number on the side and something on the roof. The google captcha is usually fairly permissive with mistakes and will show you a different question if you fail the first.
I'd like to see someone actually show an example of a captcha they were incapable of solving.
> I would have no problems identifying a minibus either. They typically have writing / phone number on the side and something on the roof.
What if the writing accompanying the phone number says "Schlüsselservice Maier"? Some kind of ... shuttle service? Is that a taxi? (It's actually a locksmith.)
Of course these kinds of situations don't really happen for English speakers. I don't think I've ever seen non-English text in a CAPTCHA... Instead you get a picture of some Californian street with palm trees or something. (I assume those streets are in California. Maybe they're actually in Jakarta or elsewhere, in which case I'm sorry for being presumptive.)
I'm from Germany and I actually do have issues every once in a while. Like traffic lights, or buses, are mostly fine, but in some cases, I wasn't sure and chose the wrong images. Is that a truck or a bus? Is that a street lamp or a traffic light? I did feel that it would have been easier for me if I was American.
If we don't push back early, there's the risk of ending up in a situation where every CAPTCHA is stuff like "Select images of arugula" or "Click every Dodge F150". The occasional fire hydrant CAPTCHA was already pretty close!
There are multiple other ones with the roof sign visible which have exactly the same look from the windows down. So its pretty easy to conclude that this one is a taxi too since it looks like the other taxis.
And if you get it wrong you simply move on to the next question.
And if you get it wrong you simply move on to the next question.
Yes, and then on to the next. They don't just profile your answers, they profile your browser too. The more obscure the browser, the more likely my answers are "invalid", regardless of what I select.
I always close the tab, it's way better for my health.
It's an argument against US cultural imperialism using a more benign example (US-centric captchas) than the politically charged ones we tend to see cropping up a lot.
> on the behalf of some theoretical person who might get confused
A theoretical person that also wouldn't get shown another captcha as a follow-up. I constantly fail captchas on some technicalities (or haste), and have never managed to fail so many to be flagged as a bot and be denied to a website.
I have failed sufficiently many to be flagged as a robot, several times. Maybe I am one?
And given how many times I usually have to solve them, if not critical for some reason, I tend to just close the tab for whatever I was going to do or read, if a CAPTCHA shows up. I have very little patience for doing free work for big corporations. If it was a non-profit with public access to the data set, sure.
It'd certainly be interesting to see how people got on with captchas where you identified plain-looking vehicles as taxis because they were obviously Hindustan Ambassadors. Or whether guesswork alone was sufficient to identify unfamiliar terms like 'layby' or'sleeping policeman'
Maybe. But I think the better question is whether there is any evidence that anybody is actually confused what what a taxi is to the point where they are unable to answer these things. I am skeptical.
I understand people being annoyed by US-centrism, but that is a different matter.
Last year there was another similar issue and each new version builds become slower and slower. I waste half of my time waiting for compilation. Never had any such issue with React/VueJS or any other framework except Angular. I really hate it.
In my country (UK) taxis are not yellow either (although yellow is a common colour for taxis throughout the world) but I am still able to guess which pictures are of taxis when the other options are of palm trees and a van.
I think that's the point of these captchas. The average human being has no problem passing the test even if they are not 100% familiar with all the details.
There is obviously a limit and indeed pictures should be chosen to be generic enough to be workable throughout the world, but I feel this is 'problem' is blown out of proportion by some commenters.
Around here the van is more likely to be a taxi than any car. So,... no. The cultural bias is nauseating. I've never yet managed to identify a crosswalk. I frequently fail at busses. All this click-click-click on items mysterious is just exhausting and leaves me feeling drained and dehumanised. These days if I see one I generally just bounce. It's less humiliating that way.
> Detect if a passage is written by human or AI. If you can't get in, it means we reached AGI.
So we've reached AGI decades ago then? Text generators aren't rocket science and carefully cherry picked results of even a simple Markov-chain will be indistinguishable from human writings.
I had one that asked for pedestrian crossings. Now, I've visited enough places to know that they change a lot, and that usually when you're there with context you can spot them. But from a weirdly angled photo I can't tell if that sequence of short white stripes is a crossing or some type of intersection marker like it could be in some places.
When I decided it looked too narrow to be a crossing, the captcha wouldn't let me through until I agreed that it in fact was a crossing.
Nonsense. I've never been to America but never had a problem recognizing any object on a captcha (except some indistinguishable letters on letter-based captchas occasionally). Everything is obvious even if it looks somewhat different from its European counterpart.
I don't know how much exactly is a nickel or a dime but it's obvious to me these are coins.
Everybody I know (and most of them have never been to America) is Ok with these American captchas too.
The only problems with captchas are they spy on you (Google doesn't actually check if you are a human, it checks which particular human you are) and annoy you.
.svg){kind=link}
