Oddbean new post about login | logout Web of Trust (WoT) is a really hard concept to communicate. I've been thinking about it a lot and trying to design something that most people can understand, and frankly there's no easy solution. The best thing people can understand is a word. A label. This is why I think the best thing to do is just to show how many people that you follow, also follow the person you're looking at. And if you already follow that person, show an icon to indicate as much. Even this icon will be confusing to some people, but people will get used to the meaning. Numbered scores, colors, or anything else we decide to use will only be misinterpreted. Ultimately, I think WoT indicators other than a textual label are going to do harm to the user's perception of someone else, or to the person who sees that they are not an ideal number or color on someone else's screen. People are not numbers or good / bad colors. We should probably never try to categorize ourselves as such.
Is this kind of like what the coracle client implements in a way? @hodlbod
Yes. It’s a noble attempt but ultimately I think numbers are not the way to go with wot display
You can disregard my above comment I saw the original thread bringing it up. I think it's a cool attempt especially for new people who get involved because of a couple of trusted connections. I can see how the issues you mention can turn it though. I essentially ignore it when I use Coracle. There's other functions on it that I'm starting learn about though. I primarily use Amethyst for nostr though
Are you talking about the circle with a number inside? If yes the only reason I know what that could be off because I saw this discussion, if not I would not have any ideia what that was
I like Coracle’s implementation and I understood it right away
But you have people wondering why they are low numbers
The algo is easy to understand imo. Low number either means the people you follow don’t follow you or a lot of them mute you or both. Could also be a bug, but numbers usually look reasonable to me.
We’ve tried to explain it in our new page about the @Tagr-bot bot for labeling. https://nos.social/tagr-bot Hopefully the posts and about page will help understand what it’s doing and the way moderation and trust works in Nostr with a WOT. Users can CHOOSE to add bots to their web or mute them.
What if we all unfollow Rabble and mute him and his bots, so they get a low score 😂 I get the score but idt most normies would. Karnage's method of simply stating X person and X others follow this person method seems more straightforward
That’d be perfectly fine. The whole point is that the ratings and labels should be subjective from your perspective in the social graph. None of us should have a single set of labels or web of trust score. Instead everybody looks at the network from their own perspective. If you hate me and what i’m doing, mute me, label my identity and content. Then if folks trust you and your views, they’ll probably agree with you. The entire point of this is that single centralized model of trust, safety, and moderation is flawed. Nostr needs to work with many communities who fundamentally have different ideas of appropriate norms. We need a system that can handle that on the same network and protocol.
Well I don't hate you lol. Can't say I'm a fan of the moderation, or agree with the methods used, but you're doing the work and I respect the effort you put in.
:-D We do also show mutual contacts in Nos because we feel that is the fastest way to see if you’d be interested in this person’s content. https://image.nostr.build/1994974e3ee4e4e6cb00131c5119f254b9783bb80e2d194cf1fbe94fa29042f5.png https://image.nostr.build/f04d03a5d14fc643376d4f736ee6bf98a0584d50c79e4612fce4f25462cac2ec.png
Nice, I haven't tried Nos but now I will. This method makes sense to everyone imo
Right and that's the thing - if you assign scores, a low score on its own isn't not a bad thing, but when people see that they have a low score, they think somehow that makes them less trustworthy. One of the points I'm trying to make is that we shouldn't use anything that gives people a negative perception of themselves or others. A simple "Followed by X and Y" (people you follow) is better than assigning scores. From there, you can decide if that person is worth following or is trustworthy.
We often use the concept of Six Degrees of Separation to explain the Web of Trust to friends, which leads them to visualize a social graph. https://en.wikipedia.org/wiki/Six_degrees_of_separation
what does this look like in practice? (UI)?
The entire Nostr social network can be seen as consisting of first-degree friends, second-degree friends, third-degree friends, and so on. When you want to determine if a newcomer is trustworthy, you can do so by calculating how many degrees of separation there are between you and them. This is our vague idea, but we are not sure how to create an intuitive UI for it.
Never thought at it like that. If just one friend of mine would suggest a film I would watch it even if the score in some site is low, because I trust his suggestion more then the ones made by others I do not know. The system you suggest is like the one on Instagram at the moment. So we know it could work. But here we are daily looking at profiles that maybe nobody follows, so you still need something for them too. Maybe you'll get different algos give different scores.
Yup 💯, there's two things you want to know: 1. Is this profile known in my network? 👉 Show how many of your Follow list follow him, or go one degree beyond. 2. Is this a possible imposter? 👉 Show warning if there are concerning matches in Name, Profile pic, Bio, Website, Npub last/first symbols, Npub color, ......
I don't think this is a foregone conclusion. Numbers measuring proximity are useful. I.e., I'd rather know how many miles from my house someone lives than have google tell me travel time will "require a vehicle". People misinterpreting WoT numbers (which are a reasonably novel concept in social media, though not unprecedented) is a problem of importing popularity-concept metrics into a new medium. There are many ways to think about this problem, and I think you may be leaping to overly-opinionated solutions too quickly.
I agree this is early to tell if a number is useless. But it can definitely be misleading. A distribution of those scores will most probably be power law - top guy having multiples higher score. To an average person it would seem as if top guy is 3x "better", while that's definitely not the case. How and when the numbers are presented is probably a very important question.
