Nostr Web Client

where can I understand better what key migration is and does nostr:npub1j8y6tcdfw3q3f3h794s6un0gyc5742s0k5h5s2yqj0r70cpklqeqjavrvg?

and what does getting the WoT built look like, nostr:npub1u5njm6g5h5cpw4wy8xugu62e5s7f6fnysv0sj0z3a8rengt2zqhsxrldq3?

david 10mo ago

What I envision is that we will see an evolution of personalized WoT relays into a sophisticated tool that will keep track of our webs of trust and use them to curate all of our content. I say “webs” plural bc there are different ways to calculate it, different ways to use it, an effectively unlimited number of categories of “trust.” For whatever topic you’re interested in, your personal WoT relay will tell you who are the most trustworthy or knowledgeable, even if those people are not already one or two hops away from you on the social graph.

Once that’s done, key rotation will be only one of a thousand problems that we will address with it.

🍇 🧠 ⚡️ 💜

Reply to this note

Please Login to reply.

Discussion

david 10mo ago

From a more technical perspective:

There are different ways that a personalized WoT relay can define WoT. One way, probably the simplest way, is your WoT = your follows + their follows. That method is great at getting rid of spam, but throws out a lot of the baby with the bathwater. Newcomers to nostr in particular may be underserved by this method of calculating WoT.

There are ways to tweak the above method for calculation of WoT, but I think the next generation personalized WoT relays will implement centralized algorithms like PageRank, which is what put Google on the map in 1998. PageRank does a better job at sorting the wheat from the chaff than the simple “follows of follows.” This will be especially important for newcomers to nostr, for two reasons: first, newcomers will have only a few followers at first, which means they’re more likely to be excluded, and for a longer period of time, using the follows of follows method. Second: newcomers will have only a few follows at first, which means their follows of follows may be too restrictive in the beginning. Many of them may quit nostr before discovering all that it has to offer.

The downsides of PageRank are that it is not contextual and not suitable for using as a weight when doing weighted averages or weighted sums. I use the term GrapeRank to refer to centrality algorithms that address those two shortcomings. They’ll require nonlinear equations (as opposed to PageRank which is linear) which means extra compute, but they’ll be powerful and versatile and I think they’ll be unexpectedly awesome. It’s just a matter of putting in the work and building them out.

🍇 🧠 ⚡️ 💜