"[T]he Sinclair ZX80 tokenized BASIC keywords at the keyboard. For instance, after typing a line number, the cursor would flash with an inverse K, indicating that the next key you hit would insert a corresponding keyword that could start a BASIC line: for instance, typing V then didn’t produce a V but instead inserted the keyword GOSUB, typing Y inserted REM, and so on." kinda reminds me of the MegaZeux Robotic editor
"[S]ometimes I wear a cool Hawaiian shirt on Fridays, and it’s commonly accepted that bad people don’t wear shirts with flowers on them. That’s just a fact."
"[T]hat’s not cool. That’s an ad hominem, and very immature of you. What do you have against a mixture of dried carbohydrates and powdered seasonings?"
"...[T]here is zero, less than zero, stress put on the relation between those two 'sides,' or their histories, or their sponsors, or their relative evidentiary authority, or any of it. Instead, what you get is a piece making the various more or less bovine noises of studious grey-lady impartiality, with the labor of anything resembling 'appraisal' surgically excised."
"There were a tiny handful of incredible nerds who thought [Zork] was fun, mostly because 3D graphics and the physical touch of another human being hadn't been invented yet. But for the most part, people would tire of the novelty because trying to guess what to type to make something happen is a terrible and exhausting user interface. This was also why people hated operating systems like MS-DOS, and why even all the Linux users reading this right now are doing so in a graphical user interface." I get that Dash is exaggerating here for effect, but the comparison between command-line interfaces and LLM chatbot interfaces rings false to me. Command-line interfaces have steep learning curves, but they're deterministic and composable, which is why we still use them today. The reason that chatbot interfaces suck isn't that they're based on textual inputs, but that they're neither deterministic nor composable. (GUIs have different affordances than command-line interfaces which make sense for some use cases, but they also have their own learning curves and limitations. A hypothetical GUI-based LLM interface would have all the same problems as a text-based one.) Plus the Zork series sold like 700,000 copies
"I think the best explanation is that in a vacuum modal editing sounds like a bad idea. The mode is global state that users always have to know, which makes it dangerous. To use new modes well you have to memorize all of the keybindings, which makes it difficult." good post but i would point to things like DAWs and trackers as modal interfaces (there's a record mode and an edit mode)
"... feel free to skip ahead to the Summary for a conveniently digestible tl;dr that spares you all the tedious details, yet still provides enough ammunition to trash-talk this post in the comments of whatever platform you stumbled upon it."
"Metastability is a peculiar problem where a digital signal can take an unbounded amount of time to settle into a zero or a one. In other words, the circuit temporarily refuses to act digitally and shows its underlying analog nature."
"Change is driven intentionally... by individuals learning from direct practice. And if you increase the size of the audience and simultaneously reduce the numbers of real practitioners, i.e. the people who drive that change in the ecosystem, you will first see diminishing returns, and then you will kill the ecosystem, and then you will kill the tool. [...] I believe the work of sense-making belongs exclusively to me. The problem is not insufficently condensed information, because the information is not condensable. The problem is an information environment that requires much more aggressive management."
this kind of feedback loop seems inevitable, and not just for Javascript frameworks, but for all kinds of things, and critics of "AI" have been saying this since the beginning. personally I think this is a really strong and obvious disincentive to use generative AI tools to generate anything, especially code: you'll always be trying to solve contextual, emergent problems with yesterday's cruft. also: "This is the new reality: If it’s not in the LLM training data, it doesn’t exist." phew what a relief
"Since I used my RetroUSB AVS for the NES footage to get its higher-quality HDMI output, and the Panasonic FS-A1F (an MSX2 machine) for the MSX footage so as to get its higher-quality RGB output, I've technically just compared the PPU (Picture Processing Unit, the NES' graphics chip) and the TMS9918A without actually using a real PPU or a real TMS9918A. Good job, Nicole." lmao