Fast, cheap AI-assisted decompilation of binary code is here. Which means code secrecy is dead.
Decompilers in themselves are not a new technology. Security researchers have employed them for years to analyze compiled malware. There’s been some limited use by others, notably by hobbyists decompiling abandonware games. But there were a couple of issues that prevented this from becoming common practice.
One is simply that running decompilers was difficult. It wasn’t as simple as feed in binary, get out source; it needed a person with specialist skills prepared to do spelunking through wildernesses of machine code and object formats. The other problem was that decompilation didn’t give you anything like the explanatory comments that had been in the original code, so you could easily wind up with code that you could read without being able to understand or modify it.
Now large language models are busily smashing both of those barriers flat. They’re better at the kind of detail analysis required to run the human side of a decompilation than humans are. More importantly, in the process of decompiling code, they rather automatically build a global model of how it works that can easily be expressed by high quality comments in the extracted code. All you have to do, basically, is ask for the comments.
I’m going to reinforce that latter point because it may not be obvious how good LLMs are at this, and how much better they’re going to get. When they decompile code and comment it for you, they’re not just working from that one piece of code you have put in front of them — they’ll have in their training set hundreds, possibly thousands of pieces of code similar to it and with comments. This will give them superhuman levels of insight not just into what it does at the microlevel, but what it means to the humans who wrote it, and what technical assumptions it’s embodying.
Compilation no longer guards your secrets. Or, to put it more precisely the expected time span in which you can still count on it to obscure them is measured in months. Possibly weeks.
What does this mean?
It means you’re in an open-source world now. All it’s going to take for anybody to bust your proprietary IP open is care enough to spend tokens on the analysis.
You will maximize your chances of survival as a software business if you get out ahead of this rather than trying to fight it.
This isn’t exactly the way I expected open source to win. But, you know, I’ll take it. Good enough.
ESR, The social media site formerly known as Twitter, 2026-04-08.
July 12, 2026
QotD: “This isn’t exactly the way I expected open source to win”
July 11, 2026
British censorship laws do not apply outside the UK’s jurisdiction
On Substack Notes, Lorenzo Warby links to a fascinating discussion about the ongoing struggle between the UK government’s Ofcom and the US-based 4chan and their legal representatives, saying “The totalitarian wannabes currently running the UK do not apparently grasp that the American Revolution and War of Independence was a thing. Also, being totalitarian wannabes, they have no sense of humour.”
A UK cabinet minister, Rt. Hon. Liz Kendall MP, Secretary of State for the Department of Science, Innovation and Technology (“DSIT”), discussed the infamous “hamster e-mail” I sent on behalf of my client 4chan to the UK’s Internet censor, on national radio today in the UK. […] My father always told me, when I was growing up, “when a cabinet minister holding the technology policy brief for a G7 Member State is talking about your e-mailed jokes to an audience of millions on national broadcast media, that is the right time to explain the joke, especially if the cabinet minister didn’t get the joke”.
That explanation follows.
The backstory β Hamster #1
The hamster joke has a bit of a history to it. Ofcom, the UK’s Internet censor, first made contact with my American client 4chan in June of 2025 in its attempt to impose British censorship law on that website. I was subsequently retained as defense counsel, pro bono.
Ofcom then “provisionally fined” 4chan on August 16th, 2025 for refusing to obey the UK’s censorship regime. We were invited to make representations to the regulator following that provisional fine decision.
We did two things in response to that. The most newsworthy response was to file a lawsuit against the regulator in the DDC. Before that, however, we explained our position to Ofcom in writing and gave them an opportunity to walk away:
To wit, Ofcom’s fine notices were not properly served and were not enforceable in the United States. Note that we also gave Ofcom fair notice that while this might have been their first attempt to enforce their censorship orders in America, this was not our first rodeo when it came to successfully refusing such orders.
No quantity of officious and haughty foreign demand letters will change our stance. The UK could even pass a bill of attainder β historically Parliament’s most extreme and powerful legislative weapon β against my client, for all I care. My client’s right to operate its service lawfully in the United States is protected by the First Amendment. There is no law Parliament could enact that would change that fact.
I am very familiar with how this movie ends, and it does not end with 4chan paying Ofcom’s fine.
It may end with the UK’s censors getting a blocking order that it serves on its own ISPs; that would be the UK visibly censoring its own people, rather than censoring my client, and doing so ineffectively, at that, as ISP blocks can be circumvented with a VPN. That is a consequence my client is prepared to accept.
England might have the Online Safety Act, but the United States has the U.S. Constitution. These rulesets do not override each other; they are, rather, mutually exclusive. In America’s domain, the Online Safety Act essentially doesn’t exist. It has about as much legal force as a pile of shredded paper one might use to line a hamster’s cage.
Peace was always an option here, but that would have required the UK to abandon the fiction that its rules override the U.S. Constitution on U.S. soil, which we are not prepared to accept.
My clients did not start this fight, but by golly we do intend to finish it.
My client sued Ofcom two weeks later.
There’s much more, so do read the whole thing.
QotD: Could airpower have broken the trench stalemate on the western front in WW1?
What about, instead of going through the trench lines, we went over them?
There are two directions to take airpower here: tactical and strategic. One wasn’t ready then (but would be by WWII), the other still hasn’t managed to accomplish its stated objectives yet, but continues to over-promise and under-deliver results.
Let’s deal with tactical airpower first. The first function aircraft were put to in WWI was reconnaissance. In 1914, that might mean locating the enemy in a fast-moving battlefield, but as soon as the trench stalemate set in, reconnaissance mostly meant identifying enemy buildups along the line and β still more importantly β serving as spotters for artillery. It wasn’t a huge cognitive leap to go from having aircraft which identified targets for the artillery to thinking that the aircraft could be the artillery. But as with tanks, the technical limitations of the platforms in use meant that actually meaningful close air support was still two decades away when the war ended. The rapid development of aircraft in these early days means that there is a truly bewildering array of aircraft designs in use during the war, but the Farman F.50 is a good sample for what the most advanced bombers in common use looked like towards the war’s end. It carried a maximum of eight 44kg bombs (352kg) under the wings, which were dropped unguided. With a maximum speed of less than 100mph and a service ceiling under 5000m, it was also an extremely vulnerable platform: fragile, slow and with a relatively low flight ceiling. The French mainly used bombers at night for this reason.
But how much airpower does it take to really move a division out of position? In 1944, at the start of Operation Cobra as part of the Normandy breakout, it was necessary for US forces to move the powerful armored division Panzer Lehr out of its prepared positions outside of St. Lo. Over the course of an hour and a half, the U.S. Eighth Air Force hit Panzer Lehr with approximately three thousand aircraft, including 1,800 heavy bombers (each of which might have had bomb-loads of c. 2-3,500kg; the attack would have been the equivalent of about 13,000 Farman F.50s (of which only a hundred or so were built!)). By this point, even medium bombers carried bomb loads in the thousands of pounds, like the B-25 Mitchell medium bomber, with a bomb load of 3000lbs (1360kg). This was followed by a hurricane artillery barrage! Despite this almost absurdly awesome amount of firepower (which, to be clear, inflicted tremendous damage; by the end of Operation Cobra, Panzer Lehr β the heaviest and most powerful Panzer division in the west β had effectively ceased to exist), Panzer Lehr, badly weakened was still very capable of resisting and had to be pushed out of position by ground attack over the next three days.
Needless to say, nothing on offer in 1918 or for a decade or more after, was prepared to offer that kind of offensive potential from the air. That kind of assault would have required many thousands of aircraft with capabilities far exceeding what even the best late-war WWI bombers could do. Once again, while close air support doctrine was developed with one eye on the trench stalemate and the role airpower could play in facilitating a breakthrough and restoring maneuver (either by blasting the breakthrough or β as in Soviet Deep Battle doctrine β engaging enemy rear echelon units to bog down reinforcements). But the technology wasn’t anywhere near the decisive point by 1918. Instead, the most important thing aircraft could do was spot for the artillery, which is mostly what aircraft continued to do, even in late 1918.
But that’s tactical bombing against military targets. What about strategic bombing against civilian targets?
The first efforts at strategic bombing were made in WWI, though once again the technology wasn’t ready. The range for fixed-wing aircraft was still very limited; the aforementioned Farman F.50 had a range of only 420km, nowhere near enough to really bring entire countries under the threat of bombing. Dirigibles β zeppelins β could manage much longer ranges and the Germans did attempt to bomb British cities with them starting in 1915. The problem was that once aircraft powerful enough to climb to the zeppelin’s altitude were developed, the slow and fragile zeppelins were sitting ducks: lighter than air airships could hardly be armored, after all. Moreover, the bomb loads of zeppelins had always been far too low to make effective strategic bombing possible beyond the initial shock of it.
What no one could have known in WWI was not merely that the technology for effective conventional strategic bombing wasn’t ready, but that it would probably never be ready. Interwar air-power theorists, seeing the potential of strategic airpower to bypass the trench stalemate by flying over it began to try to work out how this would be done. Giulio Douhet (1869-1930) argued that future wars would be fought and won in the air, with fleets of bombers using high explosives and chemical weapons to massacre enemy civilian centers, until civilians forced their governments to surrender. Douhet was not alone; his vision of airpower was shared, for instance, by the “father of the RAF”, Hugh Trenchard (1873-1956).
This concept, “morale bombing” as it is sometimes called, probably deserves its own post discussing its failures. But in brief, the concept was tested, with far larger amounts of bombs than Douhet or any other interwar theorist could have ever dreamed of, during WWII. The argument by air theorists that high altitude bombers could not be stopped was proved false when the British did exactly this, stopping German bombers over Britain in 1940. Moreover, terror bombing against civilian targets in Britain didn’t lead to surrender, but hardened resolve. Likewise, “morale” bombing against German targets by the allies didn’t lead to surrender, but hardened resolve. Later efforts to demoralize the North Vietnamese through a American bombing campaign in the Vietnam War didn’t lead to surrender, but hardened resolve. More recent efforts to demoralize or destroy terrorists and the Taliban through the use of airpower hasn’t lead to surrender, but rather hardened resolve. Likewise, efforts by the Syrian Regime to defeat various opposition groups in Syria through the use of chemical weapon-based terror bombing didn’t lead to surrender (siege-and-starve tactics did), but hardened resolve.
It turns out the fundamental premise of the entire idea of morale bombing β that being bombed will make people want to stop fighting β was flawed. Morale bombing has been, depending on how hard you squint at the US air campaign over Japan in WWII (including the use of nuclear weapons) successful either once (out of many attempts) or never. In most cases, the sustained bombing of civilian centers has been shown to increase a population’s willingness to resist, making the strategy worse than useless.
The case for strategic bombing against industrial targets is marginally better, but only marginally. While airpower advocates, particularly in the United States promised throughout WWII that bombing campaigns against German industry could lead to the collapse of the German war machine, in the end many historians posit that the real achievement of the campaign was to lure the Luftwaffe into the air where it could be destroyed, thus denying the German army of air cover and close air support, particularly on the Eastern Front. Some diminution of German industrial capabilities was accomplished (though it is not clear that this ever approached the vast resources poured into producing the large numbers of extremely expensive bombers used to do it, though the allies had such an industrial advantage over Germany, forcing the Germans to fight in expensive ways in the sky was a winning trade anyway), but the collapse of German industry never happened. As Richard Overy notes, German industrial output continued to rise during strategic bombing and only began to fall as a result of the loss of territory on the ground. Needless to say, “strategic bombing can sucker the enemy into wasting their close air support” was not the result that airpower advocates had promised, nor could it have broken the stalemate.
I don’t want to oversimplify the continued debate over the efficacy of strategic airpower here too much so let’s just say that the jury is still very much out as to if strategic airpower works even with modern technology; it certainly wouldn’t have worked with WWI era technology.
Bret Devereaux, “Collections: No Man’s Land, Part II: Breaking the Stalemate”, A Collection of Unmitigated Pedantry, 2021-09-24.
July 7, 2026
Clankers in higher education
On his Substack, David Friedman discusses the impact of AI on university exams:
Professor Roberto Serrano, who is the Harrison S. Kravis University Professor of Economics at Brown University, has detected a massive fraud in one of the classes he teaches, ECON 1170, an advanced undergraduate course in mathematical economics. He has conclusive evidence that at least 50 students cheated on the March midterm exam, making it the biggest known scandal at Brown and in the entire Ivy League
…
“Academic integrity is a value worth defending. The faculty cannot be left on its own in a battle that is decisive if we want to preserve the future of higher education,” explains the 61-year-old professor … (El Pais)
It was a closed book take-home exam; the problems were designed to test the student’s ability but proved doable by an AI. After Serrano changed the final from take-home to in-person about half the students who had gotten perfect scores on the midterm chose not to take the final.
The existence of AI, like the earlier problem of students buying papers online, reduces the ability of teachers to test their students but does not eliminate it, is inconvenient but not catastrophic. It makes some kinds of testing more difficult but not impossible; Serrano could have asked students whose midterms were suspiciously good to explain some of their answers and failed any obviously unable to do so. That would have been additional work for him and, judging by the article, not a policy Brown would have endorsed. Unwilling or unable to do that that he can base his future grading on work done in-person and adequately monitored.
[…]
It is not immediately obvious what is wrong with using AI on a test. If the purpose of the test is to generate information for potential future employers, why should they want the student tested without a tool that, if they hire him, he will have? A basketball coach does not evaluate potential team members by how well they can play with one hand tied behind their back.
Arguably the skills the employer wants tested are those that an AI cannot replace and it was up to Professor Serrano to find ways of testing for them. His take-home midterm, taken without the assistance of AI, might have provided information for him and his students about how far they had come along a path that would eventually produce skills an AI could not substitute for but not information for a future employer about the skills of the students taking the exam.
June 30, 2026
Leading the grassroots revolt against AI … Homer Simpson
Ted Gioia posted this a couple of days back, but if you haven’t read it it’ll still be new to you:
Last November I suggested that 2026 would witness a tech backlash of unprecedented intensity. And it’s now happening with a vengeance. Silicon Valley is getting skewered everywhere, and to a degree inconceivable just a short while ago.
Just yesterday, The Economist finally grasped how rapidly tech antipathy is mounting β and made AI backlash its cover story.
The latest survey numbers are devastating. Every demographic group is now opposed to AIβespecially young people, previously the most enthusiastic supporters of new tech.
[…]
Not every pushback to encroaching tech is quite so gentle.
Consider the case of “Mr. Daniels,” a 25-year-old man from England. He knows that AI will rob every music file on the web for training β so he decided to poison the data.
How did he do it? According to Tuned Into Tech, it happens like this:
He took his entire music library of 2,000 records, stripped out the original vocals, and replaced every single one of them with the voice of Homer Simpson. Then he uploaded all of them to Soulseek. He didn’t change the metadata, the file names, the artist tags, the album information. They all stayed exactly the same.
A listener might not notice at first. Some of these songs have long intros, and those are unchanged. But as soon as the singing begins, Homer Simpson takes over. When AI tries to steal this for training, it gets fooledβand contaminates its own data set.
So somewhere deep in a training algorithm’s data set is the audio of Homer Simpson which the AI will assume sounds like [for example] Madonna, Rihanna, or maybe even Sean Paul. The model doesn’t know the difference. It just ingests the data and treats that like the truth.
And that is exactly what Mr. Daniels is hoping for.
He wants “to introduce noise, chaos” into the bots that are putting human musicians out of work.
“Mr. Daniels” is not an isolated example. Musician Benn Jordan has also been “poison-pilling” music files in hopes of disrupting AI.
In recent months, he has watched in horror as “tech companies started raising millions of venture capital dollars and scraping my music without my consent”. They now use his own work to generate “shittier music with it that is inadvertently associated with my name β and then attempting to resell that in the same economy in which I make money from my music”.
As a result, he has stopped releasing music. But he hasn’t walked away from the battle β instead Jordan has developed “a type of encoding that not only makes a music file more or less untrainable by generative AI companies, but actually has the ability to decrease the quality and efficiency of their entire data set”.
“Unethical generative AI companies have made artists feel incredibly powerless for quite some time now”, he adds, “but all of that is about to change”.
June 28, 2026
June 26, 2026
June 25, 2026
June 20, 2026
“Every system on display is an answer to a question the war in Ukraine asked out loud”
Most wars are not significant drivers of technological change and military innovation. The Franco-Prussian War, the Boer Wars, World War 1, and World War 2 are some of the exceptions where the fighting accelerated innovation and adoption of new and untested technologies that were proven or discarded on the battlefield. The Russo-Ukraine war has been going on long enough and requiring new and improved weapons to such a degree that modern arms shows clearly reflect at least some of the technological changes in response to the ongoing combat:

Thales RapidStriker SHORAD, I think. Oddly, what struck me about this image was how much it reminded me of very early WW1 armoured cars, both in general outline and in its being a quick reaction development to a current combat situation.
Photo from Eyes Only with Wes O’Donnell
I was thinking recently about the good ole pandemic days; ah, what a simpler time …
At the time, I was writing for military and cybersecurity magazines about whether NASA spacesuits can be hacked and hypersonic tomfoolery.
Six years ago, a defense expo like this was mostly about better armored boxes. Things like thicker protection, a nicer turret, an upgraded engine, a fire-control system with a new acronym.
The headline acts were tanks and infantry fighting vehicles, things that go very high and very fast, and the unspoken assumption underneath all of it was that war would look roughly like it always had, just with more cowbell.
Then 2022 happened.
Then Operation Spiderweb.
Then a year of Russian glide bombs and Ukrainian refinery strikes and FPV drones turning hundred-dollar quadcopters into tank-killers.
Then the Gulf woke up to Iranian missiles in March. And the entire defense industry got the same text message at the same time, written in other people’s blood.
You can read that message on the Eurosatory floor this year.
Almost every serious system on display is an answer to a question the war in Ukraine asked out loud:
How do I shoot from farther away so I don’t die?
How do I kill cheap drones without going bankrupt?
How do I send a robot instead of a soldier?
How do I keep my tank’s roof from becoming a Thermador pizza oven set to “broil?”
Back then, I also used to write listicles, like “Top Ten Gifts for Veterans!” In that tradition, I’ve put together a hand-picked list of ten weapon systems emerging this year at Eurosatory in Paris, and every one of them is really a story about how much war has changed since 2020.
Bill C-22 passes the Commons “as MPs raced for home for the summer”
Canadian Members of Parliament care more for their summer vacations than they do for the rights of Canadian citizens. While this isn’t really news, it’s just the latest proof that our elected representatives are … well, I was about to describe their moral failings in great detail, but that could get me arrested and jailed if-and-when the many authoritarian measures the Liberals want to enact become law. Instead, here’s Michael Geist‘s summary of the way Bill C-22, the Lawful Access Bill, got sent to the Senate on Thursday night:
Bill C-22, the lawful access bill, passed the House of Commons yesterday with the government invoking a single motion to approve several bills without further debate or individual votes as MPs raced for home for the summer. Bill C-22 will now head to the Senate, where it can expect a rougher ride when study begins in the fall. Rather than use the final days of the House session to answer the privacy, security, and oversight concerns raised by the Privacy Commissioner, academics, technology companies, and civil society groups, the government spent the time ensuring it would not have to, rushing the bill through committee, cutting off debate, and maligning critics with tactics that they once decried when in opposition.
The final days of Bill C-22 in the House marked a genuine abrogation of democratic norms. The government moved a motion to shut down the clause-by-clause study in the Standing Committee on Public Safety and National Security, preventing the committee from adjourning until the bill had been pushed through. That led to a session that stretched past midnight, as MPs were barred from introducing new amendments and were left to vote on amendment after amendment without any discussion, debate, or even public disclosure of their contents. By the end of the committee session, no one could have known the contents of the bill that MPs had duly approved and sent back to the House for final approval. As noted, once back in the House, there was no further debate, discussion or even a vote. Just a motion that said the deal was done.
If the process was troubling, the rhetoric was embarrassing. I wrote earlier this week about Public Safety Minister Gary Anandasangaree’s Vic Toews moment, as he said it was time for opposition parties to “choose” whether to stand with law enforcement and victims of crime (a refrain that sounded a lot like Toews’ 2012 comment to Liberal MP Francis Scarpaleggia, who is now the Speaker of the House, that he could “either stand with us or with the child pornographers”). Government House Leader Steven MacKinnon pushed that posture further on Thursday by dismissing the bill’s critics as wearing “tinfoil hats” engaged in “paranoia.” The charge fits a broader pattern in which this government treats independent privacy scrutiny as an obstacle rather than a safeguard, seen most clearly in the Bill C-36 approach to strip the Privacy Commissioner of authority over private-sector privacy law altogether.
The committee did approve some government amendments to the bill that improve aspects of the lawful access plan but they are still likely to leave companies, security experts, and privacy advocates concerned. For example, the maximum metadata retention period the government can impose drops from one year to six months, and a category of metadata can now be mandated only where the Minister is satisfied that the category and all of its elements are essential to investigations. That is better, but still not good enough as it is not tied to any actual evidence about why six months is needed and both the costs and risks associated with metadata retention, which is not a requirement in the U.S., are largely unchanged.
As The Reclamare explains, this bill is yet another likely irritant in US/Canadian affairs, as it will expose US citizens’ data to Canadian government oversight:
– A USA person creates/maintains a social media account — lets call it “XXX”
– Using its new C22 law, Canadian RCMP develops a “reasonable grounds to suspect” of “XXX” to a CDN investigation (a low investigative hunch standard under C-22).
– RCMP obtains a Canadian judicial authorization (an “Order”) and sends the Social Media company an International Production Request, which is not a USA warrant, not a Β§2703(d) order, and not routed through full MLAT (Mutual Legal Assistance Treaty) review.
– The social media company is bound by US law (SCA/ECPA), treats the request as a formal foreign inquiry.
– The social media company discloses limited metadata: summary of login IP ranges, account country setting, and other classification signals to prove USA origin
– This disclosure happens at Canada’s “reasonable suspicion” threshold, which is lower and less scrutinized than the US domestic requirement of “specific and articulable facts showing relevance and materiality” under 18 U.S.C. Β§ 2703(d) for the exact same type of data.
– The USA user’s metadata, which would normally enjoy stronger 4th Amendment derived judicial protections, if sought directly by US authorities, is handed to a foreign government on weaker foreign grounds, without the same level of US court filtering or notice that a purely domestic US request would trigger.
– The 4th Amendment protection is effectively diluted because the platform’s good faith compliance with the foreign lower bar creates a new, easier pathway around domestic US constitutional safeguards for accounts that platforms classify as American
Canada’s Liberal government continues to chip away at our “Charter of Rights”, under the guise of “Protecting Citizens” and we are moving towards authoritarianism
While I loathe to create friction, I also hope your Rights can help slow Canada’s devolvement
It impacts you too
June 19, 2026
June 16, 2026
Universal suffrage has its drawbacks
Democracy is a better system than many others that have been tried over the centuries, but it’s far from perfect. Giving everyone the vote sounds like a good idea: you have some small theoretical degree of influence over the people who run the country (note the “theoretical” here). Devon Eriksen points out one of the problems with universal suffrage today:
The problem with universal suffrage is that the more technologically advanced a civilization becomes, the smaller the fraction of people there are in it with the native intelligence to understand how it works.
When the majority of humanity was employed in whacking at the dirt with a pointed stick, and the height of technology was a slightly better pointed stick, anyone with a triple digit IQ could understand what was going on.
Now, we have things like stock markets, the internet, transportation infrastructure, and the Linux kernel, but most people who vote are unable to conceive of these as anything but large piles of chocolate coins, or something else they can put their mouths.
Because that’s how the average monkey interacts with money. They stack the blocks, the research assistant gives them a token, they exchange the token for a banana.
It’s no good trying to explain to the monkeys what supply chain is, or how a trillion dollars worth of rockets can’t magically be converted into a trillion dollars worth of bananas just because they’re both measured in dollars, as if a six-foot man and a six-foot plank of wood were interchangeable.
Finding a slightly different explanation, or getting the monkeys to sit still and really listen, doesn’t really help.
Because the problem isn’t just that the monkeys aren’t paying attention. The problem is that the monkeys are monkeys.
Their brains simply don’t have the developmental capacity to grow the neural connections they would need in order to grasp and manipulate the concept.
In the long term, this is why universal democracy is doomed. Because societies that let retards vote will fail, and be replaced by those that don’t.
You may think that we, as a society, face a great variety of problems. We do not. We have only one. Retards. Every other problem we have is downstream from their inability to understand the consequences of their political opinions.
But to fully grasp the implications of this, you have to understand that the definition of “retard” changes over time, as technology advances, because the IQ level required to grasp what’s really going on gets steadily higher and higher.
Eventually, the category “retard” grows until it includes the average person.
This has already happened.
Nick Knudsen isn’t dumber than the average guy. But the average guy, the 100 IQ salt of the earth guy that’s sitting on the next bar stool over, can no longer understand the modern economy. And this isn’t correctable, because the problem isn’t ignorance, it’s complexity.
You can’t make Nick Knudsen smarter by telling him things. You can’t even make him less ignorant, because the bare facts aren’t believable to someone who doesn’t have the framework to understand how they fit together.
The people who understand what’s going on are so much smarter than him that he doesn’t even think they sound smart.
He thinks they sound crazy.
June 14, 2026
The “Dissolution of the Universities” draws ever closer
On the social media site formerly known as Twitter, Steve McGuire reacts to more news about the conscious dumbing-down of modern university programs:
A Berkeley history professor said he’s gone from assigning 100 pages of reading per week to 35.
Another “said the earliest version of the … course he taught required seven full books, while his most recent iteration exclusively consisted of excerpts”.
“We are now reaching a crisis point where if the number (of pages) goes down further, it’s unclear to me whether my discipline of history can really be taught”, the first one said.
To which John Carter responds:
The academic death spiral is something to behold.
Demographics are steadily reducing the size of the student body, squeezing finances and driving bankruptcies.
At the same time, standards collapse is destroying the quality of the students the universities admit.
We’re already at the point where it’s common knowledge that a degree signals essentially nothing about intellectual ability. AI is exacerbating this, since cheating is so easy now.
Kids are already starting to forgo university, since they don’t think the cost of the credential is justified. That cuts even more deeply into the number of students universities can attract.
Universities respond by reducing standards even further (thereby accelerating brand destruction), by reducing tuition (which cuts even more deeply into budgets), and by firing professors in low-enrollment majors (reducing program variety, especially in the small seminars that are generally the most rewarding experiences for students).
[…]
“How can this be reversed?”
It can’t. There are pathways for individual institutions to revive themselves, even to prosper, but the sector as a whole is cooked. The death spiral is driven by prestige collapse as well as the demographic cliff, and intellectual prestige is inversely correlated to the size of the student body. More students means lower standards. That is especially true with a demographic cliff.
The only way to survive this crisis is ruthless elitism. Stop trying to edutain the fat middle of the bell curve, and refocus on the right tail. Become a place where the smartest people gather, and from which anyone who isn’t a 2-sigma outlier is excluded. This makes the school an arena in which intellectual iron can sharpen against iron. Elitism restored, prestige follows.
Next, eliminate the 500 person intro lectures. Admin loves these, since the high student:teacher ratio makes them cash cows. But they’re functionally no better than watching YouTube videos. Refocus on small seminars. This offers value that the Internet can’t.
Schools that take this path will restore or build reputations that will enable them to survive. However, they won’t be large. There is no future in which huge institutions keep tens of thousands of professors and administrators on payroll.
Update, 15 June: Welcome, Instapundit readers! Have a look around at some of my other posts you may find of interest. I send out a daily summary of posts here through my Substack β https://substack.com/@nicholasrusson that you can subscribe to if you’d like to be informed of new posts in the future.























