You sure about that?

https://youtu.be/WEc5WjufSps?t=193

Dr. Egon Cholakian sends its regards. That is to say, the bots are getting good. LLMs made this technologically easy a few years ago, it would take a couple years to develop and deploy a sophisticated bot network like this (not for you or I, but for an org with real money behind it that timeline is correct) and now we are seeing them start to appear. The video I linked is proof that bots already deployed in the wild can take 40 minutes of dedicated effort from a capable suspicious person to identify with high conviction. Maybe it would have taken you 10, I'm not hear to argue that, but I am here to argue that it is starting to take real effort to identify the best bots and this is the worst they will ever be.

I don't care how smart, capable, or suspicious you are, within 3 years it will not be economical for you to curate your list of non-bot contacts on the basis of content (as opposed to identity).

Well on my "for you" page, I also follow a pretty niche audience of tech people, which helps :P They're a little easier to verify since they generally also have blogs, or websites, or github accounts, or youtube videos, etc that help verify they're not bots.

I also think people create bots for some purpose -- instability, political divisiveness, financial gain, etc. And I'm kind of inherently not using twitter for any of that. I don't think I could find an account on my twitter thread that mentions the word "liberal", "trump", "conservative", or any of that if I tried! I agree that's a muuuuch more likely place to find bots. What sort of bots do you notice the most in your twitter?

I have a theory, which accounts like yours would be very interesting for.

Instead of looking at it as a per user basis, if you look at it as a network or ecosystem, the issue is that the network is being flooded with spam.

Since nothing happens all at once, over time different filters will get overwhelmed and eventually impact the less networked accounts.

It would be VERY interesting to find out when, or if ever, you begin to suspect some accounts you follow.

Yeah I suppose if you are already vetting based on identity from outside the network that probably does scale. Most people aren't as careful about this as you are, though, so it'll still be a problem and it will have to get much worse before it gets better.

I'm not on twitter. I left when the tidal wave of right-wing spam started to outweigh the entertainment value of seeing Yann LeCun dunk on Elon Musk.

That's the Darwin's theory for bots: only the fittest survive on the Twitter lands.

Using the “for you” tab is the only way to use twitter these days. Their suggest algo is complete garbage. I spent a couple days trying various ways to train it and still I got was complete garbage, so I accepted reality that twitter doesn’t really have an algo for the feed, just a firehouse of crazy people and engagement trolls

You mean the "following" tab, right?

Lol you’re right, I must have not had enough coffee at that point in the day.

Oh shoot I flipped it!! Darnit, thanks for pointing that out. I also meant the following tab on my post!

> Using the “for you” tab is the only way to use twitter these days

The first 20 posts of my "for you" tab is Elon Musk, then it goes on to show me more useful content. I am wondering if following him or blocking him will make any difference.

I meant following and I just hadn’t had enough stim chemicals in my system when typing that.

I have an account I purely use to follow other accounts. I haven't posted anything aside from a "so, this is twitter then?" years ago.

I get multiple bots requesting to follow me every day, and maybe 10% of my "for you" timeline is right-wing political "discourse" engagement bots, despite never having followed or interacted with anything similar, aside from slowly increasing my block list when I see them.