Your job is to deliver code you have proven to work, Simon Willison

As software engineers we don’t just crank out code—in fact these days you could argue that’s what the LLMs are for. We need to deliver code that works—and we need to include proof that it works as well. Not doing that directly shifts the burden of the actual work to whoever is expected to review our code.

How to be right 90% of the time, and why I’d rather be wrong. Paul Buchheit

What if instead of dismissing everything new or daring, we acknowledge that the future is uncertain, and pursue some of these new ideas despite the risk? History has shown that no one can reliably pick the winners, but what if we were clever enough to pick the right ones 20% of the time? That would mean that were are still wrong 80% of the time.

Is it better to be right 90% of the time, or wrong 80% of the time? It depends on the risks and rewards. If we look at it as an investment where the losers go to zero and the winners have a 10x return, then we can see that it’s potentially better to be wrong 80% of the time:    .20 * 10 + .80 * 0 = 2

How This Was Built, How It Works. Mike Masnick

I know that it’s fun to hate on AI tools right now, but honestly, none of this would have happened without my being able to use Claude Code as a tool. And it was very much used as a tool. I pushed back on it at times, and questioned some of its ideas and decisions. And there were some things that took multiple experiments to get it to work properly.

But after 12+ years of dormancy, the blog is back, connected to the social web, and I actually want to post on it again. Neat!

Appendices - Organizing Your Social Sciences Research Paper - Research Guides at University of Southern California

Appendices are always supplementary to the research paper. As such, your study must be able to stand alone without the appendices, and the paper must contain all information including tables, diagrams, and results necessary to understand the research problem. In contemplating if information should be placed in an appendix, consider asking yourself, if it were removed, the reader would still be able to comprehend the significance, validity, and implications of your research even if that additional data or information was missing or is the information so voluminous that it would breakup the narrative flow of the paper.

Gas Town Decoded — Andrew Lilley Brinker

In the interest of making Gas Town intelligible (because, despite the prose, the idea of agent orchestration it describes will be important), I’d like to share a quick decoder for the many new terms Steve introduces. His article itself offers definitions, but those definitions reuse his insular terms, making by-hand decoding tedious. Here, I’ve done the work for you.