pull down to refresh
1503 sats \ 4 replies \ @elvismercury 3 Dec 2023 \ parent \ on: deleted by author bitcoin
It's an interesting problem. It's useful, I think, to consider what this looks like with 10x or even 100x the scale -- that's the problem that's coming. Some ideas:
-
People have really got to step up and start zapping the kind of stuff they want to see. It's the superpower of SN, people should make better use of it. All of this absurd zapping with 1 sat is ridiculous.
-
Other types of PoW could be brought into play. I know some like it, but I find such little value in random youtube links w/ literally no context. The bar is low to do a lot better.
-
Sub-boards, or whatever the term is, should help a bunch, assuming that they work at least kind of how I imagine, e.g., someone / some group is the admin, can have their own rules, can moderate hard with a particular vision. If you like their vision, join the sub. If you don't, don't.
-
Antes / stakes for posts and comments -- 10 sats is probably still too low. You could imagine this being adaptive, e.g., well-regarded stacker, 100 sats. Random newb with fresh account, 1000 sats. Different kinds of participation (commenting, reading, @-ing other users) lowers the stakes.
-
Reputation matters -- a zap from someone well-regarded is worth more in amplifying that thing. This is what PageRank was meant to solve, except you have an even stronger quality signal on SN -- zaps are more nuanced than simple hyperlinks.
-
Many of the above amount to implicitly figuring out what I should see based on what I've done in the past, combined with what other people have voted for. Could add an explicit element, too: basically, Twitter follows.
These all have tons of failure modes and edge cases, obviously. It's a hard problem. And really, there's a fundamental issue that can't really be ducked: there's tons of garbage on the internet, and people try to exploit the system in any way they can. SN will have to be robust to that. But at least the tools are there to make it possible.
Obviously, I agree with your first point. I was planning on making another post about that soon.
I actually edited out a similar idea to your second point, but the way, but I wasn't sure any of my ideas would make things appreciably more difficult for bots. I was thinking that the criteria for earning cowboy hats could be randomized over a range of desirable SN behaviors, much like the daily rewards criteria have been randomized.
Definitely subs. The only reason I didn't mention it, is because I knew you would.
The concern that has been creeping up for me about pursuing the reputation angle, is turning Stacker News into a closed community, with a sort of caste system.
reply
I actually edited out a similar idea to your second point, but the way, but I wasn't sure any of my ideas would make things appreciably more difficult for bots.
I think that's true, esp in the era of LLMs; but consider it this way: the distinction btwn "bot" and "human" is not the operant distinction -- it's "signal" vs "noise". Bots could get around a bunch of trivial PoW mechanisms that we're used to. That's inevitable, if not now, then soon.
But right now, today, a smart bot could do better than a lot of stackers do when they spam all these youtube links with basically no description, no summary of why I should care, not even copying over the youtube description text. I welcome a bot that does better.
More briefly: if a person can't do better than the bot, give me the bot.
The concern that has been creeping up for me about pursuing the reputation angle, is turning Stacker News into a closed community, with a sort of caste system.
I don't view them as exclusive -- everything is choices. If you want the full firehouse, it's available, but you're going to pay a price for confronting all the noise of that, in addition to being early to the gems. If you want to subject yourself to the caste system, there are virtues to that, too. Except in this case, the caste system is likely to be more meaningful than in real life.
reply
The caste system concern is just kind of lingering in the back of my mind, it's definitely not an argument against anything in particular.
You're right about this being about signal vs noise and I think it can be even more broadly be described as "good for SN vs bad for SN". Obviously, there's a lot of overlap between those two, but as we've discussed elsewhere, there are lots of ways different people bring value to SN.
Ideally, the only way to profit from using SN would be to add value to SN first. As you say, "if a person can't do better than a bot, give me the bot."
reply
I think it can be even more broadly be described as "good for SN vs bad for SN". Obviously, there's a lot of overlap between those two, but as we've discussed elsewhere, there are lots of ways different people bring value to SN.
This is a good reminder to me -- I often fixate on "what would bring the most value to the version of SN that I want to see" which is obviously a much narrower window than would result from the objective you listed.
It will be interesting to watch, over time, how these different types of value reveal themselves.
reply