Nostr Web Client

Have you ever considered that there might be people with valuable things to say who lack either the skills or the desire to grow their follower count? Maybe they don’t have the time. Or maybe they kinda like the idea of flying under the radar.

Here’s my essay on how web of trust can free us from the tyranny of the legacy advertisements-addicted system which worships social media “influencers” at the expense of everyone else. How WoT can give us the tools to find the users and the content we want. Even when those users don’t have a high follower count.

https://prettygoodproject.substack.com/p/the-pretty-good-way-to-calculate

Reply to this note

Please Login to reply.

Discussion

₿itcoin 1y ago

Back when nostr didn't exist I had no way to communicate with the twitter bitcoin maxi's. Now I do. Or don't.

nostr:nevent1qqszmqzp6xncxfjhys4zwxvfgcdg4tyu5uf2sl550hz38sylfwk4lyspz3mhxue69uhkummnw3ezummcw3ezuer9wcpzpef89h53f0fsza2ugwdc3e54nfpun5nxfqclpy79r6w8nxsk5yp0qvzqqqqqqyjw42fv

david 1y ago

Gonna write this essay on habla when I get the time

Only so much time in the day

Moon 1y ago

My daily struggle. Like I feel like I have a lot of share. I know I could def built something online, but I really detest the perverse incentives of audience capture and “attention-as-currency” influencing…

david 1y ago

The perverse incentives you so aptly observe can be poison for the soul

Sikto 1y ago

I’m in the same boat. I’m about to take on a grand adventure, but am resisting “selling myself and my family” to centralized media platforms.

₿itcoin 1y ago

What do you mean with built things online?

The Bitruvian Man 1y ago

Monetizable community/content. Can't refer to building software directly, but could refer to selling software.

₿itcoin 1y ago

Don't worry. Try to help people out. Follow hashtags. Post some cat memes, onboard your family and voila, you you can do this and that.

Sourcenode 1y ago

I think about WoT in regards to nostr almost daily. Thanks for taking the time to write this up

Robert Allen 1y ago

Great article. To make sure I'm clear, are you envisioning nostr users rating other users? If so, have you seen nostr:npub1arkn0xxxll4llgy9qxkrncn3vc4l69s0dz8ef3zadykcwe7ax3dqrrh43w's QTS system for scoring products? I'm extending that in a project now for ranking places. Would you use kind 1986 for reviewing users or would you use another NIP?

david 1y ago

Glad you liked it! To answer your questions: yes I do envision nostr users rating other users. The linked is an article with screenshots from my proof of concept app, where users rate other users as trustworthy or not in the context of curating a specific list of interest. It doesn’t implement the entire tapestry protocol, just the very early stages of it. The next step in the protocol will be to curate a graph of categories associated with each list. I have not yet built a proof of concept for the graph, but the methodology will be pretty similar as for simple lists.

https://github.com/wds4/pretty-good/blob/main/appDescriptions/curatedLists/overview.md

david 1y ago

I haven’t seen the QTS system but I’ll take a look. I’m using kinds 39901 and 9901 as per the part of the tapestry protocol which interfaces with nostr. The article I linked to shows examples with technical details.

The tapestry protocol is technically independent of whatever network is used to store and send data. I was building it on top of IPFS before nostr was born, but technically you could use either or both at the same time. I think nostr is the best place for me to start. There is a section of the tapestry protocol that describes how to interface with nostr, but I have a stub for IPFS and anyone could add a section for whatever other network they might want to use.

Mike Dilger ☑️ 1y ago

This creates far too much data to manage. Nobody is going to rate all their friends across all kinds of domains. IF that data existed it would be a great system, but creating and maintaining such data is not humanly possible.

I think we have to live in the real world and do simpler things like "how many of my friends follow this person?".

Mike Dilger ☑️ 1y ago

Furthermore, I don't want this data to be in any computer anywhere. I don't want the A.I. to know who I trust or how much I trust them.

david 1y ago

The grapevine will use two dimensions to define a context: an action and a category. Your grapevine will curate actions and categories as lists and will arrange them into hierarchies, so trust can be inherited down the hierarchy.

A follow in nostr would be represented by the grapevine by an attestation that translates loosely: I trust you to curate my nostr feed in the category of everything. If I really like your curation of topic X but not topic Y, I can attest accordingly, but no need for me to dig into subcategories if I don’t want to.

Any given attestation can be made public or kept private, at your discretion.

Your follows list is already public. So yes, you are willing to put at least some of it out there.

Mike Dilger ☑️ 1y ago

But my follow list is becoming private. Many of my follows you cannot see (encrypted contents).

I'm not against this idea, I'm just poking holes because that's what I'm good at - finding problems.

But I like that it is a grassroots ground-up way of solving a problem that big companies were controlling us with.

I'm not sure trust is transitive. In many cases it certainly isn't. How do you solve that? If A trusts B with a secret, and B trusts C with a secret, that doesn't mean A trusts C with a secret. In fact, that is how secrets get leaked all the damn time. Or I follow Ben, and Ben likes the keto diet. But I *hate* the keto diet. My following of him says nothing about my view on his diet.

david 1y ago

These are all good points. Which is why the big picture is that the entire protocol will ultimately be curated by your grapevine. Not at first, but eventually. Every single aspect of the protocol can be replaced with something better. As many different algos for as many different contexts as you need.

The tapestry protocol, as written by me (before your grapevine takes over from me), cannot and will not be perfect. It only has to be good enough to get to the point where it can be handed off.

What constitutes “good enough”? It needs to be capable of a handful of quasi-complicated things, and transitivity is definitely one of them. I’m not an expert in all things, and I can’t hand-pick the experts in all things, but if I can pick the people who pick the people who pick the people etc who are experts in some particular thing, then i can delegate anything and everything to a small handful of people who will probably be pretty good choices for the topic in question. And if it’s not controversial, my grapevine and your grapevine will probably settle on the same small handful of people who know and care about the topic in question. Or at least their opinions will likely overlap. I’m thinking about niche questions about our digital tools of communication, like whether a nostr note timestamp should use created_at or createdAt. No one really cares, provided we’re all on the same page, and the grapevine will usually get us all on the same page. (Unless there is controversy, which there isn’t in most cases.)

Other than transitivity, management of category trees by my grapevine is another one of the quasi-complicated but necessary things, I think. Which is why that’s also in the tapestry protocol, and goes by the acronym DCoG: decentralized curation of graphs.

david 1y ago

On the topic of privacy: the general idea is going to be that you will have the ability to reveal exactly what you want, to whom, and under what circumstances you choose. You already have that ability, in theory. But the grapevine will give you tools to fine-tune your choices. And you won’t have to publish your trust list in the same sense that you don’t have to publish your follows list. You may choose to give your follows list to a relay for the sake of performance, but that’s a tradeoff you can decide whether to make.

david 1y ago

Your comment got me thinking: the point of the tapestry protocol is for a community of users to work with their WoT to curate a decentralized digital language. In many ways, it’s modeled after the way real life works. Importantly, no one is required to have a bird’s eye view of the entire community or of all the attestations, or even to know how many users are part of the group. In fact, now that I think some more about it, communication with a small handful of peers is all that is technically required. We’ve been curating spoken languages that way for millennia, and the tapestry protocol just takes what we already do and implements it digitally.

My curated lists prototype doesn’t make the above feature evident, bc in the prototype, all attestations are public and so everyone does in fact have a bird’s eye view of everyone’s score for the curation of each list. So now I’m wondering what’s the minimal amount of the tapestry protocol that I can implement that will make the above feature evident, ie the fact that it allows us to build consensus on some question without anyone having a bird’s eye view. 🤔

david 1y ago

Now that I think more about it, my existing prototype is ALREADY capable of demonstrating that feature! All you have to do is assume that not every user has access to the same set of relays. Which is, in fact, already the case!

david 1y ago

You are correct: no one is going to rate all their friends across every trust domain that exists. Because there is no upper limit to the number of domains that can exist! Which is why this method doesn’t expect you to rate everybody on everything. A few ratings in a few general categories will be enough to get the system up and running.

My proof of concept demos curation of simple lists. Next on the docket: curation by your grapevine of a graph. In theory, any graph, defined as a set of nodes connected by edges. But importantly: we will use graphs to represent context. Your grapevine will curate graphs of context-categories that will be arranged in a hierarchy. This will allow default influence scores to be inherited down the hierarchy.

Suppose I’m in the mood to watch a drama. Alice recommends you as having good taste in movies, but she didn’t specify dramas specifically. That’s ok; dramas is a subcategory of all movies, so my grapevine pays attention to your recommendation on dramas bc it defaults to your trust score in the more general category of movies, higher up in the tree of categories.

This article discusses in more detail what I mean by context and how influence is inherited.

https://prettygoodproject.substack.com/p/context