Help me figure out sofiechan poster vouching

Sofiechan's backend technically has the capability to consider not just a poster's posts and votes, but other user's direct assesments of that poster. That is, if you were to vouch for someone, the algorithm would take that as evidence that they are solid, not a spambot or idiot. If you blackballed them, the algorithm would assume the opposite, that they might be a problem. At least it would if it trusted you on these things. The only problem is the UI currently has no way to register these vouches.

With the addition of persistent named nyms, there is more of a plausible usecase for vouching, and somewhere to put the UI. If you know someone is good either by reading their posts or because you know them IRL, you could vouch for them on their profile page if we added that. The most obvious use-case is to give your smart friend a leg up in the algorithm.

But the best signals are those that actually mean something to you, have legible consequence, and which you endorse as a way to express that kind of meaning. Does vouching clear that bar? That's less clear. Because sofiechan is anonymous by default and people could be posting under different names, and because we're avoiding personalized feeds for now, there's no obvious way do things like "follow". Follow would immediately make vouches more meaningful to you beyond a bare popularity vote, but its unclear how to do it.

Without going to a full-on personalized feed, we could do a limited form of "follow" where the system points out where people you follow are posting, without revealing who they are. Some threads would just be marked with a "your friends are posting here" mark. Specific posts could also have a "your friend posted this" mark if that could still be sufficiently anonymous. But these are of unclear value at this stage. I think they become more significant when there's too much going on to read all of it, in multiple tags. What do you think?

Another thing that might be fun is a sort of semi-anonymous matchmaking. If you vouch for each other, you're mutuals, get a match note, and can anonymously dm. Meeting the right people from the internet is fun and productive for all sorts of things, so this could be neat. It creates a direct incentive to get noticed as a good poster and to vouch for good posters you would like to talk to. But really I'd like the vouch to reflect an ongoing curation judgement, not just an initial curiosity, so we probably also want something like follow that expresses ongoing interest. Any ideas?

I've also considered doing a sort of "trust" semantics to the vouch where if you vouch for someone, that authorizes the system to reveal something about you and your posting to them. Maybe "your mutual posted this" only applies to mutuals, so the anonymity problem is less (random third parties can't stalk you that way).

And finally of course vouching for someone ties your reputation to their reputation to some degree. If they turn out to be an idiot, maybe you are an idiot too. If they turn out great, you must have good taste, etc. But this won't be a significant driver of vouching in itself, just something to push you towards good judgement on the margin, and for us to track who has good taste.

As for the UI, I envision a simple pair of "endorse"/"ban" buttons (or "vouch"/"blackball" or "follow"/"block" etc) maybe following the usual checkmark/xmark motif next to or under their name on their page. I don't think that's to intrusive or busy. But I'd like to get some more thoughts on this in case there's a better way.

Once I've got a solid direction here I'll add the UI component to the profile page and start building out the various consequences like follow and match if those are right. Would appreciate your thoughts.

I think the semi-anonymous matchmaking is much more compelling. The point of an anonymous board is the fundamental equality of posters and lack of status, but that eliminates the potential for building friendships with intellectual peers. So I think rather than thinking about it as "vouching" it should just be closer to "I would want to have a conversation with this person" and then if they do the same to you, then you enable the anon dm's and maybe mark that person's posts with a "mutual" tag that just links to a dm.

Then you keep the network aspect mostly private without an algorithm weighing these relationships for anything else.

Then the negative equivalent should just be a "mute" button so that if you really hate someone's posts, you don't have to see them.

If you really feel like a curation aspect is key, then I would posit a notion of "rapport" is better. So this would be how many times they have interacted either publicly or privately, with the idea that if you continue to engage with someone, you probably want their stuff appearing more frequently. That encourages each user to engage with what they want to see more of.

Just some thoughts though, feel free to disregard if I misunderstand your goals.

>>2623
Yeah I'm getting more excited by the matching aspect the more I think about it. As for algorithmic inequality, I don't think I believe that the value of an anonymous board is equality per se. It's rather that you are insulated from third party censure. No one knows you're a dog, etc. I think the value is the ability to speak freely among peers. But this requires some way to ensure relative peer status and peer presence. The unfortunate fact, especially going into the second quarter of the 21st century, is that the vast majority of entities capable of posting coherent english text online are not your peers, and do not act as gentlemen.

I remember the exodus in 2014 from 4chan to 8chan after moot banned gamergate. It was amazing; suddenly there were no bots, shills, and retards, just peer anons talking about politics etc at a much higher level of intellectual quality. It didn't last of course, but 8chan was briefly one of the most interesting places on the internet. My conclusion from that was that actually community selection really matters and should be optimized by any means necessary.

Of course we don't want an anonymous discussion board to turn into some kind of gamed-out credentialist popularity contest, but my view is that if we gather a large number of independent lines of evidence, including people's assessments of each other, we can quickly and accurately identify which posters are real people and make sure they are happy. I think if the platform is smart enough and uses enough different lines of evidence, the hostiles and strivers won't be able to keep up relative to the intended poster community, but we can keep it easy for a smart unknown 17 year old gentleman to come in and post.

>>2623
To be clear, most of the algorithm is already and still what you are calling "rapport". That is of course the backbone of anonymous reputation. But let's talk about an anonymous "mute". What could that mean? We could make it so that you literally just never see posts from that user again, but that's a pretty big hammer, and you won't know what you're missing. But then also you could cross-reference that with an incognito tab to de-anonymize all that user's posts. Oops. We could fix the not knowing what you're missing problem by showing it in the usual hidden format, but that's even worse from an anonymity perspective. Instead of by user we could do it by nym so that there's no anonymity problem, but a typical nym only has a few posts in a single thread, so you can just as easily hide them all manually. Still maybe that's a viable minimal consequence to make it mean something and the real effect is using your clout to punish that user in the algorithm. We could also de-emphasize threads that have muted people posting in them, but tbh if your muted people are posting regularly in threads you would otherwise see you're probably misusing the mute/blackball button and are getting muted out of the network yourself. I think mute is going to be inherently limited in personalized utility except as a political signal.

>> But the best signals are those that actually mean something to you, have legible consequence, and which you endorse as a way to express that kind of meaning.

The way to solve this is to make posts saveable as inscriptions on bitcoin ordinals, like they have done on Ark.page

https://ark.page/archive?url=https%3A%2F%2Fark.page%2Fraw%2Fb3fe3dde.txt

Then patrons can buy the inscriptions and this serves as the value signal.

There wouldn't be any incentive to buy an inscription that was less than the longest term enduring relevance, therefore Truth becomes the organizational principle.

IMO it has to be built on Bitcoin inscriptions, on an immutable peer-to-peer network, because the medium is the message, the underlying medium affects all of the incentives of the platform

>> Another thing that might be fun is a sort of semi-anonymous matchmaking

I also think anonymous reciprocal vouching would be the most parsimonious solution

Given the long history of astroturfing, vouching needs to be granted - and can also be revoked.
https://tnl.net/blog/2025/02/08/your-fake-online-friends-have-a-2000-year-old-history/