We took "Show HN" out of the title. That's for when you're introducing a project, not adding new features. Please read https://news.ycombinator.com/showhn.html.
"That sounds like a good approach; reimplementing an existing API like EC2 or DO makes it easier to integrate your software with tools like Vagrant and Terraform."
Maybe the words "reimplementing" or "existing" triggered it? For you, maybe it was "torrents"?
Cool, but it doesn't take retweets, @replies, and quoted replies into account. My worst comments over on Reddit are where I quote someone else who said something mean, and the same goes for Twitter.
And the results for HN are truly head-scratching. Use of the word "fucking" regardless of context is apparently enough to elevate a comment to "worst" status. (Watch me get a few points just for this comment alone!)
Don't get me wrong, this is a neat tool, but it's going to be misused in character assassinations on others, especially with its current presentation, where phrases are used like "troll score", "most insulting" and so on.
I'd really suggest toning that presentation down and emphasizing the fact that this is imperfect until the algorithm has been tweaked further.
This seems to be the default error message that's shown when the queue's too full to analyse a user. Which is probably not a great thing when the site's busy being upvoted on HN.
Is there an upper limit to the number of characters this will analyze? I've gotten into arguments here & on reddit that, while civil, are certainly more vitriolic than what the script found. All I can think is that longer diatribes are ignored?
From HN:
"I've researched the "why" and a couple years ago had the opportunity to discuss some of the broader ideas in the history & anthropology of locks. It was actually titled "Why do you lock your door?"
-emhart
From reddit:
"Very glad you dropped in. Nice to have someone with first hand experience offer this up, thanks."
-schuylertowne
From twitter:
"Here is the US patent filing for that taiwanese lock, so you can see the illustrations: https://t.co/K7F3aY6RTD"
-shoebox
Concept is nice, landing page and the website are nice. What needs more work is the algorithm for determining hate. Almost all the comments that Hater News chewed out of my Reddit username were not at all* trolling. Indeed some statements were even contributing to the discussion.
Broken for network = HackerNews. All usernames including the suggested one and my own return the "Hmm That's probably not a real username." etc.
And then some random youtube video plays in the background. And there's a "Still crunching numbers" animation (when I modify the css on the page to hide the overlay).
And then "Something broke, sorry :(" and then back to home.
Only Twitter worked for me, but it was kind of cool to read some long forgotten tweets.
I don't really understand the periodic table concept. My first thought was that it grouped your insults into types, like logical fallacies versus outright insults. I feel like that would be more useful for introspection.
a lot of the links to reddit comments are broken, if the title of the page in reddit has a slash in it, the slash is not being ommited/escaped in the URL and so reddit can't find it.
Apparently, this is a negative comment:
"What do you with a bunch of books and an empty pod? A racing track of course! http://bit.ly/1wThvTK "
-pmelendezu "
He is probably just making a corpus of negative words, where "less" and "only" probably have negative connotations. This is the problem with this kind of analysis, extracting it properly from context is really hard
It just means that you don't have insulting comments that it can find. Hover over titles with asterisks to better understand the context of your score and worst comment when you get your score, it just finds the "worst" one out of the bunch and if you don't have a bad comment then it just brings back something like that lol. :)
One of mine was (truncated) "What an amazing fucking museum! ... Fat man and Little Boy earrings" which was in reference to the Atomic Bomb Museum in Las Vegas.
I guess I hit the key words of calling something "fat," and dropping the f-bomb.
The "troll rate" across all my accounts is 3.X.. which is interesting. Because I'm definitely a little more.. "lively" on some accounts than others (for instance HN compared to Reddit). I would have expected one to be much higher than the other.