I think that Google and Apple will be extremely helpful in the coming months, at least in the USA, for tracing the spread of COVID-19. Google and Apple know where probably 90+% of adults/teens in the USA are located in real time. If they are able to anonymously tell you that you were in the same building as someone who has tested positive for the virus then you could take extra precautions or further isolate yourself to help prevent the spread. This wouldn't be perfect but it would be much more data than is given to people today about who is infected and it would make tracing contacts much easier for health officials.
A small number of people who read sites like Hacker News will find this an appalling invasion of privacy. Health officials will think it's a really useful thing to try. It'll be interesting to watch, if it does happen.
In the near term, I would happily try to convince everyone I meet to opt-in to such a service as I feel the health benefits could greatly outweigh the privacy concerns. But then when we do have a better long term solution for this virus I would also strongly recommend people to disable the tracking.
At the same time, the people who use this data also has to recognize its limitations.
When considered collectively, it is probably fine for an overall picture but it will have some limitations. For example: the people who are not submitting data (whether they found some way to disable it or simply don't use the technology) probably live a sufficiently different lifestyle to render the generalizations meaningless for that population.
When considered individually, it can be disconcerting. The accuracy of location data can be an issue. In urban areas, being off by a few meters can associate a person with one population instead of another. Data classification is another reason. In my area, it is currently a summary offense to enter a park. That can create problems for parks workers. The issue may easily be resolved, but it creates headaches at a time when people have other concerns on their mind. Then there is abnormal device usage. Exactly how do you treat people who leave their phone at home 24/7 because they have no use for it (currently unemployed), do not want to be tracked, or are simply unable to leave their home?
It is easy to argue that these people are professionals who know how to handle data, so limitations will be recognized. That's certainly partially true: governments, police, health officials, and the fine folks at Google are professionals who know how to handle data. The thing is, they typically know how to handle their own data and may not be familiar with handling data from other sources.
With Apple and Google they already have the permission of the tracked. It’s in the terms of service for your phone, or your phone’s OS, or any of the other vectors they control that you carry around with you all the time.
Fascinating data, the US generally showing a less than 50% change while other countries with national lock downs are >90%.
It'd be very interesting to see this data overlayed against cases of covid, you maysee them both follow a similar downward trend with a week or two delay. Possibly also determine what change in mobility is required to prevent exponential growth
The data around parks & recreation seems really odd to me, does anyone else agree?
Anecdotally, the public parks where I live (Bexar County, Texas) have gotten so busy that I rarely go on weekends anymore because social distancing has been made so difficult. Even after 5pm on weekdays, I've seen a spike in traffic similar to what a really busy weekend would look like. This has been the case at every park I've been to. Since basically any other form of recreation outside the home is no longer available, it makes sense that they're busier.
Perhaps they're including places like theme parks in that category? Bexar County, to my knowledge, doesn't have enough closed state parks to account for the drop in traffic. Obviously, I'm missing something. Maybe people just aren't searching for parks further from them and don't need directions to their closest park.
India's mobility report[1] shows a spike in outside activities just before lockdown. I hope I am wrong but I think we are going to see spike in coronavirus cases just before end of lockdown. Not sure what government will do then.
I see no evidence at all of such a spike as you describe. In the last couple of weeks, you had janata curfew on the 22nd, and things plunged; at about that time, a fair fraction of the country went into lockdown, so on the 23rd and 24th numbers came back up, but not to baseline; then from the 25th, national lockdown, and everything’s back down to levels similar to the curfew day (some figures a bit higher, some a bit lower).
I wonder if the graphs are affected by a "starting" bias, meaning is the mobility on the first date, February 17th for the charts I was looking at, used as the measure for seeing the trend on subsequent dates. The issue what if that date was particularly cold or rainy and then as we moved into March the weather got better of course you would see greater mobility movement as compared to a cold February day esp. the category of Parks and recreation
But instead had they used is sliding window for the baseline (i.e. using the average of the mobility for the past 3 years) for any given day then seeing what this year's date is as compared to that we would see (hopefully) a dramatic decline.
At least here in Finland you couldn't miss that when going outside. One factor is sunshine. Another probably kids getting unbearable after not having to go to school for many days. Probably also applies to parents working from home. Lot of families taking a walk. They closed natural parks after they got overcrowded.
I'm in Ireland .. Retail & Recreation are down 80-90% in most counties!
What I did find interesting though, is that after the initial lockdown, we show a reliable increase in "parks", that all take a sharp nosedive after it was tightened and we were told recreation should be within 2km of home.
That felt like something of a facebook-furor at the time, but it seems google's data really does back it up.
A theory i have heard from someone in the hedge fund industry is that this use of "alternative data" is mostly marketing. Investors go to hedge funds for their genius high-alpha proprietary strategies. If you can honestly say that you are feeding tracking of honeybee movements into your model, that makes you seem like far-out geniuses with loads of alpha, so you will attract more investors. For a hedge fund, attracting investors is more important than actually doing good trades.
Of course, nobody will say this out loud.
This is not to say there are not some genuine users of alt data. Of course there are! But it's not happening at the scale that everyone wants to believe.
This is a very similar story to the whole "big data" thing in Main Street businesses.
> Sergey Brin, Google’s co-founder, once said the company should start a hedge fund because it had so much information. But Eric Schmidt, Google’s chief executive, reportedly responded, “Sergey, among your many ideas, this is the worst,” since the company would face serious legal problems in starting a hedge fund.
> Regardless of Google’s plans, the company is looking for a trader of foreign government bonds, an analyst for Google’s United States government bond portfolio and a portfolio analyst for agency mortgage-backed securities, according to Business Insider.
Yeah there are stories of hedge funds buying and then selling cars all over the country to find out the serial numbers so that they can estimate their sale count before the quarterly reports in a german tanks kind of fashion. Having good information is very valuable for hedge funds.
However, I don't think that hedge funds are "Googly" products. They fail at the toothbrush test. They require massive capital investments, and while Google has lots of money, is it enough to drive enough revenue growth?
Furthermore, there is the legal danger. With businesses using Google products, Google is likely in possession of insider information on most publicly traded information. E.g. when there is an internal meeting at a company and suddenly everyone is visiting job websites that use Google CDN hosted fonts, or maybe everyone is visiting websites looking for cars to buy or houses. Or think about internal company documents hosted on Google docs or that meeting being conducted over hangouts. How can Google ensure that their products don't accidentially use that insider info? How can they separate the toxic insider info from the okay to use non-insider info? Does that separation already constitute insider information?
Lastly, Google possesses probably more information than most entities in history, and even today I'd put it into the top 10. They also have gigantic capabilities to organize that data, it's in fact the very core purpose of the company. If you possess perfect knowledge how a market will evolve, will someone want to trade stock with you? If you knew how a coin toss would turn out, would people want to make bets with you about the coin toss? Google getting into the hedge fund business might in fact not work because their buy or sell orders alone could change the stock value to a degree where Google doesn't make any money from it. I'm not an expert though, maybe there are mechanisms to counter this effect. Please correct me if I'm wrong.
Too much noise. The costs to sift through and find an identifiable market trend that has an actual upside would be very high, and you still don't have a complete picture of the market so you'd be wrong in a lot of scenarios.
Deaths have doubled in 3 days so Sweden is growing exponentially too, it just started a little later than other countries so it's in the part of the curve that feels pretty innocuous.
It's quite remarkable how similar the trajectories are starting from around the 1st death (though of course later trajectories might differ markedly depending on the measures taken now).
Those show cases not deaths, I'm not sure that's very reliable as a measure - the testing varies dramatically, it will be tempting for governments to undertest and it's hard to get hold of tests now.
Just to pick one example the US has shot up to top place by cases in the last few days mainly because they started seriously testing.
Personally I prefer graphs of deaths (normalised for population would be fine I guess), starting from 1st death or similar rather than arbitrary offsets in date as the spread only begins from the first cases (usually undetected).
It would be interesting to be able to compare countries. For example, which countries have responded most to a lockdown and which have not responded at all? The data is already broken down by state so it would be good to see this data on a map of the US.
They specifically say in the report that data differs even between individual regions in a country and are not comparable.
It's really more of a qualitative "we have these numbers and they went up/down like this" kind of a report. They don't say what the numbers mean (except some vague sense of being related to "visits and length of stay"). I find it really hard to draw any kind of conclusion from this.
Why would anybody turn on location history? (Serious question, I don't even have Google Play Services in my phone). 95 % users are known not to touch any settings. I guess Google has a trick to make more of them do it. Which one?
I strongly suspect Google has two separate classes of location history. Data which has been anonymised, and data which hasn't. I strongly suspect that your location history always exists, and that "turning location history on" only gives them permission to keep it personally-identifying.
Google make a lot of use of location data. It's used to map traffic in Maps, and is/was/? used to map wifi to improve geolocation. These reports are just another example. And they'd all work just fine with anonymised data.
(I have no proof of any of this, hence over-using the phrase 'strongly suspect'.)
Why wouldn't I? I can see where and when I was, how long I was there. It can give context to notes and photos. It's essentially an extension of memory.
If it's an extension of your memory, then why deliberately share it with others, with police, with FBI? There were many stories on HN how random people got into troubles because of this "feature".
And this is something that should be a call for change in government, a call for change in leadership, and a change in how the laws are structured. Perhaps better police/FBI training. And so on.
I'll also mention that these things are memorable because they are outliers in general, and sometimes spectacular. This is why we remember plane crashes and ignore the numerous shower accidents that kill folks.
Because it also distributes to all the devices I own so when I want to use my own data, I don't have to remember whether it's in the phone's store, the laptop's store, the tablet's store, and so on.
I'm pretty sure it is on by default: I'm both happy and paranoid about this.
I've found a phone using this. The spouse left the phone on the car: It fell off a bit past where we had stopped. This was on a mountaintop, in the countryside, traveling back from Sweden to Trondheim - not too far from the Swedish border. Location services meant we could head out and look for it. We could see if it was moving (it wasn't, so it likely wasn't stolen) and we were able to use my phone to find his. It was pretty accurate, even in that location. The phone was only a few months old. It had some scratches, but was otherwise fine. (I think the screen is now going out, but he drops his phone regularly at work and the waiting time for a new screen was 5-6 months to get the screen).
But I do understand that it means that governments could track us: Companies obviously are and I shudder to think of a motivated, angry ex or a motivated stalker. Turning the phone off doesn't help, and I don't know how far "turning off tracking" goes since there are ways to get, at least, a general location using tower information. I keep hoping that it'll turn into a greater good sort of thing, but this will only happen if governments change laws to better match people's behaviors so long as they aren't endangering others. I don't think that would happen in the US without major change, but I have hope where I'm at.
If you want to store places in maps for example, it gets turned on (last time I checked anyway). I imagine there are lots of little features like this which turn it on.
They make it very difficult to keep location data turned off. If you accidentally open Google Maps, a pop up appears where dismissing it turns on location tracking. I also think they make you enable it nowadays to use Bluetooth.
From The Verge’s report [1]
“Google said that it chose PDFs over web pages because they could be more easily downloaded and shared with workers in the field.”
I think that the location services use a mix of IP-based and GPS-based localization - GPS doesn't work too well indoors, while getting your IP from your wireless router (ie internet provider) is much more reliable.
Just a guess: people already spent more time at home than in other areas so if the absolute amount of time increased at home the same as it decreased elsewhere, the percentage increase is smaller for homes than percentage decrease elsewhere?
Governments can just ask for it and Google has to give it to them. [1][2] In fact the US Government has decided it is easier to have the service providers collect data for them and then just get it when needed instead of them separately trying to "archive" data.
Google doesn’t have armies or overzealous attorney generals (and lackeys) willing to ruin lives for their stupid conviction rate. I wouldn’t have believed police officers lie on a ticket if it hadn’t happened to me.
A small number of people who read sites like Hacker News will find this an appalling invasion of privacy. Health officials will think it's a really useful thing to try. It'll be interesting to watch, if it does happen.
In the near term, I would happily try to convince everyone I meet to opt-in to such a service as I feel the health benefits could greatly outweigh the privacy concerns. But then when we do have a better long term solution for this virus I would also strongly recommend people to disable the tracking.