It's a neat idea, but after too many sarcastic, "clever", trying-to-be-funny, defensive, et al emails in my career I have a simpler heuristic. If I have to ask myself "should I send it?", the answer is no. Leave it in drafts (with no one in the to: line!) until either things have cooled down and I delete it, or there's more perspective and it's actually ok to send.
I have too many fails to tell, but the worst ever was an email from a middle manager I didn't know at BigCo, cc: upper management. It blamed an issue on my team at the time, and I took it as an attack. For once there actually was no issue, and in my reply I managed to subtly rub in that they apparently didn't understand how their own dept. worked. CC: upper management still of course. It worked too well, and the person called me crying an hour later because they looked like an idiot to the big cheeses. As soon as we talked on the phone, they became a real person to me, and I felt like the utter dick that I had been.
Moral of the story, there are real people with real feelings on the other end of your words. It's almost never worth it to be a dick, slightly off color, or sarcastic, save that for in-person ;-)
- Does the email fit in a slack? if not, delete it, go open slack and type the message into the slack box, then copy paste that into the email and send.
My experience has been that being firm but polite over a phone call or in an in-person meeting is more effective than sending a sarcastic, clever, trying to be funny, or smiley-face-at-the-end email.
Feature request for email: Click "maybe send" to queue the email and send it to yourself in 3 hours, where you might be in the right headspace to rethink anything before hitting "send" to the original recipient.
> As soon as we talked on the phone, they became a real person to me, and I felt like the utter dick that I had been.
It reminds me when I was doing Jira & Confluence admin work for team leads at my previous job. I don't think I was a dick to the (many) people sending mails asking for admin operation, but when one came to see me in order to work together on his task workflow, this person went from an outlook contact to a real person and I realize that I could have been more polite in my email answer. Also it was weird, because from his picture I had imagined him taller.
The real value of this tool is as an objective measure of how pleasant your communications are.
Now that Codes of Conduct are the norm rude/abusive behaviour is not tolerated, but this can cause problems because one person might see rude and abusive behaviour where another person does not.
Having a tool to objectively measure incorrect words and thoughts would be a great way to protect developers from themselves.
> Now that Codes of Conduct are the norm rude/abusive behaviour is not tolerated, but this can cause problems because one person might see rude and abusive behaviour where another person does not.
I dare say "abusive behavior" has never been "tolerated", but our individual and societal recognition of "abusive behavior" is different now than earlier.
Having written my bachelor's thesis on how negation in sentences affects their sentiment: it is really, really difficult. Even just differentiating between negative/neutral/positive sentiment is successful only about ~65% of the time (depending on the source material). Text based Irony/Sarcasm detection is still an unsolved problem (most of the times even for humans, as it is strongly context dependent, not to mention missing indicators such as tone of voice and body language). Basically, you are way better of listening to your own intuition rather than using a computer to flip a coin.
Short answer: yes, crowd sourcing would work better.
Long answer: It's difficult to determine how good/bad people actually are at detecting the correct sentiment, as data sets containing phrase/sentence <-> sentiment pairs are often created by majority decision of human taggers. E.g. 7 people are given the same training examples and whatever most of them choose is then used as "correct" answer (gold standard). This might not be the real correct answer though. However, even if we accept this gold standard to actually be the absolute truth, most humans only have a correct detection rate of about ~80% (this is a very rough number, as it depends strongly on the source material, e.g. Tweets, product reviews, etc.). Still, this is way better than computers perform at the moment.
Then again, I assume those texts are written by humans for humans. So isn't the "correct" sentiment exactly what humans tend to make of it? And if humans aren't very good at detecting the sentiment, maybe the writer is at fault, not the readers.
I think letting a number of people read the text and choosing the majority vote as the text's sentiment might not actually be a very bad way of determining that.
It might be correct to say that a group of humans is interpreting the sentiment "incorrectly" if they don't have all the relevant context / information.
Completely tangential (but it's HN so that's what we're here for ey): if that video represents what normal people look like when they're having an "off" morning, then I am the actual Antichrist.
There is some real potential for comedy in that video, you're missing out :) Turn up the Grinch factor by at least 10. At 00:16, put a "beat": Change the pace, the colour balance, the framing, everything. Lovey dovey intro, music, la la, then BAM waaa waaa nuclear alarm sound, zombie hand slams on alarm, which breaks, etc. (As a matter of fact, show an actual zombie!)
Right now, that "bad morning routine" to me basically looks like... what I imagine the Dalai Lama does in the morning.
Oh and if you're gonna burn toast; actually burn the toast! Black. Burn the house down black. Smoke everywhere. And the car, just show it completely totalled. Like, just the frame is left, somehow. When the meeting's canceled, idk, the office building is just destroyed and it's this person looking out over rubble.
(I assume you probably cobbled this together from stock footage, in which case you're obviously limited by what's available. Still worth looking if you can find some much darker shots? :) basically try and exaggerate much more.)
[edit: sorry forgot to say well done. it's not bad at all, and I think you did really well on the "upbeat" bits. all the more potential for contrast!]
I appreciate you for saying that you don't store the text, but the fact that it has been sent, processed, and then returned means that this practice could change any time and I'd never know.
If I were to use this for work-related reasons it would need to live in a more on-prem / secured format where I could ensure this via controls.
This could change at any time, because extensions are automatically updated (unless you change the setting), and the permissions when it's installed allow it to, so it could start sending the text in the active text area, for example. There's also no way for them to prove that they don't store the text that's sent to them.
Chrome used to always show a list of requested permissions when installing an extension, but it doesn't anymore, unless it's asking for special permissions. Extensions can do a lot without special permissions.
> Everyone benefits, the recipient of the email, the person sending the email and the business.
It is not clear to me how you benefit.
> Your highlighted text is passed from your browser to our API where the sentiment is calculated, we do not store in any way the text you highlight. [..] We have many ideas of how to push this tool forward.
You don't store my text. How about my sentiment?
rarely.io has a GDPR page (which should really be called Privacy Policy, IMHO), but shouldisendit.io - where you are offering the extension - has not.
> Get involved. Install the extension and start using it. We would love to know how you get on and any candid feedback you have
Install? No thanks. Please first get your transparency and privacy ensurances in order, is my kind (not candid) feedback to you :)
I'm a "tech dude" who have struggled with emotional control. I used to think "I don't have that many emotions, I'm so rational" which many heated emails proved wrong. But I have worked hard at recognizing and naming feelings and emotions and trying to figure out where they come from,and I'm doing quite well now I think. So this is something I would have liked in the past but this is just a glaring symptom of a problem of people acting without knowing what feeling is in control. Sad.
(What workes for me is "non violent communication" btw, google it)
IMO add a gif of the system working right on the front page (similar to https://getstation.com/)
It’ll drive more conversions.
Also after working on sentiment analysis for a long time, you should let each user build their own model with tensorflow.js. That may be a down the line kind of thing, but otherwise your model will be fairly inaccurate.
It doesn't seem like HNProfile is particularly accurate. It seems to interpret terse comments as unhappy, and all my interests listed come from a single comment thread a couple months ago discussing Canadian regions instead of anything technical I have talked about.
I will say that I'm unhappy that you've built a chart on what you percieve my mood to be and published it without my consent.
It's not particularly accurate, which kind of is my point... I was able to get right around 80%, between myself and others manually tagging the dataset.
Funny thing, even with myself and others tagging the content, there is a high variance in our scoring. Meaning - accuracy is in part the eye of the beholder.
Also, if you think that this is intrusive -- imagine the NSA reading your personal email
Yeah.. from them I don't expect adherance to GDPR. You however are commercial, have no Privacy Policy, and collect, analyse and publish to the public user data (which in many cases is not anonymous, or can be easily de-anonymized) without their consent.
The fact it is not particularly accurate makes this even worse.
Sorry but what on earth is this demo video blathering on about? half way through it's still setting the plot to some hypothetic bad day. Get to the point!
Exactly what I was going to post. At the very least I was expecting a demo of the product after all the nonsensical "bad day" blabber. But it never came and I still had no idea what I was looking at.
On the contrary they seem like perfectly rational responses to an introductory video that is redundant at best and patronising at worst.
I was dissuaded from installing it because I assume the products quality would reflect how little the founders seem to understand people in thinking that it's necessary to accurately detail what a poor start to the day would look like.
The video should be a simple gif demonstrating how an email with a negative tone is detected and turned into a positive tone. People are smart enough to figure out how that could be useful.
The feedback might have been good but the tone of those comments had hint of snark. Considering the product is to help you remove negative sentiment from your responses I would say they were good use cases.
replace blathering and blabber with more constructive words
Looks interesting, I'm excited to try it out. I had trouble parsing the "Understanding Mood" paragraph. Consider this a pull request.
> What if you could get a mood score for emails you receive or send? A way to quickly sense check what you are about to send without your bias emotion or when receiving an email remove your current mood or maybe a currently negative relationship with the sender.
ShouldISendIt provides a mood score for emails you send or receive? When sending, quickly get a sense of the mood that your message communicates from an objective algorithm. When receiving messages, our tool can help you see past your current disposition or preconceive notions of the sender's intent.
Hey, so I've wondered for some time now how social media sites might benefit from a sentiment analysis score applied to their users, who are then given more or less moderation ability based on the result of that score.
Obviously tons of edge cases, but what do y'all think of this? Combined with Slashdot's "king for a day" moderation style, could this solve some of the lingering issues present in social media?
I'd love to sit down and talk to someone about this concept for awhile, see where the idea goes. I'm super sure it's come up before, I just wonder what the state of the art is.
Peepeth.com is experimenting with features like that. Currently, posts get a toxicity score from a Google API before they're even posted, and users are asked to confirm if it's above a certain threshold. Lots of false positives, but it seems to be reducing profanity.
Does anyone else remember that this feature was present in old versions of Eudora on the Mac? It would literally put icons of chili peppers on your emails to say how spicy they were.
I've had a conversation about something like this nearly every year that I worked in Corporate America. Interesting to see it get built (I can't speak to the history of other implementations). Those past discussions were around an MS Outlook add-on to disable/gate the send button if the email draft content was too egregious.
A few questions I had about your product:
* How would you monetize? (If you're thinking of doing so?)
* Have you considered integrating with other messaging?
Cool execution of an idea - congrats on launching!
This could get share traction if users were able to submit a message and make it public. My friends/coworkers had a phase where we did that with this tool:
Pitch it to Linus. An endorsement like that could be a huge marketing boost.
"With Should I Send It? I don't sound like a huge jerk whenever someone tries to get a patch merged that's fucking braindead^W^Wnot up to kernel quality standards. Thanks, Should I Send It?!"
I'm being lighthearted, but not really joking: Linus is really looking for something like this to help keep the rage down in his correspondemce.
As someone who recently switched back to Firefox, I've come to realise how many of these useful Gmail addons are strictly limited to Chrome only.
Firefox has been terrific for everything else, but it's surprised me how much functionality I've built into Gmail through addons and how it's almost a showstopper for me migrating to Firefox.
Can you invert it? This makes me think of Decker's wife in Do Androids Dream of Electric Sheep who adjusted her mood organ (designed to dispense psychiatric drugs each morning to stabilize your emotions) to make her depressed and miserable.
"Your message is insufficiently vitriolic. Send anyway? (y/n)"
I guess this is an English only tool. Does it account for non native speakers, like an Indian and Ukranian corresponding in
English? Either way, while I can see that developing tools like this is fun and engaging, I won't trust the results.
I've been using Boomerang's 'Respondable' feature for quite some time and it is absolutely awesome. It's rare that I'm 'off' but it's a nice way to catch when I am.
I have too many fails to tell, but the worst ever was an email from a middle manager I didn't know at BigCo, cc: upper management. It blamed an issue on my team at the time, and I took it as an attack. For once there actually was no issue, and in my reply I managed to subtly rub in that they apparently didn't understand how their own dept. worked. CC: upper management still of course. It worked too well, and the person called me crying an hour later because they looked like an idiot to the big cheeses. As soon as we talked on the phone, they became a real person to me, and I felt like the utter dick that I had been.
Moral of the story, there are real people with real feelings on the other end of your words. It's almost never worth it to be a dick, slightly off color, or sarcastic, save that for in-person ;-)