I mean, come on, look at this:

turns out software already was clay! we just had weak hands

I've often complained that interface design lacks great textbooks—still mostly taught through apprenticeship and trial/error. @joshpuckett's new interfacecraft.dev is shaping up to be a great resource, particularly for polish and craft. The step-by-step UI crit/redesigns are a really nice format. Interesting to imagine structuring material like that as Figma artboard sequences, so the student could try each step themselves before revealing the author's take (in the style of andymatuschak.org/doing-cent…)

I’m very happy to present my toy research project: Sotaku! It's a neural net that automatically discovered the rules of sudoku and learned to solve them, achieving a new state-of-the-art score of 98.9% on one of the hardest sudoku datasets, while being agnostic to the game, and beating all other sudoku-optimized neural net architectures* Read more for fun motivations, plus some extremely unconventional discoveries, e.g. reverse curriculum consistently beating curriculum (!), emergent reasoning-like capabilities, and the future of traditional programming

long timeouts are like low interest rates, show faith in the far future, faith in slow things worth doing

You might like this old paper about “read and edit wear” in software, like a scroll bar showing where you’ve read the most smg.media.mit.edu/classes/So…

Meet Gizmo: a new way to make playful, personal software—right from your phone. No code. No desktop. Just your camera, your fingers, and a good idea. [1/7]

Agreed - it’s been shocking to me how few devs I follow are working with AI. It uncovers mind-blowing opportunities that haven’t been possible before!! Too excited for the products I’m building!

Coding with ai actually feels a lot like playing civs - just one more turn

The dopamine rush of vibe coding on @cursor gives me the same rush as playing a video game I'm addicted to. I gotta watch myself!

We have two identical groups of 8 gauges with anomalous gauge at position (3,1). The degree of anomaly is identical in both groups ~ 8 deg. Humans have extreme hyperacuity with respect to detecting angular orientation of line segments—needle gauge is so much faster to read. The arc gauges all look the same even though the amount of deviation is the same, about 8 degrees. Even if you concentrate, it's hard to tell which one is off.

My mind changed re: programming with AI as soon as I realized how much of the stuff I do when "programming" is just tedious grunt work. Programming with AI actually frees me up to do more programming, not less.

What if a spreadsheet cell could hold multiple values at the same time? That's the idea behind Ambsheets, a project I've been working on w/ @geoffreylitt at @inkandswitch. It's a new spreadsheet that makes it easier for you to explore many possibilities simultaneously. 1/2

"Money can't buy happiness" is a skill issue, you heard it here first.

The story behind this is insane - idea -> proving the fundamental breakthrough working microscope in <6 mo. Could be a massive deal for ie single cell resolution across a whole mouse cortex - really, really excited to watch the applications for this new microscope. Congrats!!!

It's a sad state of silicon engineering that no startups can do this. "Silicon Valley"? More like "JavaScript Valley"

no animation only timing chart

the facebook pop framework and its downstream consequences on software: a thread (1/86)

Suppose that over the last few years, there's been an exponential rise in the amount of opium usage. It's especially prevalent among the youth. You've started to notice as you walk through airports and look at how people twiddle away their time, sneaking the occasional glance over the shoulder of a stranger, that quite a few people seem to spend those spare moments happily indulging in the drug. Obviously, this is not a great situation. Addictions rarely are. Also, some people worry about the specific opium den where most of the youth seem to go to get their fix and the way it may or may not make use of their information. People also have expressed worries about what level of involvement the government housing these dens may or may not have in running them, but maybe that's beside the point. It's a little more complicated than that because not all of this opium is bad, per se. Some of it seems to really help people, but there's no question that on the whole, it's all very addictive. Now suppose that you're a doctor, and part of your job involves prescribing drugs. Not ones like opium, at least you hope not, but ones that will actually help people and make their lives better. Moreover, a big part of the reason you got into medicine was not just to help people in general, but to help the youth in particular. In your heart of hearts, you know that there's a somewhat blurry line between the drugs that you prescribe and the opium people are addicted to. It's definitely possible for people to abuse the pharmacy where your prescriptions are dispensed, and as easy as it is to be critical of the opium dens, it's not like your own domain hasn't seen an addiction epidemic of its own. Not only that, but your own pharmacy has even recently opened a window specifically to dispense opium, and asking many doctors to try prescribing opium here, all in the hope of competing with the more popular opium dens. Still, you feel pretty comfortable that your drugs, at least, are actually helping people. The question is, should you explore channels of dispensing your drugs, or variants of your drugs, in the opium dens? On the one hand, it doesn't feel great to be a part of the growing addiction. But realistically there are a lot of people, especially the younger generation, that spend much more of their time in the opium den than in pharmacies and doctor's offices. You know it might be an avenue for reaching those people. Mostly, though, you just haven't thought about it much given that you yourself aren't an opium user. The immediate hurdle is that your drugs don't fit the form factor of what's dispensed at the opium dens. What you prescribe isn't the easiest in the world to use, it requires at least a little bit of patience and concentration, whereas most opium gives a more immediate hit. You could take some time to adapt your prescription or to create new drugs that more closely fit the spirit of the dens, but it's not really something you want to spend a lot of your own time doing. But then suppose someone comes along and offers to do that adaptation on your behalf. They'll take some of your existing drugs, pull out what may fit well in the dens, and invite people to come to your pharmacy if they like the taste of what they had. You can still spend almost all your time working on the medicines you feel passionate about, and even if it's not in the most virtuous context, a few more people may now have the chance to become aware of your practice. Worth a shot, no? --- Anyway, there now exists a 3blue1brown TikTok. My thanks go to Dawid Kołodziej for cutting up the adaptations.

i think you want to develop the habit of thinking this way whenever you see a 'list' of objects on your computer (bookmarks, tabs, git branches, songs, design layers, etc)

Every day I feel the immense frustration of wanting to quickly sketch something on my MacBook screen with my hands… but can’t. It’s too cumbersome to switch to iPad and my desktop apps don’t run there. Input formats are a tragedy.