Before people go all in on Typst, I’d like them to consider the wisdom of replacing TeX, software that is:
- Absolutely finished
- Essentially bug free
- Written in a simple, highly portable language
- Extremely general
- Backwards compatible with decades old code
…with a commercial product that relies on an enormous tangled web of dependencies amounting to millions of lines of code for something which is far less general than that which it replaces
RE: https://mstdn.social/@mfeir/117472396351742808
There is a notion that when you write, you are simply mechanically translating your thoughts from a Fodorian “mentalese” into English (or any other language, natural or otherwise). This is not true. The thought doesn’t exist until you organize it. An LLM cannot do that for you because it isn’t you.
I genuinely think that the sloppening of all major FOSS is an opportunity. Software is absolutely, utterly broken and this was already true in 2022, before any of this had happened (but how much more now). I used to be very concerned about how my work would fit in with the broader ecosystem and now I don’t care, I’m free. We have the opportunity to start afresh and build something new, and better.
And like, this has all resulted in software that is worse! Every big FOSS program basically sucks. It’s all garbage! Linux is architecturally awful and is pretty much Swiss cheese when it comes to security. None of this stuff holds a candle to the apex of what software once was and could have been.
And by the way, if you were around back in the day, you will remember that many people were clearly aware that open source would result in exactly what has happened. Every argument that this wouldn’t happen, that you could technically charge for FOSS software (ok but why would you ever *pay* for it?) has aged incredibly poorly, and it was obvious this would happen 30 years ago.
There are circumstances where I think it makes sense to release source code for free, even now. And I certainly favor the patronage model of, say, a corporation or foundation paying someone to write public domain software. But I think if you are still giving your labor away for anything that matters, you are essentially a rube.
It is very very difficult for me to see “open source” as anything but a ruse by which big tech got nerds to work for them for free. Every time I see that xkcd comic about infrastructure, my main thought is pretty much “ok well why would you do that for free?”
I am gonna be honest, I have a really hard time seeing how the status quo of the mid 2000s and before of small time proprietary software shops was “worse” than what we have now. What has “open source” amounted to? Well…
1. Free labor for corporations. In web-based apps, you can simply use even GPL-based code for free without reciprocal releases.
2. Grist for the LLM training mill
3. Killing the ability of small time software developers to make money on their own
As much as I love computer science, there is so obviously a moral rot in this field that goes back to the absolute beginning. McCarthy, Minsky (who died before his Epstein connections were made public), many many others…
We cannot move past this moment until we root out the human-eliminationist, misanthropic, eugenicist ideology.
Like, handwriting recognition obviously works decently well and is pretty old. It is, however, *not* a theory of writing. Handwriting recognition models tell you *nothing* about how handwritten text grapholinguistically functions!
And by “explain” I do not mean produce extruded textual explanations which they sometimes can (but often very poorly). Rather I mean that producing an ML model that tracks or seems to track or reproduce an object phenomenon is simply not and never can be a theory or explanation of that phenomenon.
I think a lot of people want to draw a distinction between ML and LLMs or “GenAI” the latter of which I don’t think is a scientifically useful category. I must admit I am basically a total skeptic of ML for nearly all applications, because in my mind, such methods mean largely giving up on theory building. Yes these can be useful, certainly as a crutch, but never as much as deterministic theories or at least direct theories of object phenomena because ML models **never** explain anything
I’m wondering if anyone would be interested in some kind of network for those in or engaging with academic CS or math who are skeptical/suspicious/or otherwise opposed to LLMs/“AI”? Personally, I feel a great deal of loneliness on this matter at the moment, and would appreciate the camaraderie if nothing else! I’m thinking this could be a Signal group chat if nothing else. If you’re interested, let me know with a response to this post. Would also appreciate boosts for visibility!
And I’m sorry to say but Linux itself has been a disaster for OS research and development. It’s 2026, and all we have to show for it is a clone of Unix, an utterly unremarkable early minicomputer OS, a design old enough to be my parent. If LLMs are so great, where’s the new interesting work? Where’s the groundbreaking new kernels? This field is an absolute joke sometimes.
The Linus Torvalds hero worship is out of control and has been for a long time. He is confidently wrong, not infrequently, even in his supposed area of expertise. For example, he is still repeating falsehoods about microkernel performance that are really just laughable in a world with billions of L4 deployments. Take his pro-LLM tirade as more of the same.
Very pleased to announce I’ll be starting as a Visiting Fellow with the VIEWS (Visual Interactions in Early Writing Systems) Project at the University of Cambridge this fall! I’ll be continuing my work on text encoding, digital typography, and computational approaches to grapholinguistics. Please hit me up if you’re in the Cambridge area (or London)!
My recent talk on computerizing hieroglyphic scripts just got posted online! There’s a demo of my Maya font in the presentation (with of course, a few typos I realized I made—but this is interesting in itself, the world’s first Classical Maya typo?) I hope you find it interesting and please feel free to ask me any questions you might have.
Ok now everyone who liked this follow me 💀
POV: you are using the "modern" internet on anything less than a 10 gigabit connection
After my talk at SPLASH, I had a lot of people tell me I should consider doing a PhD program, despite my pointing out that I don’t have an undergrad degree. I’ve come to the point where I feel really morally unable to continue working in the software industry, thus it’s been on my mind.
I don’t really know how this sort of thing works, but, would anyone like to work with me? I’m interested in dependent type theory, computer typography, logic programming, and many other things.