anatoly

Can you tell me more? Just really curious about this approach, and am not a Photoshop expert or even user. How usable was Photocraft on release, and what were significant gaps if any - outright bugs, missing features, subpar implementation of features? How did it become worse with time?

raylib is one possible starting place. Also might look at dos-like (https://mattiasgustavsson.itch.io/dos-like).

I don't think you are arguing in good faith.

The time-wise imbalance of deaths is a very basic fact about the ongoing war, I didn't realize you were ignorant of it and needed a verification. The Hamas-provided statistics are timestamped, you can look e.g. at https://data.techforpalestine.org/docs/casualties-daily/, download the CSV file, look in the cumulative deaths column, see that it's just over 60k for the entire period, and note that 30k occurs around 2024-03-01. So I was slightly off and it's a little less than 5 months (oct 07 to mar 01) out of a little less than 23 months (oct 07 to 2025-08-31) that account for 50% of the deaths.

There isn't any report that actually counts Gaza's population, the UN provided an "estimate" with no methodology, births are not mentioned, and it's built on figures including number of people who exited Gaza (irrelevant to the claimed decrease due to violent deaths). That's not serious.

There's no coherent notion of genocide that fails to reduce the population significantly. Yes, you can argue (and people have) that the legal definition, by using the "part of" wording, can conceivably apply to virtually any number of deaths, but again, that's not serious.

No, this data in fact suggests growing population, for the following three reasons:

- the survey recorded a surprisingly small excess of nonviolent deaths (in excess of what's demographically expected), this is discussed in the preprint. The much larger number of violent deaths is almost matched by births, so the total balance is somewhat towards shrinking, in that cohort

- however, it is well known that the violent deaths occurred overwhelmingly early in the war (so far) - according to the official Hamas statistics, something like 50% of all casualties are in the first 4 months of the war, out of 22 so far. Whether these statistics are over- or under-counted is not likely to make a dent in this huge imbalance. So as the war is ongoing - and it's already been another 8 months since the 14 covered by the survey - the death rate is still "collapsing" compared to average rate so far.

- at the same time, the birth rate has evidently not seen such a huge collapse since the first 4 months of the war; this can't be gleaned from the survey, but enough plausible reports (e.g. what @richardfeynman quoted) exist that point in that direction.

So if we consider the survey relatively representative of the entire population, the imbalance towards shrinking population after 14 months is already almost certainly repaired towards growing after another 8 months, because so few civilians are violently killed (again, compared to the first 4 months of the war) in 2025.

It's hard to get good data on current birth rates in Gaza, but the recently published preprint of a demographic study of the death toll in Gaza (https://www.medrxiv.org/content/10.1101/2025.06.19.25329797v...) provides some evidence that the death toll in Gaza is approximately balanced by births. Specifically, the project directed in-person interviews of Gazan citizens representing ~2k households and ~9k people in them, and recorded ~390 violent deaths and ~360 births in that cohort, both from 10/7 and until January 2025.

I think the definition of "profound" they chose (nonverbal, minimally verbal or IQ < 50) has much less wiggle room than the broad ASD/autism diagnosis, and also fewer incentives towards inflation. Diagnosis replacement vis-a-vis intellectual disability is still a worry, and I wish there was a way to contrast with an ID stat for the same population.

Overall am in strong agreement with you, the main thing is to nail down data and very little seems to be done towards that. I've followed these studies and articles since 2011 or so with increasing dismay. The headline-grabbing stats of "1 in X" growing every year are next to meaningless, and yet I believe much points towards prevalence of actual condition really increasing. But with scandalously amorphous definitions and abysmal longitudinal bookkeeping we don't know and can't know how much it's increasing and in what subpopulations.

If you mean non-signigicant in terms of absolute numbers, these are numbers coming from a small sample of US population (15 metropolitan communities).

There's lots of true things in that post, and it's undeniable that there's been huge broadening of criteria and it's responsible for a lot of the growth in stats. The question is, is it responsible for all of it? To understand that we have to hold severity constant and compare across time.

A recent study tries to define "profound autism" as "nonverbal, minimally verbal, or IQ<50". They found a significant increase in US children aged 8 from 2000-2016 with profound autism. Non-profound autism increased much more, which makes sense given the broadening of criteria. The study is https://pmc.ncbi.nlm.nih.gov/articles/PMC10576490/

Anecdotally, any speech therapist with a long career will likely remark on a local increase of severe autism cases over the last 20-30 yrs. It's not as "skyrocketing" as ASD stats, but prevalence has likely increased substantially.

Building a prototype of a site/app to help teach my youngest child speak and understand language. He's five and doesn't speak beyond just a few syllables and 3-4 simplest words. He has something called childhood apraxia of speech, which is basically a condition where the brain doesn't know how to control muscles of the tongue/the mouth/the lips to create complex movements necessary for speech. These movements can be learned, but it can be a very very slow process. Sometimes it's just a few sounds that need to be learned or fixed, but with my son, it's very severe. Adam says "mama" and "papa" and can pronounce several vowels together with 2-3 consonants after 1.5 years of intensive speech-production-directed therapy. This month's achievement is he learned to purse his lips, which he never could before, and which you need for sounds like 'oo' (he still can't say the sound while pursing his lips). He understands much more (hundreds of words), but mostly in isolation, following rapid speech is hard.

There are apps that help kids on the autistic spectrum to communicate, and flashcard systems, and we're experimenting with these, but they're more geared towards encouraging the child to communicate. In our case, he communicates fine with gestures, nudges, pointing at things he wants, bringing flash cards of foods he wants, eye contact etc. And he seems to have good cognitive skills in terms of puzzles, basic arithmetics and counting, memory, etc. It is learning language as an auditory system that seems to be really difficult.

Adam can 'read' in the sense of knowing and recognizing all the letters (he takes delight in that) and pronounce the few syllables he's able to when he sees them written out (mostly consonants m,n,h with vowels a,o,e). His phonematic understanding for other syllables exists but is poor (e.g. he has trouble choosing between a BAH card and a PAH card when I say one of them out loud, whereas the letters B/P in isolation are easy). My idea is to build an app/site which teaches him and reinforces three-way connections [picture]<-->[written form]<-->[sounds] by letting him "type", initially by pecking at large squares with letters on screen, rather than an entire keyboard. So for example, there's a picture of him at the top, a row of 4 big blank squares underneath the leftmost of which is blinking, and 7-8 letters strewn around at the bottom, from which he can type in sequence A-D-A-M and get a sound effect of victory. For words he doesn't know or remember, there's a mode where he just needs to repeat e.g. C-A-T which is already written in identical squares in a separate row just above, then after a few successes the hint row goes away. For an MVP in which I can quickly backfill 100-200 simple words like that, and track progress, this would already, I think, be valuable; then maybe I can add a mode where the words sounds (with or without the picture) and he needs to type it.

If all this works for simple words, and he takes pleasure in typing, the stretch goal is to turn from words into short sentences, and both teach him phrases like I WANT [X], or WHERE IS MOM?, and let him request stuff with such phrases. None of this directly addresses the apraxia problem of actually learning to move his lips/tongue/throat/etc. appropriately, but I hope it can create more scaffolding around our efforts in that area (which we try very hard to work on daily) and together help him build an understanding of language/syntax. I'm very worried that, despite ongoing (very slow) progress in both speaking and understanding, phrases, sentences, syntax seem to elude Adam's grasp, and time is running so very fast.

I've been a backend/systems developer almost all my life, with not a lot of frontend experience (although I do know basic HTML/CSS/JS), and no app development. So I'm thinking for now to prototype this as a web page/pages, maybe using a lightweight framework rather than vanilla HTML (not sure), and let him interact with it on the iPad. I'll try to get the basic visual elements (picture/rows of squares for typed letters/bag of letters to choose from below) right with CSS/JS, and see if I can iterate from that. That's the idea, currently.

It's called "shadowing". Search for [shadowing language learning] to get to discussions of this method, variants, pros and cons etc.

An electron is just an excitation of the electron field, a vast probabilistic something that's ever-present everywhere in the universe, with a different complex-number amplitude (not observable directly) at each and every point. Sometimes that vagaries of its ever-changing amplitudes cause it to locally coalesce into something that looks like an individual electron if you squint at it in the right manner, and that vision persists until a strong enough interaction with another vast and nebulous field shatters the illusion, at least in the tiny corner of the probabilistic multiverse your local version of your consciousness brazenly imagines to be the objective reality. What kind of existence is that?

(Well, the only kind on offer, really.)

Thank you for everything.

Thing is, you and people like you won't come in any of the acts. Of anything. And that's arguably just.

What is the model name in the URL?

Very cool, thank you.

1. Apart from Project Gutenberg, where do your books come from?

2. When you proofread and fix typos, do you contribute the fixes back upstream?

I also did AoC 2021 in Zig: https://github.com/avorobey/adventofcode-2021

One thing the OP didn't mention that I really liked was runtime checks on array/slice access and integer under/overflow. Because dealing with heap allocation is a bit of a hassle, I was incentivized to use static buffers a lot. I quickly figured out that I didn't have to worry about their sizes much, because if they're overrun by the unexpectedly large input or other behavior in my algorithms, I get a nice runtime error with the right line indicated, rather than corrupt memory or a crash. Same thing about choosing which integer type to use: it's not a problem if I made the wrong choice, I'll get a nice error message and fix easily. This made for a lot of peace of mind during coding. Obviously in a real production system I'd be more careful and use dynamic sizes appropriately, but for one-off programs like these it was excellent.

Overall, I really enjoyed using Zig while starting out at AoC problem 1 with zero knowledge of the language. To my mind, it's "C with as much convenience as could be wrung out of it w/o betraying the low-level core behavior". That is, no code execution hidden behind constructors or overloads, no garbage collection, straight imperative code, but with so much done right (type system, generics, errors, optionals, slices) that it feels much more pleasant and uncomparably safer than C.

(you can still get a segmentation fault, and I did a few times - by erroneously holding on to pointers inside a container while it resized. Still, uncomparably safer)

I'd love to understand virtualization/emulation better, on the technical level. If I understand correctly, in modern CPUs it's all done with the help of specialized support in the instruction set, but I've never learned in depth

   - how that works
   - how it differs between Intel/AMD/ARM
   - whether there are competing approaches
   - what are the main software project "players" in the field
   - which of them are open-source with humanly understandable 
   source code

Is there a summary/tutorial about all this stuff someone could recommend (or write)?

Can you give some links to these recent papers? Sounds interesting. Thanks!

Thank you.

You should use this opportunity to recommend an especially high-quality introduction/overview of modern history of science, approachable by a curious outsider.

You're fine with the change because of one hypothetical PoC you imagine who likes it, but you resent that others are not fine with the change and feel supported by one real PoC who dislikes it.

'master' is better than 'main' because it successfully conveys the mental picture of the branch from which others are typically cloned. 'main' doesn't carry that connotation.

You could of course have branches that evolve in parallel, even without common code if you want. But what people almost always do is have one primary branch from which exact copies are made, given other names, then continue to evolve either more slowly (stable releases) or faster (speculative work). You could call that primary branch "primary" or "main" or "premier" or "trunk" and they all work, but "master" conveys that expectation very successfully, and that's why people tend to like that name.

The point is the terrible ease of applying a double standard in how you approach an issue.

Some private actor X performs an action Y which other people Z find reprehensible. The action Y is within X's legal rights to perform.

You can focus on how reprehensible Y is and how Z are right to condemn it. Or you could focus on how X should be totally free to do Y if X so desires, even if we don't like Y.

What usually happens is that if you feel Z are right or you wish to support Z or you wish to not be seen as supporting "enemies" of Z, you will focus on condemning Y. It won't even occur to you to emphasize that doing Y is legal; if pressed you'll freely admit it is, but to you focusing on how Y is legal will look like hypocritical attempts to evade the real issue, which is the terribleness of Y.

On the other hand, if you dislike Z or like the "enemies" of Z, you will focus on how Y is legal and how Z's dangerous rhetoric about Y poses a real danger of conflating Y with actually illegal acts. You might or might not agree that Y is terrible, but to you it will seem a decidedly minor concern compared to the dangerous rhetoric issuing from Z.