Hacker News Reader

Comment by bombcar | Hacker News Reader

I wonder at what point the blocklist would surpass twitters storage mechanism for it.

Can you block every single Twitter account?

I’ve seen people on Twitter brag about having 6 figures blocked. So it at least works for that many. I know Marc Andreessen notoriously has a lot of people blocked. I wonder if it’s in the millions.

If someone spent 5 hours a day on twitter, and blocked a new person every 5 minutes... then after 4.5 years of doing this every single day, they would reach 100,000 people blocked.

As evidenced by this post, people use automated tooling that can easily make that number balloon.

Block lists. They exist, and there are tools to implement them.

Once every five minutes feels quite low in this situation. I don't really block, just mute for the most part. But I can go in and mute a dozen accounts in a minute in the comments on someone big's tweets. If you are someone with a high engagement rate, and actually want to curate the comments, you could easily end up blocking a hundred in a couple minutes.

Gotcha, makes sense. I'm not on twitter so I wasn't actually sure what to put for that.

Which is why you have a block-chain browser extension do it recursively.

Whenever I see this I wonder how much overhead blocking as a feature generates for twitter.

As a programmer I’d be more worried about the lookup cost. There are efficient ways to lookup a name in a list and very inefficient ways that would be fine for small lists, but struggle with large lists.

Maybe this is an ideal use case for a bloom filter. The answer to the question "is user X blocked?" is almost always going to be "no". And the "maybe" cases can consult the actual block list.

How exciting. I've waited my whole career for a bloom filter use case. Please no one tell me that there's a much more obvious solution.

This made me curious and I wondered how it might work...

The simplest approach would be to start from user id 1 and increment up blocking each one but then you'd block millions of inactive, dead, or suspended accounts which would probably get you suspended.

A better approach may be to choose a "recent" tweet id, block the author, count up from there, checking if each author is blocked and blocking if they're not. The best part is that you don't even need to know the author when you make a request. For example, tweet id 1508540042817376256 is Elon's and it resolves to him whether you use https://twitter.com/elonmusk/status/1508540042817376256 or https://twitter.com/Microsoft/status/1508540042817376256

This is no longer blocking "every single Twitter account" but blocking "every single active Twitter account" but may still be able to trigger whatever limits.

Compressed bitmap indexes are the way to do this in theory. Very efficient storage.

Reply on news.ycombinator.com