Nostr Web Client

Hey nostr:npub1r0rs5q2gk0e3dk3nlc7gnu378ec6cnlenqp8a3cjhyzu6f8k5sgs4sq9ac and nostr:npub149p5act9a5qm9p47elp8w8h3wpwn2d7s2xecw2ygnrxqp4wgsklq9g722q, I'd love to have a call with you both (and any other designers that are interested) to discuss some social graph stuff I've been working on lately.

I think it's important, especially when it comes to human dynamics, to start with the experience and design and then build backwards, as opposed to creating some kind of artificial trust score and then "making it pretty on the UI".

So, I'd love to learn from you.

Reply to this note

Please Login to reply.

Discussion

Niel Liesmons 1y ago

Always open for a nostr:npub1us9uc0sjjg0vyvh7vefgu2a96hx5up5gujl6pqays65hlk4de6hsnwq4g0 on this 😉

HoloKat 1y ago

What do you have in mind?

Pip the WoT guy 1y ago

To expand on what nostr:npub1wf4pufsucer5va8g9p0rj5dnhvfeh6d8w0g6eayaep5dhps6rsgs43dgh9 is doing with zap.store.

Instead of showing just 2 hops (user --> Odell --> signer of the App) it's possible to show also 3 hops, or more.

But there will be many paths, and so I want to understand how to sort them in a way that makes sense.

Also, is it the path that is important (user --> Odell --> Karnage --> signer), or is more important the last connection (Karnage)?

How to show these things in a way that solves the problem, which is :

Giving the user clues so he doesn't get fooled by impersonations.

HoloKat 1y ago

Sounds like the number of hops is important as that may indicate more or less trust. More hops = less trust from the person looking at the thing.

UI wise I imagine a slider for network “distance” the further you slide it, the weaker the trust.

But we probably want to factor in the total number of people into that calculation so the more people use this app, the stronger the trust (despite distance).

Perhaps it can be termed between trust distance and cumulative connections.

Seems like there may be other ways to do the UI then .. have to think on it.

Pip the WoT guy 1y ago

This is the user flow I have in mind right now.

- user clicks on the thing (e.g. App)

- user sees few familiar faces that follow the thing or signer of the thing (e.g. signer of the App) and the text "and 458 other you know follow the App"

- if that's enough info and the thing isn't critical (e.g. Note taking App), the user downloads it

- if the user wants more info (e.g. Bitcoin Wallet) the user expand on the view to see if the friends / friends of friends that are knowledgeable in this stuff (e.g. NVK, Odell, Gigi, Pablo) follow the thing.

This is radically different from NIP77. I think it's natural for people to think of 5 people they know that are good in a particular topic (photography, wallets...), and it's NOT natural to rate people you know on a scale from 0 to 100 publicly.

I hate fucking scores, people are not numbers.

cc nostr:npub149p5act9a5qm9p47elp8w8h3wpwn2d7s2xecw2ygnrxqp4wgsklq9g722q nostr:npub1wf4pufsucer5va8g9p0rj5dnhvfeh6d8w0g6eayaep5dhps6rsgs43dgh9

Pip the WoT guy 1y ago

I think the direction that makes most sense is "show don't tell".

Show relevant people, so the user can recall an incredible amount of social clues and make an informed decision at that particular time.

Don't tell, this is the most trusted in this bla bla.

Imagine you go to a conference, meet hundreds of people, you have thousands of social clues in your subconscious. When you go home, are you going to sit down, and start giving ratings to everyone you met? That's insane

HoloKat 1y ago

I’m probably the wrong person to ask because I actually prefer editorial picks over WoT. I trust the opinion of one professional over “50 friends use this app”

In App Store I go straight for award winning / editor picks. 🤣

But if I had to look at wot stuff then “also use this app” would be a good metric.

Pip the WoT guy 1y ago

My idea is more for avoiding impersonations, rather than giving value assessments.

franzap 1y ago

Hard problems 😅

Sure we'll incorporate multiple curators (so nostr:npub1r0rs5q2gk0e3dk3nlc7gnu378ec6cnlenqp8a3cjhyzu6f8k5sgs4sq9ac can use it).

Idea is to show various trust metrics (I also dislike scores) for the user to make a decision whether to trust/install the app or not. Domain names are fiat crap, sure, but they have value so they're shown. The app itself can completely change how trust is assessed, you might not care at all for an alarm clock app but care a lot for a bitcoin wallet.

But no matter how you slice it, the web of trust will ALWAYS be involved. Curators are simply someone you decided to trust. "Known people in the space" are also because people you trust, trust them and we need to validate that. I can assure you I can spin a full clone of the existent ~25k and ~180k connections in have in trustgraph and have Gigis, Odells, etc.

Re: avoiding impersonations, it's my next plan for trustgraph (i.e. probability of impersonator, given your 2-3 hop graph) - I was discussing that with Niel the other day.