Kristian with a K

"4623f8250e6a7b86dd90c6750b38ad6e9e284e994023f5698a3af8af51f327399605b8f29002e61b3f244ba8116f43658be8b9b14ee0b5f4b9cd7d5fcf572333"

Last Notes

2026-08-24 16:35:37 UTC

- reply

i’m listening 👀

2026-07-23 15:54:16 UTC

- reply

@npub1zq6…xf4t Upon thinking on it, I feel like a number of these could be NIPs to build on Nostr as opposed to making a whole new system. e.g.
> A post gets encrypted, then split into (say) 30 fragments where any 10 rebuild it. Fragments scatter across nodes...
Since clients are smart and relays are dumb it should be possible to conceive a "fragmented-gift-wrap" kind where it encrypts/fragments and disperses to multiple relays wrapping the real underlying kind.
> IT-PIR (private information retrieval)... IT-PIR lets you ask several servers a mathematically scrambled question — each one hands back data without learning what you actually wanted...
Probably requires a relay-side improvement with a new IT-PIR websocket request type
> Mixnet with adaptive cover traffic. Tor hides who you are; a mixnet also hides when you act.
Since this is between the client and the relay, this doesn't need to be baked into the relay end (keep the smart client paradigm, but could create server implementations that support it natively)
@npub1y0g…00ly has an example that uses mixnet (Nym) for example
> and your device emits fake traffic continuously so an observer watching your internet line sees a flat, constant hum whether you're posting or asleep
A client could continually request events, randomly varying the payload size in the request (even if the bloat get's discarded by the relay).
> Proof of personhood via zero-knowledge proofs
Curious at what level you imagine this playing in? Is this another signature on an event? That seems like it could be a client update that interacts with a proof server. I'm not super familiar with how these proofs are performed rn though.
> RLNv2 + Merkle tree. Rate limiting without identity. You're an anonymous leaf in a membership tree. Each post carries a mathematical fragment of your secret key. Stay under your quota, nothing leaks. Exceed it — post twice in one time slot — and the fragments combine, exposing your key so anyone can slash your stake and evict you. Spam punishes itself, automatically.
This is interesting, would need to learn more about the math behind this. This feels like the one thing that would be a true divergence from how Nostr uses public key cryptography.

2026-07-21 23:16:13 UTC

- reply

I know @npub1l5s…gx9z made https://git.imwald.eu/silberengel/scriptorium-obsidian/src/branch/master/README.md for easily publishing from obsidian files. Publications readable e.g. in https://alexandria.gitcitadel.eu/ and some other clients too. Not sure what OP used

2026-07-21 23:11:21 UTC

- reply

Exciting, but your website is giving a 404

2026-07-17 23:10:13 UTC

- reply

https://www.storj.io/ offers decentralized storage like an S3 bucket at a minimum (currently) of $5 a month. https://www.backblaze.com/ I think has a lower minimum cost and more competitive pricing compared to AWS

2026-07-17 23:03:08 UTC

I was today years old when I realized that `^` and `$` symbols in RegEx carry the same meaning as the motions in vim

2026-07-14 13:42:04 UTC

- reply

DMs are pretty messy. This is how I came to switch from regular Gift Wraps to Private Streams for private events where you can’t filter on timestamp but you want the metadata privacy. This way it’s only trying to decrypt that content and isn’t muddled with DMs and who knows what other Gift Wraps

2026-07-08 00:55:53 UTC

- reply

Voice would presumably use P2P WebRTC and TURN servers

2026-06-22 21:45:41 UTC

- reply

Yeah PoW definitely helps!
Consensus doesn’t work if i write a script to spam road closures/accidents from unique nsecs at the same set of locations

2026-06-21 21:44:52 UTC

- reply

Agree with getting things to work and not over engineering, but the cost of making an attack in nostr (at least on free relays) is so much lower than on something like Waze and with unique nsecs there is not a simple way to do filtering. The risk analysis is low effort + large impact (assuming people would regularly be trusting it for directions)

2026-06-19 17:59:45 UTC

- reply

Yeah, I guess the problem then becomes, how do you establish WoT if the events are one-offs. Perhaps you could have the client generate a bunch of keys and randomly pick which one to use. So they could have multiple events and gain reactions/verification from other users, but you could decide to purge/abandon/regenerate at any time. That would be less susceptible to Sybil attacks than one-offs at least

2026-06-19 15:15:18 UTC

- reply

not if you use a separate nsec for the traffic app

2026-06-08 18:30:06 UTC

@nprofile…ssyg do you have insight into these spikes? curious what ephemeral 20001 is being used for. stats on https://stats.andotherstuff.org/
https://blossom.primal.net/bc23391e718faab4e10e8a4b51b83539b838c8711df94441870afc3800c98fb9.png
https://blossom.primal.net/285cbfe535d86b366a6a1c94d8a5f67143dacbf9946ce97f8dd9d4243d7cbaf5.jpg

2026-04-02 03:35:57 UTC

- reply

ooh! how does it factor the “trusted” metrics?

2026-03-28 01:50:42 UTC

Hello World!