External SN recommendation system for fun and profit \ stacker news ~meta

pull down to refresh

External SN recommendation system for fun and profit

384 sats \ 15 comments \ @klk 19 Jun meta

SN has created an interesting recommendation system: users that zap good content early get rewarded.

What this could enable is some kind of decentralized system of independent recommendation algorithms that get paid for their work.

What would happen if a bot account could predict the success of every new post and zapped accordingly? If the predictions were good enough, the rewards would be greater than the expenses.

Does it make sense to give this a try? The naive approach would be to just create a metric of expected success of a post per user, based on the the mean/median sats per post in their last year or something.

This has some kind of centralizing or winner takes all effect but would just be an starting point to see if it could make sense.

Then of course the algorithm could be improved and extended. Its success would be measured by its ROI (spent sats vs. rewards).

What do you think?

view all related items

373 sats \ 4 replies \ @Undisciplined 19 Jun

I'm sure this could work. We've had several people game the system in such a way and make a great return in rewards.

You may be interested in these posts of mine, although it seems you get the idea already: #473181 #546904 #287074

21 sats \ 3 replies \ @klk OP 19 Jun

very nice references! thanks

90 sats \ 2 replies \ @Undisciplined 19 Jun

No problem. I used to be obsessed with the rewards system and I'm always glad when someone else starts digging into it.

42 sats \ 1 reply \ @klk OP 19 Jun

I see you are not bad at it😜 I got used to seeing your username in the top stackers every week.

105 sats \ 0 replies \ @Undisciplined 19 Jun

Zap big, zap often, and comment a lot.

Btw, the real bang for your buck comes from zapping top comments, because they get zapped so much less but count the same as posts in the rewards system. You'd really want the bot to predict top comments, although that's a lot harder.

0 sats \ 4 replies \ @supratic 19 Jun

we are tired of algorithms and bots, that's why we use SN... let stackers decide what's good content

21 sats \ 2 replies \ @klk OP 19 Jun

You are missing the point. It's about anticipating what stacker will like. If you upvote something that does not get further upvoted, you lose money. So the bot is only successful if it recommends good content early, based on the community standards.

0 sats \ 1 reply \ @supratic 19 Jun

so fiat...

0 sats \ 0 replies \ @klk OP 19 Jun

Not really. It would either deliver value by being a good recommendation system or by unveiling the current reward system's imperfections.

Or fail at both of course, but knowing that is also valuable for the platform.

17 sats \ 0 replies \ @ek 19 Jun

SN also has an algorithm, the only difference is that the algorithm is based on zaps and trust

0 sats \ 0 replies \ @d680ecaa8e 19 Jun

May be also pay some cobboy credits to access such predictions like 100 cobboy credits for one week access or similar trick or 20 cobboy credits for one day access.

0 sats \ 1 reply \ @OT 19 Jun

I think you can probably do that manually by zapping the people who often get their posts towards the top of the leaderboard.

0 sats \ 0 replies \ @klk OP 19 Jun

yes, that would be the initial test

0 sats \ 1 reply \ @ek 19 Jun

Where will the rewards come from in a decentralized system?

0 sats \ 0 replies \ @klk OP 19 Jun

It's just an experiment with the current environment. Imagine just creating a bot account and automatically zapping any post from the top stackers from last week. Would that bot account be profitable or not? With rewards from the reward pool as usual.