dimbletimbers

“The agentic future”

Yes, but Rex is built on libghostty with the multiplexing and agent integration built in, built by people I trust who don’t let the LLM write all their docs.

I prefer the TUI in ghostty and even that’s just a stopgap until superlogical is out of alpha.

These days, I constantly find myself filled with a kind of techno-dysphoria, where not only do I not want to use the tech in the way it’s being packaged/sold, but I’m filled with dread and disgust by the idea of people who do so breathlessly anticipate and proselytize its arrival.

The idea that anyone ever read a Gibson novel and thought, “can’t wait,” truly boggles my mind.

> But.... That's not knowledge?

Thank you! This is a huge sticking point for me. There is no knowing the code without experiencing the code. If nothing else, you have to read the tests to know they verify what you think they do. Even then, you might know it does what you want, but can’t know it doesn’t do anything it shouldn’t.

There are no shortcuts to true understanding.

This is a wildly online take. Humans highly value authenticity and many professional writers still write their first draft with a pen or pencil.

David Foster Wallace famously described writing with a pen and paper instead of a word processor because it forced him to slow down enough to think deeply, which I find very relevant to this whole debate.

There is no substitute for slow thinking. Your brain can’t really “move up the stack” until you understand the layer you’re on. All abstractions paper over important details. All shortcuts are cut short. You cannot produce quality writing without doing the slow, dlaborious work of choosing your own words.

I think devs and people on HN way overestimate how much the average person knows about or cares about or desires to interact with a computer. Whatever the job becomes, if you like talking to the computer and know how to get it to do anything besides show an email or youtube, you’re ahead of the curve and can likely stay there.

The most passionate, intelligent people 1. aren’t spending a lot of time on smartphones, video games and crypto and 2. aren’t as much smarter or better at predicting the future as the barely above average people who mistakenly count themselves in their ranks like to think.

It has been generally agreed since at least Plato that it’s up to people making a positive assertion to present evidence, not up to skeptics to prove it’s not possible. N.B. Russell’s Teapot Analogy.

You have to prove it is/can be conscious, not the other way around.

I feel like the AI debate is our last best chance to agree that human values are more important and have higher long-term utility (to humans) than pure financial maximization. I think it’s very very likely that money does not capture or convert to a large subset of morally good choices we can and should make to be happy.

I can’t tell if you’re defending this as good or just hypothesizing about the underlying cause but “appeal to nature” is one of the many freshman-level logical fallacies that seem ubiquitous in AI hype circles. Just because it’s what our animal brains do, doesn’t mean it’s going to be good or advantageous to our current evolutionary context.

Our energy conservation system also encourages us to eat as much high fat, high sugar food as we can get down our gullets, but that turns out to be a big problem for modern humans.

The blasé attitude with which people talk about “moving up the stack” as if it’s a foregone conclusion that it’s the right thing to do or even an accurate description of what’s happening is easily one of the most disheartening aspects of this whole movement.

Doesn’t that always seem to be the problem with the booster rhetoric. The upside is always somewhere in the hypothetical future while the downsides are already being acutely felt.

This is another case of anthropomorphic language for AI failing us. Who could be opposed to “learning?”

If I learn from a textbook I stole, some people maybe be thrilled I learned something, but stealing is still a crime and hopefully always will be.

A defense of human code review I wish I saw more often, especially in light of the concerns people have about cognitive/comprehension debt: comprehension redundancy. At the end, if taken seriously, at least two people understand how the feature works (even if that number is, on average, trending closer to between one and zero). Ideally at least one of the two also comes away with a better understanding of the wider system and how the feature fits into or stands out from that landscape.

That’s a tragically sophomoric take. 'object a has property x' is too generic to be meaningful. “AI has the property of conscious” requires settling the small matter of what the properties of consciousness are. And “behaves like it has property x” is not the same as “has property x.” I may behave like I have a million dollars and have not a cent to my name. Many things evolve to appear like many other things. And I have never seen an LLM do anything at all unless turned on and fed input tokens.

> To make a statement like that in the opening lines as if its obvious seems genuinely presumptous.

Then you must not be familiar with Bertrand Russell’s teapot analogy. The TL;DR is this: “Many orthodox people speak as though it were the business of sceptics to disprove received dogmas rather than of dogmatists to prove them. This is, of course, a mistake.”

Can you prove they are? I think “AIs are conscious” is an extraordinary claim requiring extraordinary evidence. The burden is on the claimant to provide evidence it is conscious, and I suspect any evidence you can provide has alternative explanations. The majority of claims I’ve seen like this amount to “how do you know humans are conscious in a way that’s distinct from a next token predictor.” I do not find those claims compelling.

People use online dating because they feel like they have to, but most people hate it. So maybe the analogy is even better than stated.

Not the OC, but I’ve done something similar and it wasn’t permanent but it was effective short-term after listening. My idea was to match the pitch, turn up the volume until it was fairly loud, then shut off the sound, and when I did that the pitch was gone or greatly diminished for about a day.

There doesn’t need to be a conspiracy born in smokey rooms for there to be a malignant culture of greed and moral bankruptcy. And it’s possible for one group or class to consolidate enough power to constitute an unfair advantage.

“I’m just doing what the market rewards” and “if I don’t, the next guy will” are the 21st century billionaire’s “I was just following orders.” It creates a race to the bottom.

If it were true

I’m just waiting for 0.13 so I can have Helix-like multicursor in Nvim, then I’ll never touch my config again.

I think I’m angry b/c I’m unsure but so many people are claiming to be sure without sufficient evidence.

It’s funny, I’m in a similar boat (omarchy at home, macOS at work) and recently spent a lot of time working out a system with SIP-enabled yabai for window management, kanata for the super key and other QMK-style tap-hold-chord mapping, and hammerspoon to fill in the space management commands for which yabai needs SIP disabled.

If this had existed last week, I’d have saved a few hours, but it was a kinda fun few hours and now it’s all mine. Glad to know I’m not the only one on this particular quixotic quest.

Legitimately curious, are you using the 1.4 canary release in a production environment?

Not sure how well “new ambitious software” and “____ clone” go together there.

Seems like even most critics would agree LLMs are capable of producing impressive-for-the-timescale clones of existing software, but is that valuable?

I think a majority of my friends are teachers and this could not be further opposite their sentiments. They feel strongly that it’s dishonest, nearly akin to plagiarism to use AI output in lesson plans. They don’t use it for grading or data analysis b/c of privacy concerns. They mostly complain that the kids are using it for homework, which makes it hard to grade both because it’s too uniform and because there’s no good faith effort going into the work which makes reading it unrewarding. And the consensus is that test scores are down b/c they’re offloading normal course work to AI. It’s basically the bane of their existence.

The idea that it’s actually better for diversity feels like a wild reach, almost a canned talking point.

I think the connection is pretty clear: people have always wanted to know things without having to learn them.

I’m sorry but this is a bad take. We’re suffering from a healthcare crisis, globally. We don’t have nearly enough doctors. Even in the US where wait times are relatively low, it sometimes takes months to get an appointment with my PCP, let alone a specialist. And while I do sometimes feel when I see an NP at an urgent care like I could have used an LLM for all the time and attention they paid to my concerns, that’s not a compliment to the LLM.

Besides that 60-70% is not an acceptable hit rate and there’s already early evidence of skill atrophy in AI saturated fields like radiology. We need way more people offering more personalized care, using more judgement and offering more compassion, not a chat interface that says “you’re totally right to push back, the post nasal drip was the smoking gun.”

Maybe some forms of ML, even LLMs, become part of the solution to that shortfall, but we’re a long way from proving that’s a net positive let alone a big leap forward.