I just found out that Claude 4.5 Sonnet is remarkably bad at simple refactoring and code cleanup. Splitting one file up into two while maintaining functionality turned out to represent a surprisingly difficult challenge for something that is mostly trivial copy and paste for humans. Amazing! #ai #dev

I feel like I've hit the current limit of agentic development - frustratingly early. I was just trying to add some very simple Playwright end to end tests to a React app with a Rust backend. Even after 6 hours of debugging, the success rate of those tests got stuck at around 20%! So much for the skills of Claude Sonnet 4.5 (thinking) in Cursor. Well, back to manual debugging, I guess. At least this has been a very educational experiment. #ai #dev

Really fascinating and useful report! Especially for those who see a job in the Bitcoin space like me! View article →

Yesterday I tried transferring some Btc from my #Binance account to my hardware wallet. In previous instances, that worked. But yesterday the transaction was flagged as high likelihood of a scam. Outgoing transfers were blocked twice for an hour! I searched for similar stories on Nostr during the last 30 days, but didn't find any. What is going on?

Just found this old post, and think that it's still relevant. Do you think the situation on Nostr has changed since back then? Are there any new insights? View article → #v4v

Today I experimented with agentic development with Cursor after purchasing a Pro+ subscription costing $60 US, which pays for $60 US of tokens. The overall experience was relatively good, but checking the usage list made me think again whether Pro+ would suffice for any large project work. Claude 4.5 Sonnet is really good, but burns $ at a rather rapid pace. For easy tasks, the custom AI agent Composer 1 by Cursor is rather sufficient. Also, having long-winded sessions seems to push costs upwards due to ever increasing context. Letting a second agent review what the first agent did and a third agent implement the requested fixes reduced costs to $0.50 for those steps. Compared to the roughly $20 I spent on the actual coding and debugging, that's a bargain. It was just about refactoring a small project to use a different library as dependency. In the end, I didn't need to check the documentations of those libraries. But I needed to check the output of those agents very closely. They tend to drop functionality, if they feel it's difficult to maintain it! #ai #dev

This is deeply fascinating! Economics seems to explain even more than just the parts of politics we usually think about. Long read, but really worth it! View article →

Wow, finally zapping via Primal (or anywhere on Nostr) via NWC worked for me. The last days I thought I had distributed a dozen zaps or so. I still don't know what the underlying issue was. A combination of creating a new NWC connection, deleting the old one, and opening Primal in a new tab finally did the trick. I don't know how I should feel about that. My feelings are quite mixed. My general user experience with Nostr has been one of much confusion and frustration. On the other hand, I feel some elation about finally having mastered the basics.

This technology seems quite useful for many applications running on the #nostr network! View article →

The idea of quantifying trust as threshold amount of money you expect someone to embezzle or not embezzle is quite innovative. But in a pseudonymous context large amounts of trust can only be reasonably established via reputation or personal trust. While the idea of personal trust and its implications of transitive trust are the subject of web of trust systems, I have been thinking about decentralized reputation system for quite a while - shortly after the emergence of Bitcoin in fact. I have developed a theoretical reputation system called Quantified Prestige, which is based on quantified esteem attributions. Only rather recently, I realized that Nostr provides a suitable basis for this system in a decentralized context. Those esteem attributions can easily be implemented as custom Nostr messages. It's just aggregation across relays and Sybil protection that require some relatively elaborate mechanisms. The result is indeed "merely" a reputation score, but as a global starting point for interactions with unknown pseudonymous actors that's certainly a good starting point. Here is a recent introduction to Quantified Prestige:

Yakihonne

Towards a Positive Peer to Peer Reputation-based Economy: Quantified Prestige

Quantified Prestige is a concept for a positive peer to peer reputation-based economy. Its motto is "Positive reputation done right". The system is...

Linking my #introduction article:

A Radical Visionary arrived in Nostr

Introduction of a philosophical science fiction author and full stack developer

It seems to me that articles need to get mentioned in notes to reach significant attention. Anyway, the #nostr learning curve is still quite steep in 2026.

Test. My first post on Yakihonne.