bustermellotron

Joined 3 years ago

Karma

Submissions

Member since

Apr 20

Research Software Engineer at the University of Bristol

Erdos number: 2 (https://arxiv.org/abs/2107.06674)

My dilapidated house: https://www.instagram.com/bristol.victorian.house/

View on Hacker News

Recent submissions

Comment · 1 month ago

Something about Dr. Bronner soap?

Comment · 1 month ago

There is a pre-release book on O’Reilly called “AI-Native Software Engineering” that has three chapters on spec driven design. That is where I found out about openspec.

The skills facilitate SDD, although what that actually means is vague: https://martinfowler.com/articles/exploring-gen-ai/sdd-3-too... (Note: this article is almost a year old.)

The point AI-Native Software Engineering makes is: agents misunderstanding your goal is a common cause of bad results, so spend 10 minutes writing a spec to clarify ahead of time. The spec outlives any single conversation, and provides a rubric for success.

I haven’t actually started using openspec in earnest, but it is suppose to have a way to update specs in a structured way. Also, it is one of the simpler frameworks for SDD, versus eg speckit.

So the question is more if SDD is useful, and that might be more personal than eg cutting token costs.

Comment · 1 month ago

He posted about using ChatGPT to transcribe PDFs when it first became popular. So he’s been enthusiastic about LLMs for a while.

Comment · 1 month ago

Is there an easy way to do this with eg codex? It seems like eg sol agents can’t spawn Luna subagents, so eg a “code research” subagent can save the main agent’s context, but can’t save tokens necessarily. (I suppose a tool to call codex CLI would work, but a bit unsatisfying.)

Comment · 1 month ago

I also found LSP like skills to usually have no advantage over rg. The agent needs to read the code to understand it; navigation is a small portion of that.

Comment · 1 month ago

I just went through a lot of benchmarking and the only thing that seemed better than rg was chunkhound, which sounds similar to this project. Actually a small Jina embedding model actually did better than voyage AI, but took a long time to index. Also chunkhound doesn’t work well with worktrees. In the end, I decided to stick with rg.

Comment · 5 months ago

Size of a finite set. It’s common notation in this field.

Comment · 5 months ago

The grid of squares actually gets > Cn for any C. (More in fact… C can grow like n^a/loglog(n).) The AI proved > n^{1 + b} for some small b > 0, which a human (Will Sawin) has now proved can be about b = 0.014. The grid can be rescaled so the edges are not necessarily length 1, but other pairs will have length 1; that is necessary to get more than 2n unit distances.

Comment · 6 months ago

You could use air scrubbers https://en.wikipedia.org/wiki/Soda_lime

Comment · 6 months ago

I saw Tim Gowers give a talk at the AMS-MAA joint meeting in Seattle about ten years ago where he predicted that in 100 years humans would no longer be doing research mathematics. I wonder if he’s adjusted his timeline.

At the time I thought the key missing tool was a natural language search that acted like mathoverflow, where you could explain your problem or ideas as you understood them and get references to relevant literature (possibly outside your experience or vocabulary).

Comment · 6 months ago

Go go Gadget arms!

Comment · 8 months ago

I found it easy to adapt to the x-bows keyboard (column staggered and splayed). The thumb buttons and large ctrl, alt, space are great for emacs. My only complaint is that the braces are a bit far away.

Comment · 2 years ago

I think the book "The Now Habit" discusses this well. (Could be another book though...)

I once completed a 3000 mile cycling tour across the US (and didn't take ADHD meds while on the trip) and was I was basically disappointed to reach the west coast.

Comment · 3 years ago

This covers most of what you need for "housekeeping" scripts in Python: https://automatetheboringstuff.com