pull down to refresh
847 sats \ 13 replies \ @k00b 28 Mar \ on: Some thoughts on SN rewards meta
What's fun is that we don't use any of these metrics directly. The only thing we directly measure are aspects of spending behavior.
- Your spending on items that other people who have historically spent well are also spending on.
- Your creation of items that people who have historically spent well are spending on.
It is a Keynesian Beauty Contest. But the whole point of the algorithm, at least as it's currently constructed, is to incentivize creating and surfacing a consensus of beauty.
In future algos, I'm interested in exploring more localized consensus, where rewards come from cliques of people you trust/people who trust you. As long as I'm fantasizing, I also want territories to be able to experiment with reward rules of their own making because I think any person is unlikely to have the single best idea for every territory.
As long as I'm fantasizing, I also want territories to be able to experiment with reward rules of their own making because I think any person is unlikely to have the single best idea for every territory.
You could imagine allowing territories to provide their own distribution function -- you give access to some kind of daily data bundle (computed by SN) and the territory can, in return, provide you with a JSON blob of reward allocation over users, the results (presumably) of their computation over that bundle. (This seems sort of ornate, vs the territory owner providing a function that SN uses to compute, but security risks inherent in that seem prohibitive.)
reply
I love the KBC metaphor.
Casting the reward function in that light then pushes the design space down a level -- it becomes a different grounding problem -- you define beauty based on the consensus of the users you've attracted. But you're still attracting users. Those are still choices that the definition of beautify is grounded in.
You could find that you've made choices that result in a curious definition of beauty, for good or ill.
reply
You could find that you've made choices that result in a curious definition of beauty, for good or ill.
Ah yes, it's all a bit semi-supervised in that I did a lot of data labelling early on to kick start the whole thing.
The way I've thought about this so far is that so long as the group of "people who have historically spent well" can change, ie the historical input to the algorithm is a recent window of time, the definition of beauty can progress.
The cliques are already here alive and kicking, you just have to spend any amount of time in certain territories to see it first hand. Certain members merely have to post any old comment to get zapped, they've already got their followers. Which in turn allows them to game the system.
The mere mention of cliques makes me feel sick, it's disclusionary, the try hards will try harder to suck up to the clique members and how will people get a trust rating?
Territories experimenting with their own reward rules will just make it the wild west. Those with the best reward incentives will get the most traffic forcing other territories to follow suit. It's another race to the bottom.
reply
Certain members merely have to post any old comment to get zapped, they've already got their followers. Which in turn allows them to game the system.
Building relationships and earning people's favor is a pretty virtuous "game" isn't it? What alternative game would you like to see played?
It's another race to the bottom.
I think it's a race to the top, ie "the best reward incentives will get the most traffic." Should the worst reward incentives get more traffic? I know you don't mean that but I'm not sure what you do mean.
reply
earning people's favor is a pretty virtuous "game" isn't it?
I call it sucking up, brown nosing, kissing arse, virtue signalling 😜😂 But, seriously, that's the groundings for a social credit system. I think what this showing me is that myself and a handful of other people on here are truly ungovernable 😂
reply
The word clique filled me with dread too
reply
I get what you're saying, but this seems like one of those things where everybody agrees that 90% of the internet is shit, but they can't agree on which 90%. Which is the difference btwn your description being total pathology (for me, this would be a btc maxi echo chamber screaming the catechisms at each other) and being the sign that things are working out great (a bunch of people who say interesting things being perpetually rewarded for saying interesting things).
Territories experimenting with their own reward rules will just make it the wild west. Those with the best reward incentives will get the most traffic forcing other territories to follow suit. It's another race to the bottom.
Oh wow -- for me, that sounds like a race to the top! An ecosystem of innovation.
reply
more localized consensus, where rewards come from cliques of people you trust/people who trust you
How would that work?
It sounds like it has the potential really accentuate what already happens naturally, create exclusivity, plus alienate and make it difficult for new people joining.
reply