dinobones

Just start pricing in bytes input/output. This whole "token" and "tokenizer" thing is an implementation detail that shouldn't even be leaking out into the API.

Providers change tokenizers all the time with model updates, and it's often not even possible to query/figure out how text is tokenized without actually just sending the LLM a request.

Just switch to charging for bytes of intelligence. Please. Claude Shannon figured this out decades ago.

Don't underestimate corporate waste. If it's not someone's job to care for something, they really won't.

Even before this AI wave, it was common for me to see spinning dev environments for like $3k/month that hadn't been used in months on AWS.

Yeah, this implementation and their behavior these past few weeks is especially laughable when you consider that they consider themselves “philosopher programmers” or whatever.

You would think they’d be more reflective and introspective about these brash moral decisions. Their product quality is akin to my CS capstone lab group.

Those walkways are only in between each set of gates, there aren’t any actually near gates or anywhere near seating. Where did you sleep lol?

I've noticed that all of these music generators suffer from something like "mean" collapse (as opposed to mode, you do get variance, but all results are highly centered around similar sounding songs).

The music is all just very average, it sounds like the most average song with the most average chord progression/drum pattern per genre.

I guess that makes sense if these are most likely next audio token predictors... but it'd be cool if there was a way to inject some type of creativity/novelty into these, or at least tune up the temperature.

Everything so far just sounds like stock library music to me.

Couldn't someone just uhh... patch their macOS/kernel, mock these things out, then behold, you can now access all the data?

If it's not running fully end to end in some secure enclave, then it's always just a best effort thing. Good marketing though.

Wait, the linter re-ordered the file. Let me restore it to the previous state.

whisper: There is no linter.

why tho? it's just an alternate alphabet/set of symbols.

And given that in Austin they just reached parity with Waymo (i.e. completely unsupervised robotaxi service), they are not doing badly.

There is no unsupervised robotaxi service in Austin and there won't be, for years, if ever. Just like the way "FSD" is not fully self driving and likely never will be.

There are 67 million baby boomers in the US. How can you rationally blame them all? Roughly 20% of the population.

Saying the "boomers ruined everything" is not sophisticated, we can't move forward from a blame game, we have to diagnose the actions and actors that implemented them, but of course this is much more challenging.

Ancedotally, I know plenty of poor boomers. Have you seen who works at a Dollar Tree lately?

The popular dialogue that boomer=rich and greedy, millennial=poor and exploited is not productive, it's a fabricated generational war that distracts us from the real issues.

Has any society ever behaved that way? It's already a push to get people to think of the middle/lower classes during the present.

I understand the desire to find an entity or group of people to blame, but they were acting in their own self interest at a peak time, they didn't know the party would be over soon, for many of them, it still isn't.

Can you blame them for existing during early globalization, before over the financialization of everything? It's not like they actively took more than they "should have" from anyone directly, it's a consequence of their local economy and where it was at the time.

It's not that hard to notice this, just google "{university} {degree} syllabus" and you can see all the courses that the student will take.

In my case, I have CS degree and work as SWE but I probably would've been fine with just my Data Structures & Algos course as I already had programming experience.

Are computational theory, circuits 101, discrete math, logic 101, etc necessary for being a good SWE? Probably not, but they do probably expand your mind a bit.

A brief history of programming:

1. Punch cards -> Assembly languages

2. Assembly languages -> Compiled languages

3. Compiled languages -> Interpreted languages

4. Interpreted languages -> Agentic LLM prompting

I've tried the latest and greatest agentic CLI and toolings with the public SOTA models.

I think this is a productivity jump equivalent to maybe punch cards -> compiled languages, and that's it. Something like a 40% increase, but nowhere close to exponential.

I used the early web. I miss forums, I miss the small webmaster, I miss making fun, small websites to share with friends.

And while you could make the argument that these forms of media were superior to TikTok, I’d also argue that this is mostly just taste.

While we have closed ecosystems now, they’re much easier to make and share content to than the web of the past. It’s much easier to get distribution and go viral. There’s also a well trodden path to monetization so that if you craft great content people love, you can make a living from it.

Yeah quirky designs, guestbooks, affiliate badges, page counters, all that stuff. I miss it. But only ever a very small fraction of society was going to be able to make and consume that stuff.

This new internet is much more accessible and it occasionally produces diamonds of culture, you just have to know where to look.

So no, I don’t think any amount of decentralized protocols or tooling or any technology really can change this. I think this trend is set and will continue, and I’ve had to learn to be more open minded to how I perceive internet content.

No one is going to make personal websites or change their behavior in a major way.

Look, you can still sign up for free web hosting and make an HTML page and tell your friends. There are still people that do this. But it’s naturally eclipsed by these other methods of much easier content sharing.

The point is the content itself, not the packaging. Just get over the shape of the packaging and enjoy.

You spent 3 months on this hacked together garbage when you probably could’ve just configured a pre-existing solution off the shelf with like 10 minutes of reading and understanding documentation.

This blog post reeks of “you can just do things” type of engineering. This is the quality of engineering I would expect from “TPOT” (that part of Twitter) where people talk about working 12 hour days. It’s cause they’re working 12 hours on bullshit like this.

Building some sweet custom codec or binary transportation algorithm was barely cute in like 1989. It definitely ain’t cute now.

How many of these AI and “agentic” companies are just misled engineers thinking they are cracked and writing needlessly complex solutions to problems that dont even exist?

Just burn it all down. Let it pop already.

The year is 2034. Countless attempts at re-producing the sophisticated wetware of the brain have failed. Modeling research has proved unfruitful, with the curse of dimensionality afflicting every attempt at breaking the walls of general intelligence. With only a few million of capital left, and facing bankruptcy, they knew that only one option remained.

"Bring me the rats."

It's becoming challenging to really evaluate models.

The amount of intelligence that you can display within a single prompt, the riddles, the puzzles, they've all been solved or are mostly trivial to reasoners.

Now you have to drive a model for a few days to really get a decent understanding of how good it really is. In my experience, while Sonnet/Opus may not have always been leading on benchmarks, they have always *felt* the best to me, but it's hard to put into words why exactly I feel that way, but I can just feel it.

The way you can just feel when someone you're having a conversation with is deeply understanding you, somewhat understanding you, or maybe not understanding at all. But you don't have a quantifiable metric for this.

This is a strange, weird territory, and I don't know the path forward. We know we're definitely not at AGI.

And we know if you use these models for long-horizon tasks they fail at some point and just go off the rails.

I've tried using Codex with max reasoning for doing PRs and gotten laughable results too many times, but Codex with Max reasoning is apparently near-SOTA on code. And to be fair, Claude Code/Opus is also sometimes equally as bad at doing these types of "implement idea in big codebase, make changes too many files, still pass tests" type of tasks.

Is the solution that we start to evaluate LLMs on more long-horizon tasks? I think to some degree this was the spirit of SWE Verified right? But even that is being saturated now.

I stopped listening to Lex Fridman after he tried to arbiter a "peace agreement" between Russia and Ukraine and claimed he just wanted to make the world "love" each other more.

Then I found out he was a fraud that had no academic connection to MIT other than working there as an IC.

Mega cool, I’m curious if there’s a way to burn the ISO to a disc and get this playing on a physical console?

I’ve felt the same. Also the AGI outcome for software engineers is:

A) In 5 years no real improvement, AI bubble pops, most of us are laid off. B) In 5 years near—AGI replaces most software engineers, most of us are laid off.

Woohoo. Lose-lose scenario! No matter how you imagine this AI bubble playing out, the musics going to stop eventually.

This is such a good idea!

Kids music toys are often just purely toys tap a button, make a sound... But the skill ceiling could be so much higher, offering the ability to learn and express themselves more. Awesome work.

So much text and not a single example, diagram, or demo.

I'm honestly skeptical this will work at all, the FOV of most webcams is so small that it can barely capture the shoulder of someone sitting beside me, let alone their eyes.

Then what you're basically looking for is callibration from the eye position / angle to the screen rectangle. You want to shoot a ray from each eye and see if they intersect with the laptop's screen.

This is challenging because most webcams are pretty low resolution, so each eyeball will probably be like ~20px. From these 20px, you need to estimate the eyeball->screen ray. And of course this varies with the screen size.

TLDR: Decent idea, but should've done some napkin math and or quick bounds checking first. Maybe a $5 privacy protector is better.

Here's an idea:

Maybe start by seeing if you can train a primary user gaze tracker first, how well you can get it with modeling and then calibration. Then once you've solved that problem, you can use that as your upper bound of expected performance, and transform the problem to detecting the gaze of people nearby instead of the primary user.

Thought this was going to be a good article then the author started mentioning water consumption and I stopped reading.

$30 for a sock even in 2025 seems pretty steep. In 2004 is crazy. I guess I'm forgetting how "overpriced" Apple was at the time.

Here’s another idea:

We’ve had GPT2 since 2019, almost 6 years now. Even then, OpenAI was claiming it was too dangerous to release or whatever.

It’s been 6 years since the path started. We’ve gone from hundreds of thousands -> millions -> billions -> tens of billions -> now possibly trillions in infrastructure cost.

But the value created from it has not been proportional along the way. It’s lagging behind by a few orders of magnitude.

The biggest value add of AI is that it can now help software engineers write some greenfield code +40% faster, and help people save 30 seconds on a Google search -> reading a website.

This is valuable, but it’s not transformational.

The value returned has to be a lot higher than that to justify these astronomical infrastructure costs, and I think people are realizing that they’re not materializing and don’t see a path to them materializing.

I was so confused by this article.

I was confusing it with TinyPilot, a hardware KVM made by an indie hacker Michael Lynch, that I think has since been acquired.

Dunno if this passes the bootstrapping test.

This is sensitive to the initial candidate set of labels that the LLM generates.

Meaning if you ran this a few times over the same corpus, you’ll probably get different performance depending upon the order of the way you input the data and the classification tag the LLM ultimately decided upon.

Here’s an idea that is order invariant: embed first, take samples from clusters, and ask the LLM to label the 5 or so samples you’ve taken. The clusters are serving as soft candidate labels and the LLM turns them into actual interpretable explicit labels.

This flavor of "FOMO" is new and tuned for capitalism/materialism, hence everyone wants to be go to trendy restaurants/travel/airbnb lifestyle.

But before "FOMO" used to be religious, more like "FOGTH", or fear of going to hell.

So people were mostly happy living simple, sweet lives. Spend time with family. Raise your children. Have a simple job. Just don't go to hell, so go to church, pray, don't sin. Everything will be great, as long as you don't go to hell.

Probably in like the 60s did consumerism become the mainstream religion and it's been taking over since. Now you *must* make more money to: buy a house, take a fancy vacation, live a luxurious retirement, etc.

The cringe "hustle culture" of today is because for some people, it's their spiritual fulfillment. It is their religion. Their main focal point of existence is to buy bigger, buy better. It's almost taboo to consider an early retirement, "omg, I'd get so bored! I'd go crazy!"

How dare you not follow my religion of selling B2B SaaS? You are not a go-getter, and I am. Did I mention I also have a podcast?

On paper, nothing will happen. The wealthy in America will continue to get wealthier. Real estate and stocks will soar in value.

In reality, the dollar's true value will plummet. The FED is starting to lower interest rates again. We are likely going to undergo brutal inflation.

Crashing the economy is obviously very politically unpopular. The left/right will do whatever they can to keep this charade up, even if it means dooming the working class and throwing them some kind of bone to make them think they're ok.

The COVID pandemic was a good example of this. The working class got thrown a $2,000 check while there was billions given to bail out businesses/lots of fraud. Not a lot of people cared because hey, we got a $2k check... Even though that $2k check was not even close to maintaining their relative wealth pre-pandemic due to all of the government's inflationary measures.

There won't be a recession, it won't happen on paper. But the middle/working class will continue to be squeezed. And there will be programs to "rescue us." Maybe it's low cost home programs, maybe it's community college, I'm not sure. But I am sure it will never truly benefit the working/middle class, it'll just be a token to keep them from fully dying.