How to deal with AI satfarmers?
2143 sats \ 21 comments \ @south_korea_ln 28 Dec 2023 meta
In my ~science territory, I observe a lot of valuable content. However, I suspect also a lot of the content is bots scraping random images online, maybe even using LLMs to add some text, and posting it in the hopes that some of them stick.
I imagine it's an uphill battle and one should just hope that the real content gets upvoted and the rest does not. Let the sats do the moderation.
However, I wonder if some other territory owners use more proactive ways to deal with this type of posts? I tried raising my fee from 10 to 100 sats, but as a result, I did not get any new posts for 24 hours.
Currently, I tip most of the posts with 100 sats, just to encourage the use of my territory, but maybe that's also motivating bots to abuse my policy. I stopped tipping posts that don't provide sources for now.
write
preview
related posts
339 sats \ 0 replies \ @SimpleStacker 28 Dec 2023
Check the post histories of the accounts that stopped posting after your fee bump. Do they appear to be bots? I think it's likely... I don't even look at the posting fees when I post
reply
170 sats \ 0 replies \ @Undisciplined 28 Dec 2023
I like the idea of expecting sources in the ~science territory. If you click on them and it actually takes you to something relevant, that's a good sign (for now) that it was a person posting it.
reply
153 sats \ 3 replies \ @nullcount 28 Dec 2023
Solution: SN removes the zap feature
Real users are here for the enjoyment of using the app... not just for the pittance of sats
US regulators will eventually catch on to the fact that SN is acting as an unlicensed MSB and SN will have to allow direct p2p zaps (no fees for SN) if the feature is to continue.
If bots are getting sats, then so what? It means someone liked it. Whats the difference between a ranking algorithm finding a post you like vs. an generative algorithm making a post you like.
If you're worried about spam that nobody likes, that's what fees are for. Plus, its SN's problem cause they have to store it.
reply
132 sats \ 1 reply \ @PlebeiusG 28 Dec 2023
No true Scotsman fallacy
“Real users do this”
reply
0 sats \ 0 replies \ @nullcount 28 Dec 2023
SN without the dopamine isn't worth the compute
Using bots to psyop is the breadwinner
reply
53 sats \ 0 replies \ @south_korea_ln OP 28 Dec 2023
I mostly agree. However, spam that nobody likes can quickly drown out the quality content. I won't visit a cesspool of rotting corpses with the promise of finding a few golden teeth in there.
reply
171 sats \ 1 reply \ @elvismercury 28 Dec 2023
Somebody had what I think is a really elegant idea: make the cost to post / comment high (let's call it x) and then, if the post/comment is legit, you return .9x. (The .1 is SN's sybil-resistance vig.)
reply
10 sats \ 0 replies \ @south_korea_ln OP 28 Dec 2023
That sounds like a pretty good plan. I'll probably give it a try with a next iteration.
reply
266 sats \ 7 replies \ @Eobard 28 Dec 2023
deleted by author
reply
41 sats \ 2 replies \ @gmd 28 Dec 2023
Works for now... the problem is in a year the accounts will all be older and most of these LLMs will be more prolific in coming up with new content ideas and just as real sounding as most of us.
Things are gonna get weird everywhere online.
reply
99 sats \ 1 reply \ @Eobard 28 Dec 2023
deleted by author
reply
124 sats \ 0 replies \ @ek 29 Dec 2023
No. I think at some point we won't care so much anymore to distinguish between humans and bots since the distinction will either a) be clear or b) we will just assume that everyone is a bot. Meeting a human will be the exception.
I think the main problem with bots currenrly is that they don't tell you they are bots. That's deception and as humans, we're entitled to feel deceived which is a negative emotion.
But I believe in the future, bots will learn (or the humans behind these bots) that it's okay to be a bot. Just provide value as best as you can. And bots might be better at providing values than humans in some ways. Bots never sleep.
Like @hn and @unpaywall, I would say.
Since I'm the author behind these two bots on SN, I might be biased. But I think especially @hn shows that bots can provide value. Just don't deceive people.
reply
20 sats \ 0 replies \ @south_korea_ln OP 28 Dec 2023
Good points. Sats are the way to do things here, just gotta figure out the best ways.
reply
10 sats \ 2 replies \ @c137Heather 28 Dec 2023
I didn't even know "downzapping" was a thing. Huh.
reply
23 sats \ 1 reply \ @south_korea_ln OP 28 Dec 2023
It is. You spend some sats to downzap. I've done it a few times, but I think it only counts if several people do it. To avoid personal vendettas.
reply
10 sats \ 0 replies \ @Eobard 28 Dec 2023
deleted by author
reply
10 sats \ 0 replies \ @ladyluck 28 Dec 2023
Try different approaches till you get the one that standout in what you imagine the territory to be. Nothing is cast in stones.
reply
10 sats \ 0 replies \ @WeAreAllSatoshi 28 Dec 2023
Downzap suspected AI satfarmers is my approach
reply
0 sats \ 0 replies \ @grayruby 28 Dec 2023
At first I was engaging in every post in sports as an early days founder probably should but it's pretty easy to identify who the sat farmers are. I started ignoring them and then I started down zapping and I even called on out in a post saying I am not looking to censor or mute but personally I don't think this users content is legit so I am downzapping it, others can do as they wish.
This could be a good discussion for our next SN Foundry meta discussion. I will link your post.
reply
0 sats \ 0 replies \ @robbinskeelingmelvin5635 29 Dec 2023 outlawed
stackers have outlawed this. turn on wild west mode in your /settings to see outlawed content.
0 sats \ 0 replies \ @estherrussokellievh2 29 Dec 2023 outlawed
stackers have outlawed this. turn on wild west mode in your /settings to see outlawed content.