I feel seen. 🫂

So cool.

It really pisses me off that agents ruined en/em dashes. I used them all the time before LLMs came along and now, I'm having to train myself to stop so that people who don't know grammar won't think I'm AI slop.

Yes, and… what people say and what people do differ significantly on products. Revealed preferences tell you a lot about what your product is solving for people, often stuff they’re not even sure how to articulate themselves.

I'm currently leaning towards self-hosted Umami (very similar to Plausible) for website analytics and self-hosted Sentry or GlitchTip for tracking crashes and errors. But still doing some research. And yes, I agree. A docs page which details exactly what is being collected and how would be required as well.

What do you mean served? Is this specific to zapstore for download numbers?

GM af 🌞

Why not just use one of the self hosted tools that already exists?

Genuinely curious what everyone's thoughts are on product analytics in the apps you use. I think most teams building in the nostr/bitcoin/freedom-tech space are a hard-no without thinking about how you could do them well (i.e. privacy-first, opt-in only, etc). A lot of people complain about the quality of Nostr apps but, the truth is, it's nearly impossible to build a high-quality product that delivers real value without having at least some idea of how folks are using a thing. Trolls, gtfo. I'm only interested in thoughtful or thought-provoking responses.

* From Matthew Green's recent post on WhatsApp: https://blog.cryptographyengineering.com/2026/02/02/whatsapp-encryption-a-lawsuit-and-a-lot-of-noise/ ** Oh, and Instagram just announced they are removing the option of E2EE

Spot on.

GM 🌞 An amazing use for LLMs is to give them github URLs for open source projects you're unsure about the quality on (or whether they might be unsafe) and ask them to do a full review. GPT 5.4 on extra high is a phenomenal reviewer.

I agree. It has to be driven by user opt-in. But I can see a lot of people wanting to help out and they'd opt-in. Again. it's down to the the builders to do this is a privacy preserving way but, as with all OSS, they can just look at the code to see that we're not tracking anything sensitive (or have their Clanker do it for them).

They come from a massive data indexer I built. They’re real numbers, with caveats. There’s more info on that in the site.

Interesting. I like that we have several different indexer projects in the ecosystem now.