elsat's avatar
elsat
elsat@habla.news
npub1zafc...26k5
Janitor
elsat's avatar
elsat 3 months ago
Make “it just works” the standard. Congratulations to primal team for continuously iterating to a better experience for years now. The wallet stuff is important to get right. Well done @Marko @Pavle @alex and team View quoted note →
elsat's avatar
elsat 3 months ago
Brilliant human to LinkedIN speak translation tool by @Kagi #💩 image
elsat's avatar
elsat 3 months ago
GM ☀️ Remember that team anti-human will use “nature” to scaremonger you, your kids, and your community into hating yourself. The modern (sick) human’s anti-human hypothesis is that during the Middle Ages it was suspected that the black plague would result in increased biodiversity, due to 80% human population collapse. Not only was this disproven, but it was proven that biodiversity peaked when human population peaked right before the human population collapse event. When you hear, see the TV/someone fearmonger you in that “humans bad” unsubscribe from gulagflix/tell that person to piss off.
image image image
elsat's avatar
elsat 3 months ago
No one asked for it, so I spent the past few weeks benchmarking algos coverage and performance.
GitHub
GitHub - nostrability/outbox: outbox performance measured
outbox performance measured. Contribute to nostrability/outbox development by creating an account on GitHub.
 I’ve reached a point of diminishing returns with outbox benchmarks. Key learnings: 1. Randomness > deterministic algos 2. Add Thompson learning to increase coverage by 25%+ by learning and persisting which relays return useful info 3. Nip-66 reduces latency by 40% in not checking for info from dead/offline relays Submitted a few PRs:
GitHub
outbox: add Thompson+CG3 (+4.9pp recall), NIP-66 filtering (~45% faster), connection cap by alltheseas · Pull Request #387 · nostr-dev-kit/ndk
Summary NDK's outbox relay selection doesn't learn from delivery outcomes, can't filter dead relays, and has no connection cap. This PR...
GitHub
feat: Thompson sampling relay selection + count++ bug fix by alltheseas · Pull Request #54 · hzrd149/applesauce
Summary Fix count++ post-increment bug in selectOptimalRelays — maxRelaysPerUser was never enforced (pool pruning + output capping) Add Thompson...
GitHub
feat: NIP-66 relay liveness pre-filtering with safety guardrails by alltheseas · Pull Request #55 · hzrd149/applesauce
Depends on #54 — this branch is based on the Thompson sampling PR. Merge #54 first; these commits apply cleanly on top with zero conflicts. Summ...
GitHub
Add opt-in Thompson Sampling for relay scoring by alltheseas · Pull Request #53 · coracle-social/welshman
Summary Add sampleBeta(alpha, beta, rng?) to @welshman/lib — Beta-distributed sampling for Thompson bandit relay selection Add getRelayPrior opt...
GitHub
gossip: add Thompson Sampling for adaptive relay selection by alltheseas · Pull Request #1282 · rust-nostr/nostr
Description This PR adds Thompson Sampling on top of the existing filter-decomposition (FD) gossip routing in rust-nostr. The FD pipeline (gossip_b...
GitHub
feat: NIP-66 relay liveness filtering for outbox selection by alltheseas · Pull Request #1833 · vitorpamplona/amethyst
Summary Uses NIP-66 relay monitor data to filter dead relays from outbox relay selection, reducing wasted connections and improving feed load perfo...
GitHub
Outbox: stochastic scoring, NIP-66 liveness, Thompson sampling by alltheseas · Pull Request #53 · nostur-com/nostur-ios-public
Summary Stochastic relay scoring — replaces deterministic createRequestPlan(skipTopRelays: 3) with randomized scoring (intersectionCount × rand...
 https://gitlab.com/soapbox-pub/ditto/-/merge_requests/18 Hope at least one #devstr finds this useful. cc @fiatjaf @KernelKind
↑