frognumber

This was painful to read. It become better and simpler with a basic signals & systems background:

- His breaking up images into grids was a poor-man's convolution. Render each letter. Render the image. Dot product.

- His "contrast" setting didn't really work. It was meant to emulate a sharpen filter. Convolve with a kernel appropriate for letter size. He operated over the wrong dimensions (intensity, rather than X-Y)

- Dithering should be done with something like Floyd-Steinberg: You spill over errors to adjacent pixels.

Most of these problems have solutions, and in some cases, optimal ones. They were reinvented, perhaps cleverly, but not as well as those standard solutions.

Bonus:

- Handle above as a global optimization problem. Possible with 2026-era CPUs (and even more-so, GPUs).

- Unicode :)

Years ago, I picked cell carrier because of this. When I ran out, it switched to O(200kbps), which is fine for email, basic web search, etc.

It was actually a bit ironic that, at the time, you could burn through the whole high-speed quota in seconds or minutes, if you went to the wrong web page. Most carriers would stop or bill you an arm-and-a-leg after.

I think you're wrong, and you're underestimating the transformational impact of Ad-Words.

Free internet existed before paid internet, true, but mostly because people did things for other motives (like fun). Altavista was a tech demo for DEC. Good information was found on personal web pages, most often on .edu sites.

Banner ads existed, but they were confined to the sketchy corners of the Internet. Thing today's spam selling viagra. Anyone credible didn't want to be associated with them.

What Google figured out was:

1) Design. Discrete ad-words didn't make them look sketchy. This discovery came up by accident, but that's a longer story.

2) Targeting. Search terms let them know what to ads to show.

I can't overstate the impact of #2. Profits went up many-fold over prior ad models. This was Google's great -- and ultra-secret -- discovery. For many years, they were making $$$, while cultivating a public image of (probably) bleeding $$$ or (at best) making $. People were doing math on how much revenue Google was getting based on traditional web advertising models, while Google knew precisely what you were shopping for.

By the time people found out how much money Google's ad model was making, they had market lock-in.

Pick a university, and given them $1B to never use Windows, MacOS, Android, Linux, or anything other than homebrew?

To kick-start, given them machines with Plan9, ITS, or an OS based on LISP / Smalltalk / similar? Or just microcontrollers? Or replicate 1970-era university computing infrastructure (where everything was homebrew?)

Build out coursework to bootstrap from there? Perhaps scholarships for kids from the developing world?

An isolated monastic order in the hills around the Himalayas should ideally be completely isolated from Overwatch and .docx files.

Absolutely not.

Software is bloated in part because it's built in layers. People wrap things over, and over, and over. Stripping down layers is neigh-impossible later. Starting from scratch is easy.

Starting from scratch fails in practice because you don't get feature parity in time short enough for VC (or grant) funding cycles.

If we build a tech tree around 200MHz 32MB machines, except for things like ML and video, we'd have a tech tree which did everything existing machines do, only 10x more quickly in 0.1% of the memory. Machines back then were fine for word processing, spreadsheets, all the web apps I use on a daily basis (not as web apps), etc.

Need would drive people to rebuild those, but with a few less layers.

> Love this idea and wondering where that low cost of living place would be

Parts of Africa and India are very much like that. I would guess other places too. I'd pick a hill station in India, or maybe some place higher up in sub-Saharan Africa (above the insects)

> What problem are we trying to solve that is not possible right now?

The point is more about identifying the problem, actually. An independent tech tree will have vastly different capabilities and limitations than the existing one.

Continuing the thought experiment -- to be much more abstract now -- if we placed an independent colony of humans on Venus 150 years ago, it's likely computing would be very different. If the transistor weren't invented, we might have optical, mechanical, or fluidic computation, or perhaps some extended version of vacuum tubes. Everything would be different.

Sharing technology back-and-forth a century later would be amazing.

Even when universities were more isolated, something like 1995-era MIT computing infrastructure was largely homebrew, with fascinating social dynamics around things like Zephyr, interesting distributed file systems (AFS), etc. The X Window System came out of it too, more-or-less, which in turn allowed for various types of work with remote access unlike those we have with the cloud.

And there were tech trees build around Lisp-based computers / operating systems, SmallTalk, and systems where literally everything was modifiable.

More conservatively, even the interacting Chinese and non-Chinese tech trees are somewhat different (WeChat, Alipay, etc. versus WhatsApp, Venmo, etc.)

You can't predict the future, and having two independent futures seems like a great way to have progress.

Plus, it prevents a monoculture. Perhaps that's the problem I'm trying to solve.

> Do we start from hardware at the CPU ?

For the actual thought experiment, too expensive. I'd probably offer monitors, keyboards, mice, and some kind of relatively simple, documented microcontroller to drive those. As well as things like ADCs, DACs, and similar.

Zero software, except what's needed to bootstrap.

John describes exactly what I'd like someone to build:

"To make something really different, and not get drawn into the gravity well of existing solutions, you practically need an isolated monastic order of computer engineers."

As a thought experiment:

* Pick a place where cost-of-living is $200/month

* Set up a village which is very livable. Fresh air. Healthy food. Good schools. More-or-less for the cost that someone rich can sponsor without too much sweat.

* Drop a load of computers with little to no software, and little to no internet

* Try reinventing the computing universe from scratch.

Patience is the key. It'd take decades.

As a former EE, it's not just pay.

The cog-in-a-machine corporate culture is not fun. Tech culture is much healthier.

There's no upside to big electronics companies here.

My bill for LLMs is going up over time. The more capable, higher-context models dramatically increase my productivity.

The spend prices most of the developing world out -- an programmer earning $10k per year can't pay for a $200/month Claude Max subscription..

And it does better than $6k-$10k programmers in Africa, India, and Asia.

It's the mainframe era all over again, where access to computing is gated by $$$.

I think you're talking about much shorter timelines than I am.

That's all noise.

I always consider different options when planning for the future, but I'll give the argument for exponential:

Progress has been exponential in the generic. We made approximately the same progress in the past 100 years as the prior 1000 as the prior 30,000, as the prior million, and so on, all the way back to multicellular life evolving over 2 billion years or so.

There's a question of the exponent, though. Living through that exponential growth circa 50AD felt at best linear, if not flat.

Personally, my estimate of a rational valuation would be:

$1-2T with no legal risk.

$300B assuming a rational and uncorrupt government, which should, at some point, kick them back to non-profit status, and convict people for fraud

Of course, too-big-to-fail means this won't happen.

3 second with a web search would bring up citations:

https://duckduckgo.com/?t=ffab&q=gpu+lifespan&ia=web

You'll need to scroll past the ones talking about obsolescence versus failure. Toss in 'data center' if you like.

You'll see a range of numbers -- including some lower than I cited -- but it's all in that ballpark.

> Another possibility that has long been on my personal list of “future articles to write” is that the future of computing may look more like used cars. If there is little meaningful difference between a chip manufactured in 2035 and a chip from 2065, then buying a still-functional 30-year-old computer may be a much better deal than it is today. If there is less of a need to buy a new computer every few years, then investing a larger amount upfront may make sense – buying a $10,000 computer rather than a $1,000 computer, and just keeping it for much longer or reselling it later for an upgraded model.

This seems improbable.

50-year-old technology works because 50 years ago, transistors were micron-scale.

Nanometer-scale nodes wear out much more quickly. Modern GPUs have a rated lifespan in the 3-7 year range, depending on usage.

One of my concerns is we're reaching a point where the loss of a fab due to a crisis -- war, natural disaster, etc. -- may cause systemic collapse. You can plot lifespan of chips versus time to bring a new fab online. Those lines are just around the crossing point; modern electronics would start to fail before we could produce more.

There's a numerator and a denominator. The numerator is fine for what you're saying -- the number of merged PRs.

The denominator varies wildly based on whether or not the PR is made. If codex makes nonsense, I don't ask it to make a PR.

Missing data: I don't make a codex PR if it's nonsense.

Poor data: If I make one, I either if I want to:

a) Merge it (success)

b) Modify it (sometimes success, sometimes not). In one case, Codex made the wrong changes in all the right places, but it was still easier to work from that by hand.

c) Pick ideas from it (partial success)

So simple merge rates don't say much.

It's common knowledge. See, e.g.:

https://lunary.ai/blog/openai-developer-role

False.

There is a 3-level hierarchy:

System prompt > Developer prompt > User chat

You provide that middle level.

When I bought my car, TCO for manual was higher than for automatic. Base purchase price was about $500 lower, but required a pretty frequent maintenance schedule. Automatic was nearly maintenance-free (although the little maintenance required had higher unit costs).

I ran the numbers. Automatic won for cost.

For a cheap car, manual makes little sense for a rational consumer.

Ergo, they're only left on fancy sports cars.

I'll give a few examples.

* Learning biology, memorize terms like "anabolic reaction" or "reverse transcriptase"

* Learning algebra, memorize major groups like S_n or GL_n

* Learning statistics, memorize the major probability distributions, their means, and standard deviations

* Preparing for math contests, remember things like "Chinese remainder theorem"

That's a tiny part of learning, but it dramatically accelerates the other parts. At that point, when you're working through texts, you'll understand what you're reading without looking things up or thinking about it. And when you're engage in complex problemsolving, you'll have the knowledge ready.

Do this either on or before the first (surface learning) pass, and once they're memorized, use them in more advanced contexts (e.g. reading research papers, teaching, complex problemsolving, etc.).

All this stuff interconnects, and SR gives a fast, cheap way to start building out the simpler parts of the knowledge network.

People do dumb things in stressful situations. To your "I beat my wife because I caught them cheating" example, there's a world of difference between:

1) I walked in. An argument and a fight ensued.

2) I found out about it, went of and thought, and made the choice.

There's a hierarchy, including:

https://en.wikipedia.org/wiki/Provocation_(law)

https://en.wikipedia.org/wiki/Insanity_defense

I find it entirely probable that the judge didn't know or understand, in the moment, their situation and the implications of their actions. Indeed, I will go one step further. If ICE does illegal things 100 times, then it's reasonable to expect an unreasonable reaction maybe 10% of the time.

https://en.wikipedia.org/wiki/Proximate_cause

If I were a judge, and someone came into court with an "administrative warrant," I might not want them disturbing my courthouse either. I might want parties to feel safe there, and be concerned about miscarriages of justice if parties are scared to show up.

The trick here is to have policies ahead-of-time, and especially, to let judges know about this sort of thing ahead-of-time. If police show up at my door, I might make a mistake. If they let me know ahead of time, and I have time to think, I hopefully won't.

People do dumb things in stressful situations. To your "I beat my wife because I caught them cheating" example, there's a world of difference between:

1) I walked in. An argument and a fight ensued.

2) I found out about it, went of and thought, and made the choice.

There's a hierarchy, including:

https://en.wikipedia.org/wiki/Provocation_(law) https://en.wikipedia.org/wiki/Insanity_defense

I find it entirely probable that the judge didn't know or understand, in the moment, their situation and the implications of their actions. Indeed, I will go one step further. If ICE does illegal things 100 times, then it's reasonable to expect an unreasonable reaction maybe 10% of the time.

https://en.wikipedia.org/wiki/Proximate_cause

If I were a judge, and someone came into court with an "administrative warrant," I might not want them disturbing my courthouse either. I might want parties to feel safe there, and be concerned about miscarriages of justice if parties are scared to show up.

I'm curious what's responsible for the current uptick.

I'm not sure what you're asking.

Let's say I beat someone bloody. We can play through several scenarios:

- Someone broke into my house, and I was fearful for my life

- Plain clothes police broke into my house, and I was fearful for my life

Let's say a police officer did so:

- Someone was a gang member, and the police officer did so in self-defense

- Ditto, based on mistaken beliefs

A lot of the protections in place for police and judges are based on the fact that mistakes like these happen. In general, people aren't individually liable for mistakes make in their official capacity as a government employee, unless they cross very extreme lines. They might get fired, but not prosecuted.

There are exceptions (such as handling of classified materials), but as a guideline, if a police officer beats someone bloody, but has good reason to believe they were a criminal and that this was the least force they could use to keep themselves safe, they're protected even if they're wrong.

Ergo, I posted a link to an analogous legal situation in Massachusetts.

We can agree with what the judge did, but it doesn't make it legal.

We can also agree that ICE is breaking laws, but it also doesn't make what the judge did legal. It does help a bit -- in another comment I explained why -- but not enough to change the legal analysis.

As a footnote, modern LLMs aren't worse than Fox News. They have a lot of case law in their training set. They make mistakes so shouldn't yet be used for anything critical, but the legal analysis from Claude or GPT4.1 is a lot better than e.g. 95% of forum posts here.

I usually read coverage from different sides. If you don't realize where she screwed up, look at Fox News. If you don't realize where the FBI screwed up, look at NY Times.

Fox News perspective is that she broke court procedures in order to obstruct federal agents.

Prior cases seem to support that:

https://www.ice.gov/news/releases/massachusetts-judge-court-...

Case concluded with some kind of judicial reprimand (not criminal, but administrative). This one is further over the line.

Neutral description to LLM also supports that the judge acted improperly (but LLM didn't think this would lead to a conviction). LLMs aren't great at legal analysis, but are actually pretty good at pattern-matching cases.

One thing helpful to have is a lawful plan. The courthouse might have handled ICE without breaking protocols by having protocols. Protocols should be prima facie neutral, but it's reasonable to expect people in courts, schools, and other places we actually want them to show up to feel safe there. That shouldn't involve sneaking people through back doors or hiding them in jury areas.

The truth is somewhere in the middle.

There was a similar case in Massachusetts many years back. It never went to trial, and legal analysis could go both ways. The bargain struct was it would go into secretive judicial oversight channels.

There is a strong case to be made for obstruction of justice, and an equally strong case to be made about her making an error in her professional capacity as a judge and a government employee (which grants a level of immunity). Police officers, judges, soldiers, etc. make mistakes, but they generally don't go to jail for them because (even corruption aside) everyone makes mistakes. In some jobs, mistakes can and do have severe consequences up to and including people dying. If that led to prison, no one sane would take those jobs.

In any sane universe, it'd be fair to say she screwed up, and then the FBI also screwed up arresting her. I think the FBI screwed up more, since their mistake was premeditated, whereas she was put on the spot.

I do agree with your fundamental point of fatigue. This is not something anyone has a moral high ground to hang their flag on without looking bad.

You're downvoted because ICE did not have a warrant.

ICE prints pieces of paper which they call "administrative warrants." Those were never reviewed by a judge and are internal ICE documents. An administrative warrant is not an actual warrant in any meaningful sense. It's a meaningful document (contrary to what you might read; it's not something one can just print on a laser printer and called it a day), but the "administrative" changes the meaning dramatically.

It seems like there were plenty of errors all around, in this situation, both on the judge's side and on ICE's side. However, I can't imagine any of those rose to the level of criminal behavior.

Sneaking a suspect out a back door while you stall the police is textbook, classic obstruction, but that changes quite a bit when it's a government employee operating within their scope of duty. Even if they make a mistake.

Schools don't want students scared to be there. Courtrooms want to count on cases not being settled by default because people are scared to show up. There is a valid, lawful reason for not permitting ICE to disrupt their government functions. That's doubly true when you can't count on ICE following the law and might ship someone off to El Salvador.

Asking an LLM, whether or not the judge broke laws is ambiguous. It is unambiguous that they showed poor judgment, and there should probably be consequences. However, what's not ambiguous is that the consequences should be through judicial oversight mechanisms, and not the FBI arresting the judge.

As a footnote, a judge not being able to rely on ICE following lawful orders significantly strengthens the government interest argument.

Then you have issue with reading comprehension. The whole point of the discussion was this:

"The time to do this was about a decade ago. Apps, open formats, open USB protocols, open wifi / bluetooth protocols, and semi-open firmware (with a few proprietary blobs for color processing, likely) would have led things down a very different trajectory."

And the rest of your posts also misquote what I said and, ironically, just as often, what you said. There are also minor technical errors: diffraction limits are basic physics. It's a simple relationship between (a) the radius of the circle of confusion (in units of angle); (2) the frequency of light (in linear units, typically nanometers); and (3) the radius of the aperture (in linear units, typically mm). There is no voodoo with "sensor size, pixel pitch, and the lens optics." Most of your post is taking statements like a basic rule-of-thumb of what you need for decent photos and exaggerating to statements like "diffraction doesn't affect sharpness." Of course it's easy to beat up a statement if you misquote it. That's called a strawman.

So I think I'm done here. Give me your downvote, and I'll argue somewhere else.