alt Hacker NewsAlright, thanks. I read your explanations and they do answer some of my questions.
I'm still surprised that the percentage of this activity here is so low, below 0.1%, as you say. Given that the modern internet is flooded by bots—over 60% in the case of ProductHunt as estimated by the article, and a third of global internet traffic[1]—how do you a) know that you're detecting all of them accurately (given that it seems like a manual process that takes a lot of effort), and b) explain that it's so low here compared to most other places?
[1]: https://investors.fastly.com/news/news-details/2024/New-Fast...
Replies
intended's reply is correct.
Most of the bot activity we know about on HN has to do with voting rings and things like that, people trying to promote their commercial content. To the extent that they post things, it's mostly low-quality stuff that either gets killed by software, flagged by users, or eventually reported to us.
When it comes to political, ideological, nationalistic arguments and the like, that's where we see little (if any) evidence. Those are the areas where users are most likely to accuse each other of not being human, or posting in bad faith, etc., so that's what I've written about in the posts that I linked to.
There's still always the possibility that some bad actors are running campaigns too sophisitcated for us to detect and crack down on. I call this the Sufficiently Smart Manipulator problem and you can find past takes on it here: https://hn.algolia.com/?dateRange=all&page=0&prefix=true&que....
I can't say whether or not this exists (that follows by definition—"sufficiently" means smart enough to evade detection). All I can tell you is that in specific cases people ask us to look into, there are usually obvious reasons not to believe this interpretation. For example would a sufficiently smart manipulator be smart enough to have been posting about Julia macros back in 2017, or the equivalent? You can always make a case for "yes" but those cases end up having to stretch pretty thin.
From what I understand - users accuse others of being shills and bots, and are a largely wrong.
Dang and team use other tools to remove the actual bots that they can find evidence for.
So yes, there are bots, but human reports, tend to be more about disagreements, than actual bot identification.