Hacker News: bustermellotron

New comment by bustermellotron in "A misalignment of AI in mathematics"

bustermellotron — Fri, 11 Sep 2026 22:04:20 +0000

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

New comment by bustermellotron in "RTK reports token savings, but our cost benchmarks disagree"

bustermellotron — Fri, 11 Sep 2026 21:53:46 +0000

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.)

New comment by bustermellotron in "RTK reports token savings, but our cost benchmarks disagree"

bustermellotron — Fri, 11 Sep 2026 21:47:07 +0000

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.

New comment by bustermellotron in "RTK reports token savings, but our cost benchmarks disagree"

bustermellotron — Fri, 11 Sep 2026 21:45:05 +0000

Similar for ponytail, I don’t know if it saves tokens, but there is less output to read (and usually less over engineering). Occasionally I have to push for more complex code, but that is much nicer than constantly asking for simpler code.

New comment by bustermellotron in "RTK reports token savings, but our cost benchmarks disagree"

bustermellotron — Fri, 11 Sep 2026 19:21:56 +0000

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.

New comment by bustermellotron in "Sum-product, unit distances, and number fields"

bustermellotron — Thu, 04 Jun 2026 20:00:43 +0000

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

New comment by bustermellotron in "An OpenAI model has disproved a central conjecture in discrete geometry"

bustermellotron — Wed, 20 May 2026 20:36:19 +0000

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.

New comment by bustermellotron in "I let AI build a tool to help me figure out what was waking me up at night"

bustermellotron — Tue, 12 May 2026 05:10:57 +0000

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

New comment by bustermellotron in "A recent experience with ChatGPT 5.5 Pro"

bustermellotron — Sat, 09 May 2026 05:12:28 +0000

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).

New comment by bustermellotron in "After a 40-year wait, technology enables three-sided zipper design"

bustermellotron — Wed, 06 May 2026 05:39:52 +0000

Go go Gadget arms!

New comment by bustermellotron in "A beginner's guide to split keyboards"

bustermellotron — Fri, 20 Feb 2026 06:25:22 +0000

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.

New comment by bustermellotron in "The manager's unbearable lack of endorphins"

bustermellotron — Sun, 23 Jun 2024 09:23:22 +0000

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.

New comment by bustermellotron in "The Case for Bash (2021)"

bustermellotron — Fri, 19 May 2023 09:18:53 +0000

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

New comment by bustermellotron in "Only one pair of distinct positive integers satisfy the equation m^n = n^m"

bustermellotron — Thu, 20 Apr 2023 19:53:30 +0000

On the other hand, your proof really only needs the binomial theorem and geometric series.

New comment by bustermellotron in "Only one pair of distinct positive integers satisfy the equation m^n = n^m"

bustermellotron — Thu, 20 Apr 2023 19:45:22 +0000

The “claim” more or less proves that k-1 th root of k is less than 2 if k is larger than 2. So I think your argument is equivalent.

New comment by bustermellotron in "Only one pair of distinct positive integers satisfy the equation m^n = n^m"

bustermellotron — Thu, 20 Apr 2023 10:06:22 +0000

Here is an elementary proof:

Since m and n are distinct, we may assume that m > n >= 2. From the equation and unique factorization, we know that n divides m, so write m = nd.

Then (nd)^n = n^(nd). Hence d^n = n^{n(d-1)}, which yields d = n^{d-1} >= 2^{d-1}.

Claim: if k is an integer greater than or equal to 3, then k < 2^{k - 1}.

Proof: the base case is clear: 3 < 4. Suppose k > 3 and k - 1 < 2^{k - 2}. Then k < 2^{k-2} + 1 <= 2^{k-1}, where the last inequality holds because 2^{k-1} - 2^{k-2} = 2^{k-2} >= 1. QED

So, d must be less than 3. Since m = nd and m and n are distinct, d is not 1, so d = 2. Since d = n^{d-1} and d - 1 = 1, we have n = d, so m = 4.