nostr:npub1manlnflyzyjhgh970t8mmngrdytcp3jrmaa66u846ggg7t20cgqqvyn9tn nostr:npub1wqfzz2p880wq0tumuae9lfwyhs8uz35xd0kr34zrvrwyh3kvrzuskcqsyn nostr:npub1wmr34t36fy03m8hvgl96zl3znndyzyaqhwmwdtshwmtkg03fetaqhjg240 nostr:npub1nu9m6k6dca28f9humvq0ad334a3czu0qrevw0pd4ml2pzaumx0mqyr3ars
We need an "inner web" layer to WoT. We could call them "trustees". 🤔
I've thought about it this a lot recently. Just like in life, there are layers/levels to the amount your trust or want to pay attention to someone.
I've always thought it would be a cool UI pattern to have a dial or a slider for all feeds on Nostr, where you can control the aperture of the feed based on their WoT proximity to you, how much you trust them, or how much you want to pay attention to them.
nostr:note1t8s5u6kjgpvcg2w2m5wcvlmfaw06tj3vfta4p2rqn4qglfananeqdltrmv
Someone in Madeira told me that someone build this WOT slider interface already.
Agreed.
In my opinion, though, you should also be able to trust some *for something specific*, not trust them across the board. For instance, I may trust the entity "shitcoin bots tracking collective" to provide me with a list of npubs to block, or I may even grant it the rights to automatically block them for me. (At a later point, when I notice that the "shitcoin bots tracker" has become overzealous, or maybe has quality issues, I can revoke that privilege.)
This is different from trusting that entity unconditionally for everything, there should always be a scope.
Yup. Hard to do this in the digital realm in a way that people will actually do but you’re right, we do this sort of delegation of decision making based on trust all the time.
Yes, we do it intuitively.
Making it explicit is a difficult job, I agree. Because the abstraction needs to be simple enough that people actually want to use it.
People don't know how to rate products. Letting them rate how much they trust Alice fir X and Bob for Y is a UX nightmare.
Why ask users to do something (they can't) that competing algorithms could probably do a waaaay better job at?
Almost all data is public and users literally spend money on what they wanna see more of.
Even the most simple algo nostr:npub1l2vyh47mk2p0qlsku7hg0vn29faehy9hy34ygaclpn66ukqp3afqutajft could come up with after children bedtime would do better job then humans indicating trust levels ever could.
I’m not talking about a dial or slider for the trust you have in people; I’m talking about it for the feed. Basically dialing up or down whatever algo/heuristics you’re using.
I'd autonate that too.
Updating these things all the time sucks.
It's the sane problem with ratings in general. If you've only seem Mean Girls, think it's the best movie ever and give it a 10. The Lord of the Rings comes along and you have to update EVERYTHING.
I think you're missing my point. Here's the UX.
1. I'm looking at primal but want to see only my closest friends content. (Primal knows this based on WoT and other signals I've given it over time - it's on them to come up with an algo that makes sense - I'm not manually doing anything other than using the app).
2. I change a setting on the feed page (takes less than 1 second).
3. My feed updates to only show my closest contacts.
4. I decide I want to see lots of new content. I change the setting. My feed updates again.
This can be made very very easy for users. The heuristics behind the scenes are doing most of the work, which is why no one has made this yet. It's quite complicated.
Also, there is the obvious issue with cold starts - your filter is shit for a while until Primal builds up enough info.
Aha lol, then we mean basically the same thing 😂. Scuzi!
With that setting change being:
You selecting from competing algos that ideally any client can access (of course "latest from following" doesn't have to a DVM etc)
Trust aside, it's little too abstract concept by itself.
But I like the sliders as an idea for a simple personalization control for the feed algos. With it, you can signal to the algos
- give me more Gigi, fitness, and cooking content
- give me less Bitcoin, github, and finance content
- and give me zero Trump/Biden drama
It will offer more flexibility than a single synthetic algo personalizing your feed without your explicit inputs just based on your zaps, likes and some mysterious WoT score.
The problem is you are putting trust in a box.
There's no equation for trust. Web of Trust models rely on a BINARY relationship with trust.
Trust is not binary.
What you said is important- there are layers/levels to trust! It's clearly more than a binary relationship.
If you list out every user you encounter on Nostr - could you sort them into meaningful "defined trust relationships?"
Sure.. but what's the goal? To what extent are you defining the context of trust?
All of the variables you define as "trusted" are subject to change.
What I see is encrypted npubs as an identity layer. These are free to generate and offer no verification to create. That is inherently an untrusted network. No amount of Webs of Trust will change the level of trust users have in the network itself.
Webs of Trust should be decentralized. They should be defining context. They should rely on endorsements.
But they shouldn't put trust in a box.
We need to organize information for ourselves before we can curate it via decentralized methods.
We need to cultivate trust before we determine what trust algorithms should look like.
