squircle

Why is that? Did I make myself look dumb? Or, is this a comment to the author of the original article?

May be? Stand in a shady park and then walk next door to the asphalt parking lot. Is there anything inherently wrong with strong thermal updrafts? Perhaps not but, prove it. Or, prove in a bio regional sense this sort of top down, market driven planning isn't offloading entropy on others. How can we go on pretending that we are not all intimately connected and all related in one way or another? You would not want your children to suffer but it is okay for others?

I work across the stack but necessarily use the browsers end users are working with. Also, I can't pay my fecking credit card on Linux or any browser other than Edge. Whatever tho. A hammer is a hammer.

Here is the distinction the author makes:

> Not having fucks left to give is when you stop caring what others think.

> Not giving a shit is when you stop caring.

Burnout? Sounds fair. However, there's a point beyond burnout, IMO, when failure is not an option, where you exceed your capacity and limitations and are no longer tethered or limited by feelings of exhaustion. Then, the question is: why should I care? For what should I care about?

I fundamentally disagree with the notion that AI is a species or alien intelligence. It is a reflection of humanitys individual and collective intelligence fused with computers. If we reify it and give it agency, sure, it seems alive, but it isn't. (Talk about false idols...) Now, here we are, where so many beautifully intelligence and well meaning people are advocating we put all our trust and agency into a clock? This is no better than freaken ancient cultures divining meaning and answers from the position of stars. It's a regression.

Imagine a still pond. If I throw a stone, the water ripples. That's all AI is. Are you struggling with this? Consider the information flowing through the algorithms like water or a new/emerging natural force that is lighter than air.

> Instead of having an hour commute to move a few miles, you could have a half hour commute to move 30 miles.

Traveling faster than what the human body is capable of on its own feels like time travel to me. Horse, buggy, car, whatever... like stepping into or activating a warp bubble where your consciousness arrives at a place faster than humanly possible. Similarly, having access to information or experience (different manners of vehicles) is potentially a huge advantage or major pitfall. (Perhaps why some ancient maps indicate, "here be dragons!")

Inversely, moving slow when others are traveling fast allows you to witness where those paths lead without having to go down proverbial rabbit holes.

Eh, one man's trash is another man's treasure. I agree with you, I think, that the grass is always greener...

Having a child similarly changed my life. (My father died when I was 7, so I understand intimately what it would mean to her if I was gone. By now I have outlived him and have zero correlates for navigating life, except what I've observed in others paired with the narrow and provincial life experience gathered trying to survive.)

Here's the rub though... if there was a threat to your son's life, or anyone you care about really, would you have the aptitude to put yourself in harms way, take a bullet or otherwise lay down your life so they might live to see another day? If your family was starving in the middle of winter, would/could you adeptly brave the elements to keep them alive or, all else lost, allow yourself to die so they could eat your flesh and live another day? Rather dark, I know. But, modernity with all its trappings does not prepare us for the unknown. Yet the unknown is part of our individual and collective evolution.

The downside of surviving and returning to a safe and comfortable existence is anhedonia. Everything is grey. Every day feels like a slow death. There is nothing important to do because the first world problems and mild activity of daily life are now incomparable and inconsequential. And the vast majority of people you meet have never had to make truly hard decisions. This is an observation, not a judgement. It is a lonely feeling.

They would be arrested for over prescribing now. If you look at the state of the world 20 years ago or more it looks much different from here.

It sounds like they have little skin in the game. Does your org reward hard work and hitting targets? Is the company doing mission critical work or feature farming? I dont think there's a company I've ever worked for where the status quo wasn't a full on regression to the mean average of the culture of the place. Conway's law? They were hired for a reason but no one can be firehosed with bullshit for long, before simply giving up if there's no compelling reason to continue striving.

So, have you spoken to them? Or, put them in a position where there failure to deliver will have real world negative consequences to people on the team they care about?

For me personally, I seem to only operate well under intense pressure and heavy load. Maybe they're bored. Maybe there's confounding variables outside of work they need to put time and attention into before they're unblocked and consistently in a high delivery flow state. (Or, perhaps, you need to tell them: we want you to try harder to fail.)

Edit: also, work expands to fill time. Intellectual and creative work accomplishes things in leaps and bounds or fits and starts IMO. Punching a time clock is terribly draining... Is there a shared team vision?

Paywalled... however, it now seems poignant to save and invest in printed texts, invest in personal and community libraries, etc. (And be highly skeptical of anyone advocating the burning of books for whatever reason.)

Cognitive decoupling is an awesome concept.

Interesting this advocates against unbridled free-market capitalism. An issue I see right now is corporate personhood. This, coupled with the tendency for humans to anthropomorphize and see aspects of themselves in LLMs may be a hazard. But hey, computers are really just an oscillators/clocks at their core, right? (If there's any built in logic to the US simultaneously witnessing "No Kings" protests paired with a dictatorial display of military power makes me think the trend is/must necessarily be to finally and completely turn away from cults of personality in any form or fashion.)

Also, noting the bio-singularity, if our built environments change our internal clocks and biorhythms to the point where we're not compelled to mate, perhaps its time to similarly consider the tendency of the "free market" (which, ostensibly, is not free at all if it's bottlenecked through institutional, organizational, and individual decision makers rife with their own biases.) But, at the same time, to automate economic decisions feels similarly foolish. Imagine the bias an LLM with tool-calling capabilities might exhibit.

Literally nothing at all.

Years ago I was having terrible chest pains and irregular heartbeats which, I don't know, somehow intuitively landed on my blood being too "thick" (this is what it felt like, along with cognitive fog.) I gave blood, upped my intake of dark leafy greens along with herbs like cilantro/parsley (and cut down on bananas.) Problem solved.

Thanks Dustin. This would be acceptable... however, this is an interdisciplinary effort and, perhaps ironically, absolutely requires mentors and peers to fill in the gaps (because my knowledge is inherently imperfect.) I could use LLMs to help with the research but... what to do about the matter of provenance, tradition, and lineage? Is this not the spirit of the thing?

And, well, is this a good thing?

Ah, okay. How about JSDoc style comments?

Have you considered using a README (or a ./docs folder) as a source of truth? I was working in this direction (before and after gaining access to LLMs.) I rolled about half the tooling for this (in Rust and JS) and have proven the effectiveness, at least for my own use cases. But, its so ridiculously fast that... I don't know what to do with it. Make a bunch of open source tools? Launch innumerable SaaS'?

The build step is crucial (along with solid system prompts.) This keeps the human in the loop and makes for something that truly feels like rapid/iterative development. Build steps also allow for checkpoints, and provide for something that, from a layman's perspective, feels like reversible computing.

Although I'm leaning heavily away from being passionate about software development, this is a cool project, and its freaken awesome how anyone can now reinvent the wheel from first principles.

Was detachment 418 already engaged? Joking aside, is there anyone on this forum who has ever been to war who would advocate for war being a desirable end in itself? That is, must conflict, if an inevitable aspect of life, always perpetuate itself? Or, if tension in itself can be good and necessary, does it have to resolve around suffering and destruction? (Its a thrilling experience to explore the unknown, push yourself beyond known limits, or fire a weapon, for instance... but none of this necessarily implies harming others.) Are there examples of win-win scenarios where all sides involved feel heard, appreciated, and respected? Invariably, technology (as embodied culture) is the double edged sword of civilization which is useful and necessary for individual and collective health and well-being. At a certain point, might the lot of us as a globally aware individuals realize that, through historical contigency, have lost the thread and are clinging to false dichotomies (man vs nature, etc)? What happens when an immovable object encounters an unstoppable force?

Thank you for the thoughtful response.

> The consequence of imperfect knowledge is that you're going to be wrong.

How would you even know? Does breaking your thoughts and experiences into discrete signals help you, as an author, understand your own biases and why they are valid or not?

I suppose a tangential question then is, why publish? Is there value in putting ideas into the world beyond entertaining/entraining an audience or having your ego stroked? Is publishing a self-sacrificing act of, here is where I am at, perhaps it will provide some value to you (the audience) or, perhaps you will tell me (the author) what I'm not seeing?

My gut feeling here is that complex software systems have critical functions whose usefulness is deterministic reliability. On the other end of this spectrum, interface and experience are primarily subjective and relatively non-deterministic . That is, relying on automated tests at the level of HCI, as opposed to rigorous testing by engineers and an open feedback loop with end-users, forces all humans working with these systems into a reinforcing negative feedback loop where only the squeekiest wheel is greased. This is an abstract and off the cuff reaction (without having read the entirety of this article) but, I think the point I'm attempting to make is: process and method are important so, what approaches actually lend themselves to the best outcomes for all involved?

So, a bit like separating the grain from the chaff or, organizing all these different life experiences into an understandable and meaningful whole?

Of course... "What about Bob?"

You might enjoy roasted dandelion root tea. The taste is about as earthy as it gets. Also, as a bitter herb it's great for digestion.

I think it's both. If you look at the response coming back from claude you can get an idea for how the front end library probably works. (When updating an artifact, claude's response is inefficient. Perhaps this is more reliable at scale but... why not use a diff format similar to git?)

I was "vibe coding" a rust project the other day and, working on tests, the LLM removed the safety function added to keep the thing from accidently nuking my system. The the test case was confidently resolved.

Thank you. Some variation of Circuit Tracing[1] is what I was thinking of. It would be useful to visualize the tokens effect on inference in real time, and tweak the input parameters, individual tensor weights, etc., and watch the inference collapse on a different result (does output of LLMs need to be next token based? Would larger, pre trained/synthesized tokens with known and expected output allow for faster inference, or is this simply tool calling?)

I understand LLMs and ML models are not linear functions but, they are functions (no?) and some complex higher level math can be grasped intuitively when visualized.

[1] https://news.ycombinator.com/item?id=43495585

Edit: thoughts.

Forgive my ignorance but is it possible to visualize neural net activation during inference?

If your game is web based, toneJS may work for developing a prototype. Using basic wave forms and messing with various filters can produce piano like tones.

Not until its published? Have not read the article but I thought this was common knowledge.