pull down to refresh

70 sats \ 7 replies \ @k00b 8 Apr \ parent \ on: Stacker Saloon

To be clear:

  • I like the auto-item-mute feature idea
  • I don't like plutocracies deciding content
  • Without some kind of collective memory or a constant zap war being waged
    • n people will need to pick up the same m pieces of trash
    • folks who are logged out or lazy or whatever will see lots of trash and leave

The benefit of SN is the commons. It's a level playing field - you don't need to be an influencer to influence - just someone that shared something valuable to someone(s). That becomes less true as:

  • a plutocracy decides content, or
  • it turns into a trash heap that non-trash doesn't want to inhabit/visit

imo the commons needs some memory, or zap militants, to avoid becoming a trash heap. Neither demands a plutocracy - just something clever in terms of incentives, signaling, or mechanism that I'm unaware of.

write
compose
145 sats \ 6 replies \ @optimism 9 Apr

So I don't think that zap wars are the solution because wars consume. Perhaps trust/memory can be reformulated/reconceptualized as reputation?

reply
92 sats \ 5 replies \ @k00b 9 Apr

reputation is a form of memory, the way I mean it, but that's not important

what's the important difference between trust and reputation to you?

reply
146 sats \ 4 replies \ @optimism 9 Apr

I'd say that most distinctively trust is a belief, so it needs a starting point (iirc that's what we had - trust flows from central points.) Reputation, how I see it, is more factual and measurable. Its not based on a belief but on merit. You build and lose it over time, but its not hierarchical. So your reputation doesn't have to influence mine in our interactions.

I understand that that doesn't solve the Sybil problem on its own though. I think John had written something about that not too long ago, but that was localized signal instead of globalized - and inherently still based on trust because of a LinkedIn style "connection of connection" legitimization. I don't think that that is suitable for a globalized "commons" approach, because the bias starts where the trust starts.

So I'm thinking a matrix more than a tree?

reply
92 sats \ 3 replies \ @k00b 9 Apr

I get you.

So an example of reputation might be: your zaps/boosts have limited power until your account total_earned_zaps - total_earned_downzaps > 100k?

And that's better because that's not as subjective.

reply
145 sats \ 2 replies \ @optimism 9 Apr

Maybe! If it decays.

With evergreen features, the best posts ever written would still generate reputation through the years. But full glory must belong to the young (just not too young)

reply
145 sats \ 1 reply \ @SimpleStacker 9 Apr

Maybe a good time to revive some old collaborative filtering ideas I've had?

https://github.com/stackernews/stacker.news/discussions/2648

These methods allow you to construct similarity scores between posters and and between items, and individually preference rank items by posters.

Doesn't solve the problem of people with no data behind them, but their rankings could be some social aggregate using people with more data.

reply
68 sats \ 0 replies \ @k00b 9 Apr

If we start experimenting with personalization again, I'll for sure seek your input. Right now at least, I'm biased toward proving/ruling out solutions that are explainable to a five-year-old that provide a relatively raw commons worth inhabiting.

reply