I configured Claude Code to use a local model (ollama run glm-4.7-flash) that runs really well on a 32G M2Pro macmini. Maybe my standards are too low, but I was using that combination to clean up the code, make improvements, and add docs and tests to a bunch of old git repo experiment projects.

I want one of those t-shirts.

I was incredibly lucky to have been funded to write StarLisp code for the original CM-1 machine. CM-1 was a SIMD architecture, the later models were MIMD. Think of the physical layout being a 2D grid or processors with one edge being for I/O. That was a long time ago so I may have the details wrong.

You can order them at the bottom, lucky you :)

A friendly counterpoint: my test of new models and agentic frameworks is to copy one of my half a zillion old open source git repos for some usually very old project or experiment - I then see how effective the new infra is for refactoring and cleaning up my ancient code. After testing I usually update the old projects and I get the same warm fuzzy feeling as I do from spring cleaning my home.

I also like to generate greenfield codebases from scratch.

I like the idea of a smaller version of OpenClaw.

Minor nitpick, it looks like about 2500 lines of typescript (I am on a mobile device, so my LOC estimate may be off). Also, Apple container looks really interesting.

I surprise myself: at the international AI conference AAAI in 1982 some of the swag was a bumper sticker “AI It Is For Real” that I put on my car and left it there for years.

With that tedious history out of the way, even though I respect all the good work that has gone into this and also standalone mostly autonomous LLM-based tools, I am starting to feel repulsed by any use of AI that I don’t directly use for research, coding, studying non-tech subjects, etc. I think tools like Gemini deep research and NoteBookLM are superb tools. Tools like Claude Code and Google’s Antigravity are so well done, but I find it hard to get excited about them or use any tool like these tools for more than once or twice a week (and almost always for less than 10 minutes a session.)

I was going to agree with this sentiment until I realized that I listen to a few podcasts put out by individuals who spend a lot of time producing content. For me, it is pod quality vs. how many ads.

I read years ago that Hetzner placed data centers near inexpensive power, but I understand that the EU’s energy situation has deteriorated. So you are correct, they have the larger energy problem to contend with.

If they wait a year or so, the new AI chips being used now in China will probably be available for LLM inference in Europe. It seems unfortunate for small and medium size countries, and also for the EU to be dependent on any IT infrastructure only from China or the USA, but perhaps being flexible enough to be able to switch venders or use both is safer?

exactly. in HPC we all understood that it was a tradeoff between money and time, and that the curve was exponential. if you wanted to race ahead of todays capabilities, you could, but you couldn't go very far without burning alot of cash.

because of the investment story about being first and building a moat, we have companies torching 100s of billions of dollars to see who can climb that exponential the furthest.

we have so much work to do, in infrastructure, and distributed computation models, and programmability, quantization, and information theory...just relax a little. you dont have to compete with OpenAI. OpenAI is just a giant waste of money. take your incremental gains and invest in research and I assure you we can get there without directing our entire economic output into buying the latest highest margin parts from Nvidia only to use them at 30%, if you're being generous.

This kind of reminds me of an old paper: https://arxiv.org/abs/astro-ph/9912202

Of course, we are no longer in the clock speed doubling era, but computer does get faster still.

It's true that the progress on clock speeds has slowed. Now we have to address the parallelism problem in order to keep moving forward. And we haven't done a very good job. Progress on that front will get us back on the acceleration curve. Saedfly, the current framing of 'who buys the most hardware', while I providing a nice marketing story, isn't netting us that much progress except what Nvidia spends internally.

Proton drive is fine, their docs service is usable but could use improvement. Their secure and private file and docs sharing with other Proton users could be a great feature, if you need it.

EDIT: I just re-tried Proton docs and spreadsheets - much improved docs, and I think the spreadsheets are a new feature; looks OK but I am on mobile right now so minimal testing.

I have used Hetzner off and on for years, nice products and services.

I don’t care what provider you use, if your business or app use case needs any sort of reliability have a plan for reinstalling code and data on alternate providers quickly as possible.

There are horror stories of people and companies being cut off because of pressure from the US government, or having one of the Google/Microsoft/Amazon tech giants cancelling accounts.

Really, in today’s world, why totally rely on anyone?

EDIT: it seems prudent to maintain a cloud account in Europe, US, and Asia and have a plan for moving application code and data around if required. Outside the US I have mostly only used Hetzner, but Alibaba has impressive looking services.