Hacker News Reader

Comment by WheatMillington | Hacker News Reader

This justification makes zero sense to me. Why would they need to use rate limits to detect bots? They can trivially observe account behaviour without using rate limits.

View on HN

It's corporate talk for "we f up and systems are failing". Musk's act of not paying people now bites him back. I guess they don't have enough servers to sustain the service.

The new "you need to log in to view this tweet" page has a bug that causes it to repeatedly try to load the tweet, about 10 times per second. This explains the increased server load.

I'm pretty sure this is provided solely for the benefit of Musk's weird fans, who can now go "see? Dear Leader broke the site as part of the 4D chess!" Normal people probably shouldn't take much notice of it.

About the only thing I could think of is, limit the rate and see what accounts keep trying to post even though they fail?

But there would be so many false positives on that idea. Would punish many that are legitimate users and probably harden those that have more sophisticated posting techniques.

Like you said, there are better ways.

> Why would they need to use rate limits to detect bots? Accounts that hit rate limits are more likely to be bots.

Any ideas on how to do this more trivially than toggling a global rate limit flag?

The post itself says that publicly announcing the policy in advance would be detrimental. The policy is now public and they clearly notify the offender of violation. It probably took bot maintainers less than an hour to update their scripts to pause attempts/switch accounts on first failed request. It's the humans who keep trying. Any heuristic they'd want to apply to detect bots would be applied to action history, tuned to acceptable false positive rates and silently rolled out. This circus doesn't align with the statements at all.

Turing test.

Reply on news.ycombinator.com