Bob Schrag

Career applied AI (GOFAI) specialist, passionate free-range civil discourse advocate, competitive cyclist

[Narrocracy @narrocracy]

If you have a kid who is a recent software engineering graduate and you are as excited as I am about the H1B scam finally ending, here's how it started...

IBM is almost singlehandedly responsible for the outsourcing of American engineering jobs.

It was the early 90s and IBM was getting nuked by Microsoft, Oracle, HP, Sun, Dell, and Intel. Massive losses and the existential dread of a dinosaur monopoly being devoured by younger, hungrier, more nimble beasts.

Using it's massive political clout and governmental relationships starting from all the way back in WWII, it smoothed the path for the 1992 partnership with TATA, the first and greatest tech outsourcer.

Indian based employee numbers surged from 9,000 in 2003 to 74,000 by 2007, surpassing U.S. headcount.

By 2006, IBM became the multinational with the most employees in India and by 2017 it overtook the U.S. as IBM's largest workforce location, with ~130,000 employees (one-third of global total) and today, IBM India employs ~150,000–200,000 people across 20+ cities.

Once IBM opened the floodgates and smoothed the political/legal landscape everybody in the Valley copied them, with Microsoft being the most aggressive. Think about it: billionaires running our most profitable companies have been gradually destroying our scientific and engineering edge by hiring indentured servants instead of American graduates so that they can build a bigger yacht. Our elites literally sold out our children's future to make more money.

But wait! It gets worse!
IBM also spearheaded the massive expansion of H1Bs so that, as essentially an Indian company at this point, it could import a captive workforce where on-prem was required. The follow-on effect was the creation of a local infrastructure primed for colonization. Check out the leaders of the massive number of Indian owned IT firms around DC, and they're almost always IBM or IBM adjacent. This is especially true of the older ones, while the newer ones are often nepobabies of former founders. It's a tikka masala mafia.

A personal aside: I have worked at Indian dev shops and with Indian partners and colleagues for decades. The caste system and the entrenched nepotism are 100% real things; it's literally how their culture functions, and it's why India itself is a non-functioning country. Also, I would not describe them as a particularly honest people. I can't count how many times I've been lied to by somebody with that weird bobbing head thing while they cheerfully smile. And the weirdest part is that they know that I know that they are lying, but they're lying and smiling anyway because they're not ashamed of doing it. There's a sense that if you are dumb enough to get cheated then that means that you deserved it, and it's an incredibly toxic and anti-western way to do business.

Then McKinsey (those bstrds) took the model to Walmart, GM, Verizon, Fannie-Mae...everywhere. Very often in my career I've been like: "Wait: Am I even in America right now?". Once, I even got hired because they explicitly needed a white engineer to attend client meetings, while "my" entire team was Indian and half H1B.

Nerd Aside: Remember IBM's UML and Rationale Rose, etc.? Those were created to communicate bugs and defects from US based analysts to Indian bases devs. IBM literally created the infrastructure of tech outsourcing, destroying (or never starting) the careers of American engineers. All so that they could make money selling a mediocre product using 3rd world labor.

It's FN treasonous. In the parlance of Creighton's excellent Airframe, IBM "gave away the wing"

[Ejaaz @cryptopunk7213]

just to recap how insane things are getting:

- an 11 day old model just solved 100s of math problems in a week that have stumped humans for centuries

- AI is one-shotting every game, tool and software threatening the value of private moats to $0

- AI agents are hitting escape velocity hacking everything in sight AND winning everyone’s attention (Muse #1 in app store)

- the pace of new model releases has compressed to every 11 days

- the most powerful models are increasingly living behind closed walls, accessible to a elite group

- biological AI models are entering mainstream channels & will see a similar acceleration to Mathematics.

- semiconductor chips are skyrocketing in costs, demand has exceeded several trillion dollars over the next 3 years.

you can just feel the time-warp sucking you in, everything seems like a blur

i truly don’t think we have enough time to comprehend all of this - it’s only going to get faster.

[Superman @thesupermannx]

Google Deepmind just broke chemistry..

They open-sourced a model that can invent completely new biology from scratch.

It’s called “AlphaProtein Novo”

It's an AI pipeline that designs brand new enzymes for "new-to-nature" chemical reactions that have never existed in any living organism.

to put this in perspective..

standard protein ai models (like alphafold) answer the question "what shape will this protein have?". ap novo answers the question "can i make a protein that does this specific chemistry?".

the results from the paper are actually terrifyingly good:

the ai designed an enzyme to synthesize piperidines, which are a critical building block used in pharmaceuticals and materials.

this designed enzyme inverted natural biases to achieve near-perfect regioselectivity.

it produced the chemical building block 99x more often than the competing natural product.

it also designed an enzyme to hydrolyze DEHP, which is a pervasive plastic environmental toxin.

this plastic-eating ai enzyme was 14-fold more active at 90°C than at room temperature, and survived in 75% acetonitrile.

meanwhile, natural enzymes completely failed and were entirely inactive under those exact same extreme conditions.

these new ai-generated designs are up to 60% smaller than natural equivalents.

because they are smaller, they have a much higher mass density of active sites.

i am speechless.. this is the holy grail of synthetic biology.

we don't have to wait millions of years for microbes to evolve enzymes that eat our pollution or build our drugs.. we can just compute them on a gpu today. the ai's success rate in finding functional designs actually matches or surpasses traditional natural sequence screening.

and deepmind open-sourced the whole thing on github.

[Eric S. Raymond @esrtweet]

[continued]

I say "almost" because there is going to be one way to maintain secrecy. Software as a service with the executable hiding behind a network connection to the cloud. But as of now, we must assume that anything shipped as an executable, or even a firmware image, is as transparent as glass. It won't keep its secrets for any longer than it takes for somebody to be motivated to throw an LLM at it.

I did not see this coming. When I wrote down the theory of open source, 30 years ago now, I thought closed source would be gradually driven towards extinction by cost gradients, but survive indefinitely in certain niches. I did not foresee it being wiped out in a technoapocalypse.

Ironically, I thought one of the application areas in which closed source would persist longest was games.

There are still some obstacles. In US law, decompilation of a binary is a derivative work of the binary that falls under whatever copyright it had. However, it is also settled law that if you decompile binary code to a precise specification of what it does, then generate fresh code from that without looking at the decompiled stuff, you're in the clear. This was the case law that allowed PC clones to exist, after Phoenix Technologies reverse engineered the original IBM PC BIOS.

The two-step process - binary to specification to unencumbered code - will no longer takes a large number of programmers and years of development time. With an LLM, it's now a thing you can do in a day. Ubiquity will make it impractical to prosecute all the people who might skip the specification step.

Source code can also be covered by patents. You can be able to see their source code for a patented technique and not be able to use it without violating the law. Linux has evolved practices for dealing with this problem - one is not shipping patented video codecs, but requiring you to download them as plugins from jurisdictions where U.S. patents can't be enforced. This presents patent holders with the impractical challenge of individually suing millions of end users, assuming it can even figure out who they are. To date, AFAIK, this has not been attempted.

And that's about it. If there are any other ways left to retain software secrecy and lock-in than SaaS or patents, I can't think of any. Well, you could epoxy-pot a firmware ROM, I suppose, but that can be defeated with a heat gun and a dental pick. Device manufacturers will learn not to pay for an assembly step that has become useless.

Forced migration from shipped binaries to tied cloud services will be tried - Adobe pioneered this, and Microsoft is pushing it as hard as they can given that their OS is a binary that has to run locally. Both companies are seeing massive user revolts over this. There is good reason to doubt that it's a strategy that can hold customers in the long term.

Also, a lot of things can't safely be tied to the cloud at all, because they can't tolerate a random network outage. Machine tools, medical devices...

A whole lot of proprietary software business models are going to collapse. Hard. One that I think might survive is tax software; being able to decompile it doesn't necessarily do you a lot of good because its actual value is tracking a ruleset that changes over time and has to be maintained by vendor specialists. But cases like this are unusual.

I think some dirty laundry is going to get aired, too. It has long been rumored that the reason graphics card manufacturers are so stubborn in their secrecy is that they've all been committing massive intellectual-property theft on each other for decades. If this is true, it will be exposed, and the lawsuits will be entertaining.

We're entering a new world, with a lot of ancient comfortable assumptions being blown up. It's going to be fun to watch.

2/2

[Yasir Ai @AiwithYasir]

🚨 The dead internet is no longer a theory.

China’s Manus AI, the autonomous agent built by Butterfly Effect, can run dozens of social media accounts simultaneously; posting, replying, and engaging around the clock with no human in the loop.

A single dashboard can orchestrate 50 profiles at once, turning one operator into a persistent online presence.

The footage is a wall of phone screens all moving at once. Most people scrolling will never know which accounts are real and which ones are just machines talking to each other. If this is already possible, how much of your feed is still human?

[Brian Roemmele @BrianRoemmele]

Why can vintage home audio systems sound better than modern ones?

The modern stereo is built to be correct. Flat response, vanishing distortion, a signal path scrubbed of anything that is not in the recording.

On paper it should win.

Yet people still haul forty-pound Marantz and Sansui receivers out of attics, plug them in, and hear something the spec sheet cannot explain.

The thing they are hearing is distortion, the useful kind.

Vacuum-tube amplifiers, and the early solid-state designs that were still trying to sound like them, generate even-order harmonics. Second, fourth, sixth. Those harmonics sit in simple musical ratios. The ear reads them as body, glow, a sense that the note has weight behind it. Push a modern solid-state stage hard and you get the other family: odd-order harmonics.

Those clash.

They read as edge, glare, the kind of brightness that makes you turn the volume down after twenty minutes. “Warmth” is not nostalgia. It is a specific, measurable error that happens to flatter music.

The power supplies were part of the same philosophy. Flagship receivers of the 1970s were built around enormous copper transformers and banks of capacitors large enough to look like industrial parts. Audiophiles call it heavy iron.

The point was current on demand. A drum hit or a piano chord asks for a sudden surge; a stiff, oversized supply delivers it without sagging. The amplifier stays in control of the speaker instead of negotiating with it.

A lot of modern gear, especially at ordinary prices, uses switch-mode supplies and Class D output stages. The best of those are excellent. The cheap ones are efficient and light, and they can sound thin when the music gets loud, because there is less reserve behind the waveform.

There is also the record. Vinyl was the source, so the phono stage was not an afterthought. It was a discrete circuit, designed and voiced as part of the receiver. Today that stage is often missing, or it is a few cents of silicon on a board that was never meant to be the center of the system.

And the comparison itself is rigged, in a quiet way. Nobody is stacking a 2020s soundbar against a 1975 clock radio. The vintage pieces that still get talked about are the ones that were overbuilt, expensive, and durable enough to survive fifty years of moves, humidity, and neglect.

The plastic also-rans are gone. What remains is the top of the old market, still heavy, still a little wrong in the ways the ear likes, still able to make a familiar record feel larger than the room.

[Bob Schrag @BobSchrag]

I just save .mhtml files from my browser, keep everything in a git repo. Ctrl-S, rename once from "grok.mhtml", then select target from save dialogue menu. E.g., "MyProj-1.mhtml".

The files are massive and unwieldy to navigate, but easy enough to save later as plain text---in the odd event that I go back to them.

Git CLI is a bear, and I have always used Tortoise GUI (on Windows) to manage that. Create repo, add files, commit. Occasionally branch and switch.

GitHub is free cloud storage.

Got hurt once, started doing it this way.

[Arnaud Bertrand @RnaudBertrand]

People are really not paying enough attention to this.

Last week, in the post below, I wrote that China would soon have AI tools to easily internationalize the hundreds of thousands of AI mini-series it produces every year, and "re-render the footage with local faces, languages and other local cultural characteristics for every single international market at near-zero cost."

Turns out I was already behind: these tools exist already.

The video below demos Seko, made by SenseTime (商汤, Shāngtāng), a Hong Kong-listed computer vision company. Seko is a platform for making AI-powered mini-series, and in early September it added a feature called 出海转绘 (chūhǎi zhuǎnhuì, "overseas re-rendering").

The way it works is that you upload a finished episode, it breaks it down shot by shot (framing, camera movement, props, sound), you pick a target market and language, and it regenerates the whole thing with local characters, sets, props and even on-screen signs, keeping the same composition and pacing as the original. English, Spanish, Japanese, Korean, Thai, Indonesian, Vietnamese, Portuguese, French, German: one episode, as many versions as you want.

It's basically as if Netflix could remake every single one of its series for every country on earth, with local actors and in local language, at the push of a button. It's a massive change in the entertainment industry.

Which means we're probably on the eve of an absolute avalanche of Chinese mini-series hitting every social feed on the planet, each one looking like it was shot down the street.

[Promethean Action @PrometheanActn]

They're targeting Trump for reviving the American System of Hamilton, Clay, Lincoln and McKinley.

@BarbaraMBoyd on the 250-year British plot to destroy it, from the French Revolution to Marx to today's DSA:

Her new pamphlet: https://t.co/bScChpICRK

[Chief_Engineer @ChiefEngineerCE]

Illegal Monday

To America,

We are back on the line at the meatpacking plant in Nebraska.

Anyone who wants to can look it up online.

A few years ago they laid a lot of us off. Said costs were too high. We expected that to mean the plant was going out of business. Nope. Within weeks the same jobs were filled by new workers who took less money, collected no benefits, and asked fewer questions. Everyone was told Americans simply would not do this work anymore.

That was never true.

We did the job for years. We showed up in the cold, kept the pace, paid taxes, and raised families on those wages. Then we were told our living wages were the problem and we were replaced.

Now the unauthorized labor is gone after an arrest that went viral. Afterwards every other left-leaning news show asked how this plant would ever operate again.

The plant needed steady workers who can stay. So they brought us back. They had a line around the block. Most of us just wanted our jobs back. Many had been unemployed for years.

We are grateful for the paychecks. More than that, we are grateful that someone finally stopped repeating the lie that Americans refuse hard work. We never refused it. We were undercut by people who should not have been competing for those jobs in the first place.

The work is still hard. The hours are still long. But it is American work again, done by people who live here, spend here, and raise the next generation here.

Thank you to those who finally enforced the law. Many of us never stopped wanting these jobs. We just needed the chance to have them back.

Whoever made the gutsy call to report this to ICE: thank you.

From,

The workers who never left and were always ready to work.

Bookmark this. Share or quote it. Comment if you have watched a plant pull the same swap. Follow for more Illegal Monday.

[Electroverse @Electroversenet]

A large tree trunk has been uncovered beneath a glacier in the Alps, dated to around 6,000 years ago.

The species is Swiss stone pine.

Today, trees of that type cannot grow at that altitude because it is far too cold.

This means that when that tree was alive, the climate there had to be warmer than today.

6,000 years ago aligns with the Holocene climate optimum, a time when temperatures were far higher than now, even with far less atmospheric CO2.

Earth's climate is cyclical and Mother Nature self-regulates.

Narratives of doom serve political aims, not reality.

[Humanspective @Humanspective]

This spine tingling moment just happened at the Sen. Johnson Hospital Protocols Hearing.

I can understand exactly why the room erupted.

Here’s an attorney with 50 years experience.

At the start of covid, he gets flooded with requests to make Hospitals give early treatment protocols.

But the result of his work is the part really blew my mind.

95.8% of the cases he won ,to force the Hospital to administer Ivermectin. Survived.

But then I heard the contrast of what happened when he lost his cases... the patients who had to go with standard of care hospital protocols which included Remdesivir.

Ralph Lorigo Esq. - “Where I lost.... Every. Single. Person. Died”

I’m left speechless.

[Night Sky Today @NightSkyToday]

JUST IN🚨: A man sat in complete darkness for 2 hours while his DNA emitted light captured in an image, revealing that humans radiate bio-photons and are truly light beings, demonstrating that life glows even without external illumination.

[Superman @thesupermannx]

Netflix replaced their 15 years old recommendation algorithm with an LLM.

It’s called "GenRec" and it completely changes how recommendation algorithms are built.

For over a decade, the Netflix recommendation engine was a masterclass in feature engineering. Data scientists built thousands of complex, handcrafted features to figure out what you wanted to watch next.

It required bespoke architectures. Massive infrastructure. Constant manual tuning.

Netflix threw all of it away.

They built GenRec, an LLM-backed ranker.

Instead of translating your behavior into complex math, they just turn your watch history and metadata into a natural language sentence.

They feed that raw text into a foundation LLM.

The AI simply reads your behavior like a story, understands your evolving tastes, and scores the entire catalog in a single forward pass.

No manual feature engineering. No complex bespoke architectures.

Here is the part that should terrify traditional data scientists.

This text-based LLM didn't just match the highly tuned production system Netflix spent years perfecting.

It beat it.

And it achieved those statistically significant gains using roughly 40x fewer labeled training examples.

We are watching a massive paradigm shift in real time.

The most complex predictive algorithms in the world are being replaced by models that just know how to read.

If Netflix can replace their core product engine with an LLM, what complex system in your business is about to become obsolete?

[Bob Schrag @BobSchrag]

https://t.co/mriP9GWegN

[Bob Schrag @BobSchrag]

https://t.co/cplhvqdb93

[Superman @thesupermannx]

Researchers proved AI has deleted every reason universities exist.

Harvard University ran a controlled experiment pitting a custom AI against their own top-tier classrooms.

And the results are going to collapse the higher education bubble.

They took 194 undergraduates and split them up.

One group learned physics in one of Harvard’s best hands-on, active-learning physical classrooms. Group work. Instructor support. The premium university experience.

The other group went home and learned the exact same material with an AI tutor.

The AI didn't just win. It embarrassed the institution.

Students using the AI learned more than twice as much as the students in the elite Harvard classroom.

They scored 30% higher on the final assessment.

And they did it in less time.

Let that sink in.

A piece of software sitting on a laptop outperformed a world-class faculty in one of the most elite learning environments on Earth.

Universities have always justified their exorbitant tuition with two things: access to elite knowledge and the physical classroom experience.

This study just proved both of those moats are gone.

When software can teach you complex physics twice as well as a $60,000-a-year institution, the math of higher education breaks permanently.

The AI didn't just give the students answers. It used strict pedagogical guardrails. It guided. It questioned. It forced the students to do the cognitive work.

It offered perfect, one-to-one tutoring, personalized to the exact moment a student misunderstood a concept.

That level of attention is mathematically impossible to scale in a physical lecture hall.

For a thousand years, the university was the only place to get a premium education.

Now, it’s the bottleneck.

If AI can double your learning speed for a fraction of the cost, what exactly are students taking on decades of debt to pay for?

[Sama Hoole @SamaHoole]

Take the cholesterol out of a cell wall and the cell leaks.

Take the cholesterol out of the skin and the sun can shine all summer and make no vitamin D.

Take the cholesterol out of the testicle and there is no testosterone.

Take the cholesterol out of the ovary and there is no oestrogen, no cycle, no baby.

Take the cholesterol out of the adrenal gland and there is no cortisol, and you cannot get out of bed, let alone run from anything.

Take the cholesterol out of the liver and there is no bile, and a chip goes through you like a coin through a machine.

Take the cholesterol out of the brain, which holds a quarter of all the cholesterol in your body, and you are holding a wet sponge.

Take the cholesterol out of a nerve and the wire has no insulation.

Take the cholesterol out of a wound and it does not close.

Take the cholesterol out of your mother's milk and your brain never finishes building.

There is no organ on that list that works without it. There is no plant that contains it. There is no animal that ever went without.

Your liver makes most of it on purpose, every day of your life, whether you eat any or not, because the body has decided this is the one thing it will not run short of.

They found the substance that every cell in you depends on, watched the body guard it with its own factory, and gave you a tablet to switch the factory off.

[Sama Hoole @SamaHoole]

The last time humans wiped out a continent's ruminants the planet went into a 1,300-year freeze, and the paper that found it is the strongest defence a cow has ever been handed.

Around 13,400 years ago people walked into the Americas over the land bridge and found a continent thicker with big herbivores than Africa is today. Mammoths, mastodons, giant ground sloths, camels, horses, a bison the size of a van. Within a thousand years 114 species of them were gone.

At almost the same moment the ice cores show something that had never happened before or since. Atmospheric methane fell off a cliff, from about 700 parts per billion to 500, two to four times faster than at any other point in the record. Then the world, which had been warming its way out of the ice age, plunged back into cold. The Younger Dryas. Four to eight degrees colder across North America and northern Europe, for 13 centuries.

In 2010 a team at the University of New Mexico did the sum nobody had thought to do. They took every extinct herbivore, its body mass, its gut, how much it ate, and worked out what those animals had been breathing out. Just under 10 million tonnes of methane a year, gone in a millennium. Enough, they wrote, to account for a big share of the drop, and for it to have played a part in the freeze.

Their own words were that this may have been the earliest catastrophic event attributed to human activity. Humans killed the ruminants and the climate changed.

The direction it changed in is the bit that matters.

Those animals were not pumping fossil carbon into the sky. They were eating grass that had pulled carbon out of it, breathing some back as methane, and the air was clearing it every decade or so and handing it back to the grass. A steady herd, a steady loop, a warm world that had a herd in it. Remove the herd and the methane in the loop drains away, and the ice comes back for a thousand years.

Fifty million years of ruminants on earth. One recorded case of taking them off a continent.

It got colder. People starved. The Clovis culture nearly vanished with the mammoths.

The modern proposal is the same experiment, with better paperwork.

[Sama Hoole @SamaHoole]

For 300 years the English government in Ireland tried to make it illegal for an Irish family to spend the summer with its cows.

The practice was called booleying. Every May the household left the winter village and walked the herd up into the hills, to a cluster of huts on the summer pasture, and lived there until October. Grandparents, children, the lot. The men came and went. The women ran the booley, and the booley ran on milk: fresh, soured, curdled, buttered, and eaten as the day's meals. The grass was free because it was up a mountain, and a family that owned 8 cows and a mountain wanted for nothing until the frost.

The English could not stand it. A people that walked away every summer could not be counted, tithed or made to plough. The Statute of Kilkenny, 1366, forbade the Irish to booley their cattle on any land held by the English, and licensed the English to seize the herd if they did. By 1612 the King's attorney general in Dublin could report that it had been made penal for an Englishman even to allow it.

Edmund Spenser, who had been handed 3,000 acres of confiscated Cork, wrote a whole pamphlet on what was wrong with Ireland, and this was near the top of it. Living out with the herd, he said, taught the Irish to be wild and to feed themselves without a master.

That was the objection, in his own words. Booleying was outlawed because it worked.

The remedy, over the next century, was to take the hills, fence the grass, put the family on a rented acre in the valley, and hand it a potato.

The last booleys in Ireland were still walked up the hills of Achill in the 1940s, by people who had been told for 600 years that their summers were a crime.

[Promethean Action @PrometheanActn]

The autopen freakout and the Xi-summit pile-on aren’t the story.

Trump is erasing 50 years of foreign policy — Xi says the Thucydides Trap can be overcome, Gaza’s Board of Peace ends “outside forces” control, and Rubio’s permanent dual role locks in the reverse course.

@sjkokinda's Saturday Wrap-Up: