for those of you who still pollute yourself with reddit, have you seen it anywhere since? it must have been purpose-built for CTH because I haven't seen a single reddit user mention it happening post-ban, not even a warning
for those of you who still pollute yourself with reddit, have you seen it anywhere since? it must have been purpose-built for CTH because I haven't seen a single reddit user mention it happening post-ban, not even a warning
I keep seeing people say this, but I don't know why they'd need to be able to identify all the hate speech on the website to have some measure of it.
to reduce something by a percentage you have to know how much of it there was in total, right? you need a denominator for your fraction
EDIT: note that "reduced reported hate speech by 18%" is a vastly weaker and shittier claim and probably what they really did
You'd need a denominator for something like "percent of comments on reddit that are hate speech," but they just claimed a shift in the total.
They probably did something like measure a reduction in reported hate speech, yes. But that's accurate insofar as hate speech is correlated with hate speech reports, which is a reasonable enough measure that I wouldn't fault them for just calling them the same thing.
The only actions they could take to reduce hate speech to 0% using that metric would be to remove the report button (obviously bad, there are limits to unthinkingly following metrics) or removing the entire website (obviously good, not sure why they haven't to be honest).
then you'd need to know the total in both cases
this is where we disagree
I know it's more humid today than it was yesterday, and I have not weighed all the water in the sky.
also I think you missed the addition, but I completely disagree that fudging the difference between "reported hate speech" is reasonable
the internet is fucking full of botnets and armies, have you not seen how easy it is to flip the switch on a twitter account
you can tell sign (direction of the change) but you can't tell proportion without some kind of tally of the total.
for this to track you'd have to have a verifiable way to determine whether a comment was hate speech
EDIT: verifiable automated way
They can sample their reports queue.
It's not like my humidity sensor is perfectly accurate either.
and again I completely disagree that reports are a good substitute. bet you could buy 5000 reports for ten dollars
Okay, sure?
At this point you're talking about something very different from what you originally said though. Why not just say that they're probably measuring reports and you think those are bullshit? Instead of phrasing it in the form of a weird rejection to the concept of measuring things.
the fact that they dropped "reported" was the point I guess mainly because reddit admins are lying assholes
Catch me wildin out with the total probability denominator of the Bayes law term.
What's the total probability?
I dunno, 1, I guess.
-The entire branch of statistics I'm most familiar with.
deleted by creator
Maybe they measure it by the hate speech complaints they receive
heh I just edited with the same thought
even if they did that... why would they not just "reduce verified hate speech by 100%" that is, hate speech which was reported and then verified.
that they couldnt even delete one out of five reported hate messages (because they couldnt verify?) is hilariously bad too