Links and Notes for October 17th, 2025

Below are links to articles, highlighted passages[1], and occasional annotations[2] for the week ending on the date in the title, enriching the raw data from Instapaper Likes and Twitter. They are intentionally succinct, else they’d be articles and probably end up in the gigantic backlog of unpublished drafts. YMMV.

Table of Contents

• Public Policy & Politics
• Journalism & Media
• Labor
• Economy & Finance
• Art, Literature, & Cinema
• Philosophy, Sociology, & Culture
• Technology & Engineering
• LLMs & AI
• Programming
• Fun

Public Policy & Politics

Against Chutzpah by Patrick Lawrence (Scheer Post)

“In history chutzpah has been variously cast as an admirable trait in the mode of “gotta be me,” and alternatively as an odious disregard for others. I have always been of the latter persuasion. I find chutzpah in any manifestation — whether it is a case of table manners, the conduct of public discourse, or any other small thing — repellent. It is one thing to liberate oneself from deadening orthodoxies. It is altogether another to hold oneself, garishly and abusively, above others.”

“Israel proposes to live and act in the community of nations, I mean to say, not according to law or what we know as morality or common forms of decency but according to what amounts to a biblically authorized project of subjugation and domination in the name of a righteous presumption of superiority. And with Zionist-nationalist fanatics now in control of the country’s direction, Israel has chosen this moment to insist that the world beyond its borders swallow this project as legitimate in the 21st century.”

There’s Now a Casino in Everyone’s Pocket. For Some Young Men, It’s a Near-Fatal Gamble by Paul Solotaroff, Eli Senor (Rolling Stone)

“The portals and drivers for much of this action were the giant sports-bet apps. On the party-colored killing floor of online gambling, FanDuel and DraftKings own most of the take, cornering 80 percent of the mobile bet market in this country. Eight years ago, Americans placed around $5 billion in sports bets. Last year, that number zoomed to nearly $150 billion; by 2028, we’ll have bet — and lost — a trillion dollars since 2018. That was the year the Supreme Court reversed a federal ban on legalized gambling, freeing each state to partner with Big Sports Bet and feed their residents, especially the young ones, to the wolves.”

““And that,” says Levant, “is why I chose this place.” He points to the flat-panels mounted above the tables, 50 or 60 sets tuned to Fox Sports 1 or the umpteenth rerun of “First Take.” Every last one of them posts a ticker at the bottom: Odds brought to you by either FanDuel or DraftKings. “This is what these guys have to live with,” says Levant. “They can’t run from sports or those fucking apps. All they can do is change their response.””

“Every major pro sports league followed football’s lead, selling their data for a slice of the sports-bet pie. The effect on problem gamblers was catastrophic. “I went from betting money lines on baseball games to betting the number of runs scored in every inning,” says Frankie, a client of Levant’s in his late twenties with a South Philly brogue and a shiny widow’s peak. “Any money left at the end of the night, I’m flipping to FanDuel’s casino. Then it’s slots and blackjack till I bust, and now I’m betting Chinese ping-pong at 3 a.m.””

“Those microbets and parlay packs that hooked Levant’s clients are the SBOs’ profit centers. How do we know this? Because the apps themselves say so: They’re the bets featured in their ads. Kevin Hart, Rob Gronkowski, Tom Brady, LeBron James: You can’t shut them up and make them go away when they’re touting props and parlays in every promo. Nor can you squelch their motormouthed peers on the pods and sports-bet shows: the Bill Simmonses and Charles Barkleys and Scott Van Pelts, who’ve merrily boarded the gravy train as “ambassadors” for the SBOs. (Approached for comment, Simmons, Barkley and Van Pelt declined to speak.) “Among the dangers of celebrity endorsements is the normalization of an addictive product,” says Levant. “They’re accepting enormous sums to push [that] addictive product on an increasingly younger audience.””

“Diana Goode, the executive director of the Connecticut Council on Problem Gambling, who likens the legalization of gambling to the opioid crisis. “It’s literally the same thing they did with pain pills. These companies hand out free samples [i.e., welcome bonuses] to get [young men] addicted to betting.””

“They’ve grown up immersed “in a stew of ads” from the Big Two betting apps; been chased across the web by their pings and promotions; and been told by the celebrities they trust most to think that betting’s how winners have fun. It normalizes gambling as “something cool to do with your friends,” she says. Now layer on the male-skewing lubricant of sports, and you’ve built “a mass addiction machine,” says Matt Gaskell, the clinical lead for the NHS Northern Gambling Service in England. “These companies engineered a product that exploits the reward pathways” of young brains. “The constant crackle of dopamine keeps them playing” — and then a big bump, equivalent to a “spike of heroin,” is triggered by “a win on their team.” Eventually, though, the wins and losses cease to matter. What keeps these kids in action is “that neurochemical feed that fires the desire centers in the brain.””

“Rather than confront the SBOs by slapping limits on their ads and promos — “our kids see 1,600 gambling logos in a 90-minute [soccer] match onscreen,” says Gaskell — the British government lamely lists “gambling disorder” as an official cause of death. “This industry has captured our policymakers with its billions, as I expect it’s done with yours. So the warning from over here is, expect disaster.””

“For every person hiding a gambling disorder, six people in their orbit are impacted financially, according to the World Health Organization. The collateral impacts of new gambling addictions are just now being charted by clinicians. Among states that have legalized sports-bet apps, bankruptcies are up by 30,000 a year, per a USC-UCLA study still in progress.”

These companies will never stop voluntarily. It’s just another form of plunder, funneling value away from the base animals—the wretched, stupid, and undeserving poor—who are nearly always solely responsible for their own victimization. It’s never the fault of the machine that plunders, which nearly always not only keeps its plunder but grows in power and wealth and retains its business model undisturbed. Our society not only does nothing to stop it—this is what it prefers, what it encourages.

“WHAT’S A YOUNG MAN TO DO when all the outlets he watches — ESPN, Paramount+, Peacock, Fox Sports — either own or have partnered with a sportsbook? When FanDuel and DraftKings push him their bet boosts while he’s scrolling reels? When SportsCenter plates him up a side of “Bad Beats” to pair with its “Top Ten Plays”?”

“Since grade school, we’ve been trained to blame the addict for addiction: a failure of will and want-to in the weak. Even when the truth emerges, we still default to that warhorse, character, as the root of personal ruin. It’s only when the operators are forced to pay out fortunes that we finally fault the poisoner, not the poisoned. Hundreds of billions recovered from the tobacco companies, not counting the giant verdicts they keep losing. More than seven billion from the Sackler family.”

The wheels of justice turn far too slowly. It’s always decades behind, allowing the next wave of scam artists—or just another business model from the same scam artists—to plunder, rape, and pillage to their heart’s content, all the while purchasing PR that lauds them for their altruistic and eminently praiseworthy dedication to bettering society with their latest scam.

“The complaint they filed was a strategic one: a tautly focused claim of consumer fraud. “Plaintiffs allege that the offer of the $1,000 bonus … was and is unfair and deceptive because, among other things, a new customer would, in order to get a $1,000 bonus, actually need to deposit five times that amount and then, within 90 days, place $25,000 in bets with only certain odds of return,” the suit reads. “In other words, the ‘$1,000 Bonus’ is structured so that it is inordinately expensive to obtain $1,000, and the new user is, instead, statistically likely to lose money by chasing the bonus.””

The Art Of Trade War by Indrajit Samarajiva (Indica)

“The Communist Party of China follows methodical five-year plans while the American government is just an insider trading club that is now pumping-and-dumping their entire economy every few weeks.”

“Trump enjoys holding up his signature and issuing edicts saying 100% tariffs on this, 30% tariffs on that. But this is light work, statements, not statesmanship. It’s just the music on Titanic, steering into an iceberg they could have avoided but hubris. China, on the other hand, speaks softly and carries a big stick, as Teddy Roosevelt said back when America was no less evil but far less stupid.”

“All America can do in a petulant fury is tax its own importers, effectively blockading its own ports. They didn’t even bother carving out exemptions for inputs they need, it’s just blanket tariffs that Trump clings to like a blankey because he’s an intellectual man-baby. America has no concept of heavy vs. light, they’re just trying to go heavy while being philosophically light.”

“China happily traded rare earths with America for years, but now that America is obviously trying to lynch China, they’ve stopped selling them rope. And can you blame them?”

“However, Americans approach elder civilizations with such basic disrespect that they’re incapable of learning anything. Even if China and Iran are enemies, there is no greater teacher than the enemy, as Mazer Rackham said in Ender’s Game. But America has outsourced its manufacturing and then manufactured those same countries into enemies. It’s literally self-defeating, and I for one am here for it. As Napoleon said, when your opponent is defeating themselves, why interrupt? America’s policy—especially under its idiot it in Trump—is shoot first and ask questions never, including where do we buy our buckshot?”

“America has marched into a trade war with only enough tinder to blow their own feet off. Which they have done, through tariffs. And what are they marching on? Their own supply lines, which China has just cut off, without firing a shot. This is why you don’t attack your own supply lines or start multiple land wars in Asia, but Americans ‘know neither the enemy nor themselves’ as Sun Tzu actually said, so they ‘will lose every battle, certainly.’ Now witness a trade war that’s going to go like every American war I’ve ever seen. They’re going to lose, and lose ugly.”

The U.S.A. will lose. It’s rulers will, as usual, win, for their narrow, unphilosophical definition of winning. Unfortunately, their definition of winning is also the working definition used by the entire world, as it somehow continues to look up to these self-nominated masters of the universe, who continue to amass power and wealth—and, BARF, admiration—from a world of sycophants whose only goal is to be trodding down rather than being downtrodden. Jesus wept.

How to fix the UK housing crisis by Cory Doctorow (Pluralistic)

“As housing prices went up, housing could be used as collateral for still more loans, which encouraged homeowners to stake their homes to borrow money in order to buy more homes to rent out. Because they have so much collateral (an overpriced home), they can borrow so much (from banks that can create money) that they are able to outbid people who don’t have a home yet and just want to buy a home so they can live in it.”

“The UK housing situation has been vapor-locked, because there’s a powerful voting and donating bloc of homeowners who want to keep house prices high, both to maintain their personal net worth, and to avoid having their “chained mortgages” collapse when prices fall and they suddenly no longer have enough collateral and the banks demand repayment.”

Ponzi! ⚅ ⚅ ⚅ ⚅ ⚅

“Here’s [Thomas] Edison:”

“[Ford] thinks it’s stupid, and so do I, that for the loan of $30,000,000 of their own money the people of the United States should be compelled to pay $66,000,000—that is what it amounts to, with interest. People who will not turn a shovel of dirt nor contribute a pound of material will collect more money from the United States than will the people who supply the material and do the work. That is the terrible thing about interest.”

“As Keen points out, it’s not merely that the banks that currently issue mortgages don’t “turn a shovel of dirt or contribute a pound of material” – they simply will not issue a mortgage to a median buyer. The median buyer can’t get a mortgage, so the system is rigged to make them pay someone else’s mortgage through their monthly rents, every month until they die.”

“The loser is the investment sector, the City boys who buy and sell mortgage debt. And you know, fuck those guys.”

God willing.

What Japan Taught me About American Trains by Quico Toro (Persuasion)

“It’s maddening. Because New York-Washington ought to be the perfect route for high-speed rail. At Japanese speeds, you could hop on in New York and hop off in D.C. about an hour and 40 minutes later. The Shinkansen, at peak cadence, moves around 20,000 people per hour in each direction. The Acela, less than 400. In a world where 16 Acelas per hour were leaving New York and reaching Washington in 100 minutes, how many airlines could compete? Not many. And that, one suspects, is why no such service will ever be allowed to exist.”

Borders and Scars by David Masciotra (CounterPunch)

“The working definition of “political violence” is an assault or murder with political motives committed by someone without political power. When those with political power plan, order, and execute acts of violence, even on a mass scale, it is excusable, justifiable, or even praiseworthy.”

“No major media figure or Democratic politician has pointed to the Grand Canyon-sized contradiction of claiming that “violence is not the answer,” while also promising to exercise State violence against a defenseless human being.

“Helen Prejean writes in her book, Dead Man Walking,”

“If we believe that murder is wrong and not admissible in our society, then it has to be wrong for everyone, not just individuals but governments as well. And I end by challenging people to ask themselves whether we can continue to allow the government, subject as it is to every imaginable form of inefficiency and corruption, to have such power to kill.”

“[…] calling to mind the John Lennon lyric,”

“There’s room at the top, they’re telling you still
But first you must learn how to smile as you kill…”

That’s from the song Working Class Hero.

US Politics Is Just Nonstop Fake Revolutions Now by Caitlin Johnstone (Substack)

“It’s two plutocrat-owned warmongering imperialist parties whipping their respective bases into the mass delusion that they are participating in a heroic act of revolutionary defiance by voting Democrat or Republican. They get everyone fighting a fake revolution so that nobody thinks about fighting a real one.”

Israel Flipped Out And Killed 45 Palestinians After Running Over Their Own Bomb by Caitlin Johnstone (Substack)

“In today’s news, Israel’s stupid fucking genocidal rapists ran over an unexploded ordnance from their own evil carpet bombing campaign, blamed Hamas for the explosion, started bombing the fuck out of Gaza again, killed scores of civilians, said they were once again cutting off aid to the enclave, and then quietly backed down on urging from Washington.

“Rather than report that Israel violated the ceasefire agreement as blatantly as any agreement could possibly be violated, the western press have been referring to this as a “test” of the ceasefire. Killing Palestinians is so normalized and accepted as a baseline expectation in the western press that CNN called it the “first major test” of the ceasefire after Israel killed people in Gaza every single day since the ceasefire agreement was signed.

“I hope the “WHY AREN’T YOU CELEBRATING?” crowd have gotten their answer by now. We weren’t celebrating because we know more than you. We’ve actually been paying attention, so we know Israel is going to seek out every excuse to kill Palestinians and torch this fake “ceasefire”.”

“Imagine thinking this is a good argument. Imagine thinking it’s perfectly reasonable to blow up a car full of children if they cross a made-up invisible line.

“[…]

“Imagine if that was happening in your country. If police just blew up your vehicle if you accidentally turned onto a one-way street or made an unauthorized U-turn. If they could send a drone to go pick you off if you were walking down a street they didn’t think you should be on.”

Journalism & Media

The Imperial Propaganda Machine Is Failing In Unprecedented Ways by Caitlin Johnstone (Substack)

“This entire dystopia is sustained by mass-scale mind control, and the mind control machine is getting weaker and weaker by the day. More and more people are waking up to the fact that we are ruled by tyrants, that our politicians and media have been deceiving us, and that everything we were taught to believe about our nation, our government and our world was a lie.

“So while in the short term things might look darker than ever before, what’s spelled out in the trends we are seeing tells us that the bars of our cage are made of melting ice. We are freeing our minds from the artificial delusions that have turned us into docile and obedient gear-turners, and awakening the healthy animals within us.

“I find it impossible to feel hopeless under such circumstances. I don’t feel certain that everything will work out perfectly fine, but I find it impossible not to have hope.

“They’re on the back foot. This has never happened before.

“We’ve got a real shot at winning this thing.”

Labor

Tech jobs bloodbath continues with Amazon announcing new round of layoffs by Dan Conway (WSWS)

“It is becoming clear that the recent round of tech layoffs is not part of a typical hiring boom-and-bust cycle. It is the result of a permanent restructuring process across the industry in which highly skilled workers, at least those who remain, will be facing ever greater exploitation and be forced to work even longer hours for even lower pay. The current job cutting process is underway while most large tech concerns are still experiencing massive increases in profits and stock valuations.”

“Throughout 2025, US companies have thus far issued 2,745 WARN notices affecting 216,545 employees. WARN (Worker Adjustment and Retraining Notifications) are required by law whenever companies with more than 100 employees terminate the employment of 50 or more employees within a 30-day period. Federal government layoffs are exempt from the WARN Act.”

Economy & Finance

Minsky Moments and AI CapEx by Paul Kedrosky (Substack)

“[…] Minsky divided financing behavior into three regimes:”

1. Hedge finance, where borrowers can meet all debt obligations from cash flow.
2. Speculative finance, where they can service interest but must roll over principal, and
3. Ponzi finance, where repayment depends on ever-rising asset prices or new borrowing.

“Over time, Minsky argued, as stability breeds complacency, economies drift from hedge toward Ponzi finance, creating a self-reinforcing boom driven by optimism and easy credit. Eventually, a shock—often minor—exposes cash-flow shortfalls, forcing asset sales and deleveraging. This abrupt reversal, the “Minsky moment,” as Paul McCulley coined it in 1998. famously triggers a cascade of defaults and falling asset prices, turning stability into crisis.

“Where are we in that cycle today with respect to data center financing? After all, the sums keep spiraling, with every year seeing regular revisions higher. Consider this: as the following figure shows, 2026 capex forecasts for the top 4 hyperscalers alone grew almost 50% during the year.”

OpenAI Needs $400 Billion In The Next 12 Months by Ed Zitron (Where's Your Ed At?)

“Broadcom and OpenAI have announced another 10GW of custom chips and supposed capacity which will supposedly get fully deployed by the end of 2029, and still the media neutrally reports these things as not simply doable, but rational.

“To be clear, building a gigawatt of data center capacity costs at least $32.5 billion (though Jensen Huang says the computing hardware alone costs $50 billion, which excludes the buildings themselves and the supporting power infrastructure, and Barclays Bank says $50 billion to $60 billion) and takes two and a half years.”

“Abilene’s 8 buildings are meant to hold 50,000 NVIDIA GB200 GPUs and their associated networking infrastructure, so let’s say a gigawatt is around 333,333 Blackwell GPUs at $60,000 a piece, so about $20 billion a gigawatt.”

“OpenAI has now promised 33GW of capacity across AMD, NVIDIA, Broadcom and the seven data centers built under Stargate, though one of those — in Lordstown, Ohio — is not actually a data center, with my source being “SoftBank,” speaking to WKBN in Lordstown Ohio, which said it will “not be a full-blown data center,” and instead be “at the center of cutting-edge technology that will encompass storage containers that will hold the infrastructure for AI and data storage.””

“There is not enough time to build these things. If there was enough time, there wouldn’t be enough money. If there was enough money, there wouldn’t be enough transformers, electrical-grade steel, or specialised talent to run the power to the data centers. Fuck! Piss! Shit! Swearing doesn’t change the fact that I’m right — none of what OpenAI, NVIDIA, Broadcom, and AMD are saying is possible, and it’s fair to ask why they’re saying it.”

“Number must go up, deal must go through, and Jensen Huang wouldn’t go on CNBC and say “yeah man if I’m honest I’ve got no fucking clue how Sam Altman is going to pay me, other than with the $10 billion I’m handing him in a month. Anyway, NVIDIA’s accounts receivables keep increasing every quarter for a normal reason, don’t worry about it.””

“OpenAI is saying it wants to build 250 gigawatts of capacity by 2033, which will cost it $10 trillion dollars, or one-third of the entire US economy last year.”

“In February, Goldman Sachs estimated that the global data center capacity was around 55GW. In essence, OpenAI says it wants to add five times that capacity — something that has grown organically over the past thirty or so years — by itself, and in eight years.”

“[…] build capacity assuming that literally every single human being on Earth uses this all the time.”

“I’m sorry, but what exactly is it that OpenAI has released in the last year-and-a-half that was worth burning $11.7 billion for? GPT 5? That was a huge letdown! Sora 2? The giant plagiarism machine that it’s already had to neuter?

“What is it that any of you believe that OpenAI is going to do with these fictional data centers?”

“I realize that it’s tempting to write “Sam Altman is building a giant data center empire,” but what Sam Altman is actually doing is lying. He is lying to everybody.

“He is saying that he will build 250GW of data centers in the space of eight years, an impossible feat, requiring more money than anybody would ever give him in volumes and intervals that are impossible for anybody to raise.

“Sam Altman’s singular talent is finding people willing to believe his shit or join him in an economy-supporting confidence game, and the recklessness of continuing to do so will only harm retail investors — regular people beguiled by the bullshit machine and bullshit masters making billions promising they’ll make trillions.”

The AI that we’ll have after AI by Cory Doctorow (Pluralistic)

“When the AI bubble pops, what will remain? Cheap GPUs at firesale prices, skilled applied statisticians looking for work, and open source models that already do impressive things, but will grow far more impressive after being optimized.

“The AI bubble companies are scams. They’ve spent most of a trillion dollars in capital expenditures, and by their own (very cooked and dishonest) numbers, they are grossing a total of $45b/year, industry-wide.”

“To recoup their existing and announced investments, AI companies will have to bring in $2 trillion, more than the combined revenue of Amazon, Google, Microsoft, Apple, Nvidia and Meta.

“And they have to bring in that $2 trillion before all those GPUs burn out…which is, again, about 2-3 years.

“Or sometimes just 54 days.”

“it’s far cheaper to pretend to be spending a lot of money than it is to actually spend it, and they’re doing plenty of that, too. Meta has promised to spend $72b next year on data-centers. However, Meta’s annual free cash flow is $52.1b. OpenAI says it will spend $60b/year on data-centers, which is five times its annual revenue of $12.7b (and the company is losing $9b/year). As The American Prospect’s Brian McMahon writes, “How can OpenAI plan to spend five times what it brought in?””

“Those people are going to get wrecked. And so are the rest of us. You don’t need to be an AI investor to get wiped out by the AI investment bubble, either. With 30+% of the S&P 500 tied up in seven AI companies’ stock, the coming crash will definitely escape containment and crash the whole damned economy.

“So the bubble is bad. Really bad. But even so, there will be things we can salvage from it: open source models, skilled programmers, cheap GPUs bought out of bankruptcy for pennies on the dollar. It would be better if we created that stuff without burning the world’s economy to the ground and emitting a heptillion tons of CO2, but ignoring the productive residue of the AI crash won’t bring the economy back, or suck the carbon out of the atmosphere.”

“There are a ton of these open source Chinese models, and they all perform like crazy. China does a lot of AI optimization because US embargoes prevent Chinese AI companies from accessing the most powerful GPUs, so Chinese coders tighten up their code and outperform US companies even though they’re using far less powerful computers.

“After the crash, everyone will be in a similar position to those Chinese AI optimizers: Chinese companies can’t buy advanced GPUs because of the embargo; and everyone else won’t be able to buy advanced GPUs because the AI crash will have cratered the economy for a generation.”

“This privacy-preserving, cheap-like-borscht component adds a voice-activated, conversational assistant to a device, sipping power like the clock on your microwave, running on a processor that costs less than a pack of AA batteries. It’s seriously fucking cool.”

Anatomy of a crypto meltdown by Molly White ([citation needed])

“In the span of minutes, Bitcoin plummeted around 10%. Altcoins plunged even more steeply, with the popular Solana token diving 40% and Trump’s own memecoin falling more than 60%. The trading firm Wintermute reported that the median crypto token price drop was around 54%, and more than 90% of tokens lost more than 10% of their value.”

“CoinDesk reported that “market depth collapsed by more than 80% across major exchanges within minutes.” Market makers — institutions that normally provide liquidity and price stability by taking the opposite side of trades — came under fire as some accused them of amplifying the crash by withdrawing liquidity during this crucial period. The Coinwatch crypto tracking platform accused market makers of “desert[ing] their responsibility”, and blockchain analyst YQ alleged “they executed a coordinated withdrawal at the optimal moment to minimize their losses while maximizing subsequent opportunities.” Others characterized these institutions’ pullback as a normal risk management response to elevated volatility, and the predictable actions of firms with no mandate to maintain market stability at the expense of their trading books.”

“Binance’s site went completely down at one point, and customers reported unexplained account freezes, unsuccessful trades, and automated protections like stop-losses failing to trigger. Several tokens intended to be maintain pegs to other assets, such as USDe, de-pegged on Binance’s Earn program. Coinbase’s status page claimed there was “latency or degraded performance when transacting”, although customers widely reported not being able to trade at all. The Kraken app showed customers a vague “something went wrong” screen, and customers reported similar issues with trades not completing and stop-losses not triggering. Robinhood users also reported the app freezing, and attempted trades not going through. Other exchanges including OKX, Bitget, and MEXC had intermittent outages, delayed trades, or inaccurate price information.”

When you would need to trade to stop losses and capitalize on your own gains, the platforms mysteriously stop working.

“Some have accused centralized exchanges of minimizing their own losses at their customers’ expense by intentionally halting trading or withdrawals under the guise of “technical difficulties”. Indeed, it is suspiciously common for supposedly highly sophisticated centralized exchanges to suddenly experience glitches or announce urgent “maintenance” under far less volatile circumstances.”

This is obviously what is happening. There is no regulation to prevent them from robbing their customers. And their customers keep coming back for more because it’s a cult.

“As prices fall, those trading on leverage are often given an opportunity to restore their positions to a “healthy” state by adding more collateral, thus increasing their margin level. But with the often slow process of converting fiat currency into cryptocurrency, often the only option for traders to obtain more crypto to use as collateral in an emergency is to sell off other crypto assets. This contributes to overall sell pressure as traders panic-sell assets to shore up their leveraged positions. And in rapidly falling markets, traders can be wiped out before they have any chance to add collateral.”

“[…] crypto exchanges routinely offer leverage up to 100× or more, accept volatile cryptocurrencies as collateral, and operate with minimal oversight. Traditional markets also have circuit breakers and trading halts that can pause cascading liquidations, and brokers typically follow careful procedures with multiple warning thresholds before forcing positions to close. In crypto, a position can be liquidated before a trader even knows they’re in trouble.”

Tesla profits fall 37% in Q3 despite healthy sales by Jonathan M. Gitlin (Ars Technica)

“Even though revenues grew by 12 percent to $28 billion compared to the same period last year, Tesla’s operating expenses grew by 50 percent. As a result, its operating margin halved to just 5.8 percent. And so its profit for the quarter fell by 37 percent to $1.4 billion.”

That company is still making $1.4B profit per quarter. Stop reporting this as if it were an unadulterated tragedy.

“Q3 saw a bigger profit decline than last quarter, and the first quarter wasn’t great either, but despite that, the automaker isn’t in much danger of falling behind on the rent. Free cash flow grew by 46 percent, and between cash, cash equivalents, and investments at the end of September, Tesla had $41.6 billion with which to pay for its future plans.”

You’ve got to be kidding me. This is ridiculous. It gets worse, though.

“The hit to profitability has come from several sides at once. It only took in $417 million in regulatory credits, compared to $739 million this time last year. That’s a problem that’s only going to get worse; in the US, the government is no longer enforcing the regulations that fine automakers for selling inefficient cars and trucks.”

The peerless injustice that is being transgressed against Tesla is that a company with $41B of cash reserves has to make ends meet with a 40% smaller government subsidy! But the government subsidy is still almost half-a-billion dollars.

500,000 Amazon jobs on chopping block due to automation in next few years by Tom Hall (WSWS)

“The question is not the technology itself, but who controls it. Under a rational and humane social system, automation could be used to vastly improve access to necessary goods, shorten the working day with no loss in pay, and fund pensions, healthcare and other social needs.

“But under capitalism, it is being used as an instrument of class warfare on a vast scale. These new technologies are being deployed to intensify exploitation in anticipation of another global recession and new economic crises caused, in the final analysis, by the massive and uncontrolled growth of financial speculation. Ever greater sources of surplus value are being drawn from the working class to keep financial bubbles from bursting.”

Art, Literature, & Cinema

The World is Insane and Thomas Pynchon Knows It by Ron Jacobs (CounterPunch)

“[…] our daily reality provides us with daily events that suggest this world is heading to its end. The media presents us with their version of those events, usually tailored to the sources of their funding. It’s a reason things often don’t make sense. Pynchon’s novels provide a different version, beholden not to money and its evils but to visions deeper, stranger and often darker. Ultimately, I would argue that they probably contain more truth. This novel is both prescient and a cleverly composed fiction reminding the reader who knows history how often it repeats itself yet never becomes any clearer.”

The Kafka Challenge by Paul Reitter (Hedgehog Review)

“Mann’s opening sentences are so full of extended modifiers and internal clauses that an acclaimed recent Anglophone translation simply drops one of those clauses for the sake of getting the sentences into literary English. In contrast to Mann’s fiction, moreover, Kafka’s largely avoids local references and also dialects, two things that can bedevil translators. Whereas Mann cultivated a musical style, at times echoing the rhythms of Wagner’s compositions, Kafka strove, as Mark Anderson has put it, to make his prose “non-musical,” even boasting of his “unmusical” nature in letters to his Czech translator Milena Jesenská.”

The tyranny of literacy by Victor Mair (Language Log)

“These ‘myths’ are not fiction. Most of the ancient myths of long-established cultures have an empirical core. They are not inventions but observations, filtered through worldviews from potentially thousands of years ago and clothed with layers of narrative embellishment before they reach us today. Framed within the science of their day, they represent knowledge often from times far earlier than those in the world’s oldest books.

“The ‘tyranny of literacy’ makes us sceptical of knowledge being retained in oral societies for such a long time.”

“My earliest encounters with people who could neither read nor write (and nor, in this case, speak English) were in the Pacific Islands where I lived and worked for more than two decades. As a geologist, my research took me to some of the remotest corners of the Pacific region, where my self-belief as a conventional scientist gradually eroded and was replaced with an appreciation of other worldviews equally as valid as that with which I had been inculcated. I also became disabused of the belief – held by most Western-educated literate people – that orality is inferior to literacy. As carefully explained by Walter Ong in his classic book Orality and Literacy: The Technologizing of the Word (1982), not only has literacy transformed human consciousness, shifting it from sound-focused to sight-focused, but is has also ‘weaken[ed] the mind’. For, as Ong wrote: ‘Those who use writing will become forgetful, relying on an external resource for what they lack in internal resources.’ Plato’s Socrates noted the same thing, arguing that writing ‘destroys memory’, something that sustained oral societies in every part of the inhabited world for tens of thousands of years.”

You know, I guess, maybe. Maybe I would be even more prolific without the written word. Maybe I would be an even more intense locus of intellectual power, shining an even brighter light, more intensely, without the written word. But I kind of fucking doubt it. Maybe I’m too unenlightened to even consider the possibility, too enshrined in my benighted world of the written word but I’m not sure I’m ready to gird myself for this battle. I may have missed the boat and, for once, I don’t really care. I don’t see any room for self-improvement by spending even more time than I already do in gathering information, because I would have to commit it to memory. In a way, now that I’m considering it, this is already what I do: I use all of these operations on the written word—the reading, the highlighting, the note-taking, the highlighting of emphases within the highlights, the expansion to more notes—all to help commit what I’ve read to memory, so that I can repeat it orally for those who don’t want to read, for those who prefer to hear me tell stories of that which I’ve read. I find it nearly impossible to even consider the possibility that this is inferior in some way to a purely oral tradition, that the imposition of the written word has somehow robbed the knowledge or wisdom of its purity, its power. That seems ridiculous on its face, not even worth measuring.

“Many people I know, including family, friends, professional colleagues, and, yes, readers of Language Log, engage in days long colloquies with ChagGPT and Ask AI Anything.”

What a sad waste of time. It’s a mirror dressed up asa toy dressed up as a serious tool for adults. Get a real hobby, you pathetic omphaloskeptics!

“Vedas are śruti (“what is heard”), distinguishing them from other religious texts, which are called smr̥ti (“what is remembered”). Hindus consider the Vedas to be apauruṣeya, which means “not of a man, superhuman” and “impersonal, authorless”, revelations of sacred sounds and texts heard by ancient sages after intense meditation.

“The Vedas have been orally transmitted since the 2nd millennium BCE with the help of elaborate mnemonic techniques. The mantras, the oldest part of the Vedas, are recited in the modern age for their phonology rather than the semantics, and are considered to be “primordial rhythms of creation”, preceding the forms to which they refer. By reciting them the cosmos is regenerated, “by enlivening and nourishing the forms of creation at their base.””

Philosophy, Sociology, & Culture

Generative AI has access to a small slice of human knowledge by Deepak Varuvel Dennison (Aeon)

“I find it hard to believe my dad’s herbal concoctions worked, but I have also since come to realise that the seemingly all-knowing internet I so readily trusted contains huge gaps – and in a world of AI, it’s about to get worse.”

“[…] the digital world reflects profound power imbalances in knowledge, and how this is amplified by generative AI (GenAI). The early internet was dominated by the English language and Western institutions, and this imbalance has hardened over time, leaving whole worlds of human knowledge and experience undigitised. Now with the rise of GenAI – which is trained on this available digital corpus – that asymmetry threatens to become entrenched.”

“The underrepresentation of Hindi and Tamil, troubling as it is, represents just the tip of the iceberg. In the computing world, approximately 97 per cent of the world’s languages are classified as ‘low-resource’. This designation is misleading when applied beyond computing contexts: many of these languages boast millions of speakers and carry centuries-old traditions of rich linguistic heritage. They are simply underrepresented online or in accessible datasets. In contrast, ‘high-resource’ languages have abundant and diverse digital data available. A study from 2020 showed that 88 per cent of the world’s languages face such severe neglect in AI technologies that bringing them up to speed would require herculean – perhaps impossible – efforts. It wouldn’t be surprising if the status quo is not too different even now.”

“[…] one study on medicinal plants in North America, northwest Amazonia and New Guinea found that more than 75 per cent of the 12,495 distinct uses of plant species were unique to just one local language. When a language becomes marginalised, the plant knowledge embedded within it often disappears as well.”

“Gramsci argued that power is not maintained solely through force or economic control, but also through the shaping of cultural norms and everyday beliefs. Over time, epistemological approaches rooted in Western traditions have come to be seen as objective and universal, rather than culturally situated or historically contingent. This has normalised Western knowledge as the standard, obscuring the specific historical and political forces that enabled its rise.”

“As climate change accelerates, these glass buildings are gleaming reminders of the dangers of knowledge homogenisation and epistemic hierarchies. Ironically, I’m writing this from inside one of those very buildings in Bengaluru in southern India. I sit in cooled air with the soft hum of the air conditioner in my ears. Outside, people saunter through a gentle drizzle. It looks like a normal monsoon afternoon – except the rains arrived weeks ahead of schedule this year, yet another sign of growing climate unpredictability.”

“[…] they often turn to elders from the Neeruganti community for advice. Their insights are valuable but their local knowledge is not written down, and their role as community water managers has long been delegitimised. Knowledge exists only in their native language, passed on orally, and is mostly absent from digital spaces – let alone AI systems.”

“LLMs also tend to reproduce and reinforce the most statistically prevalent ideas, creating a feedback loop that narrows the scope of accessible human knowledge.”

“For example, if pizza is commonly mentioned as a favourite food across a broad set of training texts, the model is more likely to respond with ‘pizza’ when asked ‘What’s your favourite food?’ Not because the LLM likes pizza, but because that association is more statistically prominent.”

“LLMs are optimised to predict the most probable next ‘token’ (the next word or word fragment in a sequence), which leads to a disproportionate emphasis on high-likelihood responses, even beyond their actual prevalence in the training corpus. Together, these two principles – uneven internal knowledge representation and mode amplification in output generation – help explain why LLMs often reinforce dominant cultural patterns or ideas.”

“This uneven encoding gets further skewed through reinforcement learning from human feedback (RLHF), where GenAI models are fine-tuned based on human preferences. This inevitably embeds the values and worldviews of their creators into the models themselves. Ask ChatGPT about a controversial topic and you’ll get a diplomatic response that sounds like it was crafted by a panel of lawyers and HR professionals who are overly eager to please you.”

“The most lucrative users – English-speaking professionals willing to pay $20-200 monthly for premium AI subscriptions – become the implicit template for ‘superintelligence’. These models excel at generating quarterly reports, coding in Silicon Valley’s preferred languages, and crafting emails that sound appropriately deferential to Western corporate hierarchies. Meanwhile, they stumble over cultural contexts that don’t translate to quarterly earnings.”

It’s the same as WEIRD (Wikipedia), which is the observation that nearly all psychological studies were performed on and reached conclusions about an extremely narrow section of the population that is “Western, Educated, Industrialized, Rich, and Democratic” (also, mostly white and speaking English)..

“LLMs predominantly reflect Western cultural values and epistemologies. They overrepresent certain dominant groups in their outputs, reinforce and amplify the biases held by these groups, and are more factually accurate on topics associated with North America and Europe. Even in domains such as travel recommendations or storytelling, LLMs tend to generate richer and more detailed content for wealthier countries compared with poorer ones.”

“With each training cycle, new models increasingly rely on AI-generated content, reinforcing prevailing narratives and further marginalising less prominent perspectives. This risks creating a feedback loop where dominant ideas are continuously amplified while long-tail or niche knowledge fades from view.”

“The AI researcher Andrew Peterson describes this phenomenon as ‘knowledge collapse’, a gradual narrowing of the information humans can access, along with a declining awareness of alternative or obscure viewpoints.”

“Peterson also warns of the ‘streetlight effect’, named after the joke where a person searches for lost keys under a streetlight at night because that’s where the light is brightest. In the context of AI, this would be people searching where it’s easiest rather than where it’s most meaningful.”

“All this means that, in a world where AI increasingly mediates access to knowledge, future generations might lose connection with vast bodies of experience, insight and wisdom.”

And they will have been trained not to care. They will never be able to miss what they will never be taught.

“The rationale isn’t that research-backed advice is always right or risk-free. It’s that it offers a defensible position if something goes wrong. In a system this large, leaning on recognised sources is seen as the safer bet, protecting an organisation from liability while sidelining knowledge that hasn’t been vetted through institutional channels. So the decision is more than just technical. It’s a compromise shaped by the structural context, not based on what’s most useful or true.”

“The marginalisation of local and Indigenous knowledge has long been driven by entrenched power structures. GenAI simply puts this process on steroids.”

“I have my doubts about whether Indigenous knowledge truly works as claimed in every case. Especially when influencers and politicians invoke it superficially for likes, views or to exploit identity politics, generating misinformation without sincere enquiry. However, I’m equally wary of letting it disappear. We might lose something valuable, only to recognise its worth much later”

It’s Not the Crime, It’s the Coverup by Freddie deBoer (Substack)

“[…] Sarah Manavis points out that the sharpest indictments of consumer culture often come from voices who maintain their integrity by refusing to participate in the very systems they dissect; when those voices cease resisting and instead become part of the machine, the critique collapses into complicity. And as a man who believes that, actually, selling out does exist, it is bad, I love that attitude. The sweaty communal effort to deny that selling out “is a thing” has been a poisonous turn in human culture. Because, you see, the profit motive really does distort and cheapen and poison artistic and cultural production, even if it would be more convenient for everyone if that wasn’t so. As human beings, we have values that go beyond the merely pecuniary, or at least I hope we do, and we have impulses that are driven by something other than self-interest, or at least I pray we do. When we have erased the critique of selling out as anachronistic, we’ve pretended that we have no choice but to sacrifice our deepest beliefs on the alter of commerce. And that’s stupid and bad.”

What Is Tim Dillon Doing? by Benjamin Y. Fong (Jacobin)

“When Socrates says that “god-sent madness is a finer thing than man-made sanity,” he means, among other things, that the experience of being disturbed allows us insight into the nature of the soul and some access to the truth of our condition. The experience itself can be a difficult one, involving “feeling contempt for all the accepted standards of propriety and good taste.” But it is being “sick with passion” in this way that creates the wonder that is the origin of the pursuit of truth.

“The “Life on a Boat” rant is a dreamlike presentation of life in late capitalism (and for those skeptics of that term, we can now define it as a form of capitalism wherein the Tim Dillon Show exists). It is disorienting and disturbing, but it is also captivating to lots and lots of people; if that is so, it’s because it reflects back to us the disorientation and disturbance of contemporary society in pseudo-personalized form. I say “pseudo” because nobody wants to identify with the “you” of Dillon’s story. But the magic works anyway, and we’re jolted into a fantasied confrontation with the horror and unsustainability of a world we barely understand.”

Technology & Engineering

Why Signal’s post-quantum makeover is an amazing engineering achievement by Dan Goodin (Ars Technica)

“The overhaul here adds protections based on ML-KEM-768, an implementation of the CRYSTALS-Kyber algorithm that was selected in 2022 and formalized last year by the National Institute of Standards and Technology. ML-KEM is short for Module-Lattice-Based Key-Encapsulation Mechanism, but most of the time, cryptographers refer to it simply as KEM.”

Interesting, because lattice-based is being marketed hard, despite being wobbly.

“The mechanism that has made this constant key evolution possible over the past decade is what protocol developers call a “double ratchet.” Just as a traditional ratchet allows a gear to rotate in one direction but not in the other, the Signal ratchets allow messaging parties to create new keys based on a combination of preceding and newly agreed-upon secrets. The ratchets work in a single direction, the sending and receiving of future messages. Even if an adversary compromises a newly created secret, messages encrypted using older secrets can’t be decrypted.”

“[…] when Alice sends Bob a message, she creates a new ratchet keypair and computes the ECDH agreement between this key and the last ratchet public key Bob sent. This gives her a new secret, and she knows that once Bob gets her new public key, he will know this secret, too (because, as mentioned earlier, Bob previously sent that other key). With that, Alice can mix the new secret with her old root key to get a new root key and start fresh. The result: Attackers who learn her old secrets won’t be able to tell the difference between her new ratchet keys and random noise.”

“Also known as trapdoor functions, these problems are trivial to compute in one direction and substantially harder to compute in reverse. In elliptic curve cryptography, this one-way function is based on the Discrete Logarithm problem in mathematics. The key parameters are based on specific points in an elliptic curve over the field of integers modulo some prime P.”

“The technical challenges were anything but easy. Elliptic curve keys generated in the X25519 implementation are about 32 bytes long, small enough to be added to each message without creating a burden on already constrained bandwidths or computing resources. A ML-KEM 768 key, by contrast, is 1,000 bytes. Additionally, Signal’s design requires sending both an encryption key and a ciphertext, making the total size 2272 bytes.”

“What does Alice do when she wants to send a message? What happens if we can lose messages, and we lose the one in fifty that contains a new key? Or, what happens if there’s an attacker in the middle that wants to stop us from generating new secrets, and can look for messages that are [many] bytes larger than the others and drop them, only allowing keyless messages through?”

“To manage the asynchrony challenges, the developers turned to “erasure codes,” a method of breaking up larger data into smaller pieces such that the original can be reconstructed using any sufficiently sized subset of chunks.”

“For those who care about the internal workings of their Signal-based apps, though, the architects have documented in great depth the design of this new ratchet and how it behaves. Among other things, the work includes a mathematical proof verifying that the updated Signal protocol provides the claimed security properties.”

How close are we to solid state batteries for electric vehicles? by M. Mitchell Waldrop (Knowable Magazine)

“Liu points to a prime example: the roll-to-roll process used for the cylindrical batteries found in most of today’s EVs. “You make a slurry,” says Liu, “then you cast the slurry into thin films, roll the films together with very high speed and precision, and you can make hundreds and thousands of cells very, very quickly with very high quality.”

“Lithium-ion cells have also seen big advances in safety. The existence of that flammable electrolyte means that EV crashes can and do lead to hard-to-extinguish lithium-ion fires. But thanks to the circuit breakers and other safeguards built into modern battery packs, only about 25 EVs catch fire out of every 100,000 sold, versus some 1,500 fires per 100,000 conventional cars—which, of course, carry around large tanks of explosively flammable gasoline.”

“Solid-state technology does have a geopolitical appeal, notes Ying Shirley Meng, a materials scientist at the University of Chicago and Argonne National Laboratory. “With lithium-ion batteries the game is over—China already dominates 70 percent of the manufacturing,” she says. So for any country looking to lead the next battery revolution, “solid-state presents a very exciting opportunity.””

There it is.

“So score one for solid-state batteries: Not only do the best superionic conductors offer a faster ion flow than liquid electrolytes, they also can tolerate higher voltages—all of which translates into EV recharges in under 10 minutes, versus half an hour or more for today’s lithium-ion power packs.”

“Standard lithium-ion batteries don’t use lithium-metal anodes because there is too high a risk of the metal forming sharp spikes called dendrites. Such dendrites can easily pierce the porous polymer membrane that separates anode from cathode, causing a short-circuit or even sparking a fire. Solid-state batteries replace the membrane with a solid barrier.”

“Major investments have come from startups such as Colorado-based Solid Power and Massachusetts-based Factorial Energy, as well as established battery giants such as China’s CATL and global carmakers such as Toyota and Honda.

“And there’s one big reason for the focus on superionic sulfides, says Wachsman: “They’re easy to drop into existing battery cell manufacturing lines,” including the roll-to-roll process. “Companies have got billions of dollars invested in the existing infrastructure, and they don’t want to just displace that with something new.””

LLMs & AI

We’re all going to be paying AI’s Godzilla-sized power bills by Steven J. Vaughan-Nichols (The Register)

“The AI companies’ plans are fantasies. There is no way on Earth the electric companies can deliver anything like enough juice to power up these mega datacenters.”

I remember living in New York City in the 1990s when there were brownouts every summer. I’m supposed to believe that the infrastructure has been improved not only to prevents brownouts—I read about them again last summer—but also to supposedly have a ton of extra capacity to subsidize whatever shenanigans our lords and masters in the tech world get up to? This is frankly unbelievable.

“The utilities will certainly do their best so they’re pushing their building plans as fast as possible. There’s only one little problem with that. Recall the project manager’s mantra: “You can have something that’s good, cheap, or fast – pick two.” Guess what? They’ve picked “good and fast,” so someone has to foot the bill. Guess who?”

“A Bloomberg News analysis of wholesale electricity prices shows “electricity now costs as much as 267 percent more for a single month than it did five years ago in areas located near significant datacenter activity.” Those bills are going to skyrocket in the next few years.”

Andrej Karpathy — AGI is still a decade away by Dwarkesh Patel (Substack)

“I do feel like the agents work in very specific settings, and I would use them in specific settings. But these are all tools available to you and you have to learn what they’re good at, what they’re not good at, and when to use them. So the agents are pretty good, for example, if you’re doing boilerplate stuff. Boilerplate code that’s just copy-paste stuff, they’re very good at that. They’re very good at stuff that occurs very often on the Internet because there are lots of examples of it in the training sets of these models. There are features of things where the models will do very well.

“I would say nanochat is not an example of those because it’s a fairly unique repository. There’s not that much code in the way that I’ve structured it. It’s not boilerplate code. It’s intellectually intense code almost, and everything has to be very precisely arranged. The models have so many cognitive deficits. One example, they kept misunderstanding the code because they have too much memory from all the typical ways of doing things on the Internet that I just wasn’t adopting. The models, for example—I don’t know if I want to get into the full details—but they kept thinking I’m writing normal code, and I’m not.”

Exactly this. I am writing code as she should be written, as we’ve all promised to write maintainable, extendible, testable, secure, and SOLID code. That is not what 99% of the code that these models inhaled during their training looks like. So they constantly try to correct your code or introduce new elements in a different style, so that, if you’re not careful, your style erodes down to the mediocre, barely passable code that forms the majority of code out there.

“You have eight GPUs that are all doing forward, backwards. The way to synchronize gradients between them is to use a Distributed Data Parallel container of PyTorch, which automatically as you’re doing the backward, it will start communicating and synchronizing gradients. I didn’t use DDP because I didn’t want to use it, because it’s not necessary. I threw it out and wrote my own synchronization routine that’s inside the step of the optimizer. The models were trying to get me to use the DDP container. They were very concerned. This gets way too technical, but I wasn’t using that container because I don’t need it and I have a custom implementation of something like it.”

This is a great example. Whereas the agents using the models can sometimes pick up unique stylistic patterns from the context, they will often be overwhelmed by the “weight” of the rest of the training data that insists that a certain library belongs to the pattern. A model is never going to know where my programs store IOC registrations because they’re not in the Program.cs like everyone else’s.

“They kept trying to mess up the style. They’re way too over-defensive. They make all these try-catch statements. They keep trying to make a production code base, and I have a bunch of assumptions in my code, and it’s okay. I don’t need all this extra stuff in there. So I feel like they’re bloating the code base, bloating the complexity, they keep misunderstanding, they’re using deprecated APIs a bunch of times. It’s a total mess. It’s just not net useful. I can go in, I can clean it up, but it’s not net useful.”

“I also feel like it’s annoying to have to type out what I want in English because it’s too much typing. If I just navigate to the part of the code that I want, and I go where I know the code has to appear and I start typing out the first few letters, autocomplete gets it and just gives you the code. This is a very high information bandwidth to specify what you want. You point to the code where you want it, you type out the first few pieces, and the model will complete it.”

“The other part is when I was rewriting the tokenizer in Rust. I’m not as good at Rust because I’m fairly new to Rust. So there’s a bit of vibe coding going on when I was writing some of the Rust code. But I had a Python implementation that I fully understand, and I’m just making sure I’m making a more efficient version of it, and I have tests so I feel safer doing that stuff. They increase accessibility to languages or paradigms that you might not be as familiar with. I think they’re very helpful there as well. There’s a ton of Rust code out there, the models are pretty good at it. I happen to not know that much about it, so the models are very useful there.”

This is a by-now classic fallacy. He’s literally suffering the Gell-Mann amnesia effect (Wikipedia) from one sentence to the next! In the first case, he knew exactly what he wanted and, so, was in a position to judge that the models were leading him astray. As soon as he admit that he didn’t know what he was doing as much, he deems the models trustworthy. A perfect fit!

Where’s the AI design renaissance? by Erik D. Kennedy (Learn UI Design)

“[…] so far as I’ve found:”

• There’s no evidence of massive designer productivity increases due to AI
• There no evidence of designer job loss due to AI
• I’ve not been able to significantly speed up my overall design process using AI
• I’ve not talked to any designers who have significantly sped up their design process

“If you had told me in late 2022 I’d be saying these things 3 years later, I would’ve been pretty surprised. “B-b-but − the tools are improving so fast! Your own workflow isn’t even noticeably improved!?”

“Don’t get me wrong. I’ve had some incredibly productive moments with AI design tools. But I’ve had at least as many slogs, where I can’t get it to do some basic thing I should’ve done myself 45 minutes ago. And even those productive moments are generally for less important, less business-critical, less live-in-production design stuff.”

“[…] one-off chats with an LLM are a terrible way for a non-designer to end up with a great design.

“Why do I say this? Because one-off chats with a human designer are a terrible way to end up with a great design!”

“AI design will be safe. If you ask it to be bold, it will be bold in a safe, reasonable, well-trod way.

“If your design has an opinion, something the median half-decent design would never touch, then the LLMs are already steering away from it. They may help you build it, but they won’t replace you in building it.

“They’ll be busy building “slightly above 2025 average”. But in a world inundated with average, what’s great will shine all the more. “Proof of humanity” will increasingly feel like a breath of fresh air in an onslaught of slop.”

This is similar to what Karpathy was saying above about writing good programming solutions.

Programming

Result isomorphism by Mark Seemann (Ploeh Blog)

“[…] languages that support exceptions have very specific semantics for that language construct. Specifically, an unhandled exception crashes its program, and although this may look catastrophic, it usually happens in an orderly way. The compiler or language runtime makes sure that the process exits with a proper error code. Usually, an unhandled exception is communicated to the operating system, which logs the error, including the stack trace. All of this happens automatically.”

“[…] you lose static type information about error conditions. Java is the odd man out in this respect, since checked exceptions actually do statically advertise to callers the error cases with which they must deal. Even so, in the first example, above, IllegalArgumentException is not part of the statically-typed method signature, since IllegalArgumentException is not a checked exception. Consequently, I had to invent the custom StatisticsException to make the example work. Other languages don’t support checked exceptions, so there, a compiler or static analyser can’t help you identify whether or not you’ve dealt with all error cases.”

The cost of design iteration in software engineering by Oren Eini (Ayende@Rahien)

“[…] in software, every modification demands a careful assessment of the existing system, long-term maintenance, compatibility with other components, and user expectations. This intricate balancing act is at the core of the engineering discipline.”

“While software designers might not grapple with physical forces, they contend with equally critical elements such as disk usage, data distribution, rules & regulations, system usability, operational procedures, and the impact of expected future changes.”

“This is a simple change, no? Just a few characters on the screen. No physical cost. But it is also a full-blown Epic Task for the project − even if we aren’t in production, have no data to migrate, or integrations to deal with.”

“I simply very strongly disagree that there is zero cost (or indeed, even low cost) to changing software once you are past the “rough draft” stage.”

How I provide technical clarity to non-technical leaders by Sean Goedecke

“I do other stuff too. I run projects, I ship code, I review PRs, and so on. But the most important thing I do − what I’m for − is to provide technical clarity.”

“In an organization, technical clarity is when non-technical decision makers have a good-enough practical understanding of what changes they can make to their software systems.”

“These people may have been technical once. They may even have fine technical minds now. But they’re still “non-technical” in the sense I mean, because they simply don’t have the time or the context to build an accurate mental model of the system. Instead, they rely on a vague mental model, supplemented by advice from engineers they trust.”

“Suppose a VP at a tech company wants to offer an existing paid feature to a subset of free-tier users. Of course, most of the technical questions involved in this project are irrelevant to the VP. But there is a set of technical questions that they will need to know the answers to:”

1. Can the paid feature be safely delivered to free users in its current state?
2. Can the feature be rolled out gradually?
3. If something goes wrong, can the feature be reverted without breaking user accounts?
4. Can a subset of users be granted early access for testing (and other) purposes?
5. Can paid users be prioritized in case of capacity problems?

“Finding out the answer to these questions is a complex technical process. It takes a deep understanding of the entire system, and usually requires you to also carefully re-read the relevant code. You can’t simply try the change out in a developer environment or on a test account, because you’re likely to miss edge cases. Maybe it works for your test account, but it doesn’t work for users who are part of an “organization”, or who are on a trial plan, and so on.”

“[…] you can be an impactful engineer without doing the work of providing technical clarity to the organization. Many engineers − even staff engineers − deliver most of their value by shipping projects, identifying tricky bugs, doing good systems design, and so on. But those engineers will rarely be as valued as the ones providing technical clarity. That’s partly because senior leadership at the company will remember who was helping them, and partly because technical clarity is just much higher-leverage than almost any single project.”

“[…] when you’re talking to the company’s decision-makers, you should commit to a recommendation one way or the other, and only give caveats when the potential risk is extreme or the chances are genuinely high.

“At the end of the day, a VP only has so many mental bits to spare on understanding the technical details. If you’re a senior engineer communicating with a VP, you should make sure you fill those bits with the most important pieces: what’s possible, what’s impossible, and what’s risky. Don’t make them parse those pieces out of a long stream of irrelevant (to them) technical information.”

“Effectively simplifying complex technical topics requires three things:”

1. Good taste − knowing which risks or context to mention and which to omit.
2. A deep technical understanding of the system. In order to communicate effectively, I need to also be shipping code and delivering projects. If I lose direct contact with the codebase, I will eventually lose my ability to communicate about it (as the codebase changes and my memory of the concrete details fades).
3. The confidence to present a simplified picture to upper management. Many engineers either feel that it’s dishonest, or lack the courage to commit to claims where they’re only 80% or 90% confident. In my view, these engineers are abdicating their responsibility to help the organization make good technical decisions.

Exploring PostgreSQL 18's new UUIDv7 support by Alexander Fridriksson & Jay Miller (Aiven)

“Using UUIDv7 is generally discouraged for security when the primary key is exposed to end users in external-facing applications or APIs. The main issue is that UUIDv7 incorporates a 48-bit Unix timestamp as its most significant part, meaning the identifier itself leaks the record’s creation time.

“This leakage is primarily a privacy concern. Attackers can use the timing data as metadata for de-anonymization or account correlation, potentially revealing activity patterns or growth rates within an organization. While UUIDv7 still contains random data, relying on the primary key for security is considered a flawed approach. Experts recommend using UUIDv7 only for internal keys and exposing a separate, truly random UUIDv4 as an external identifier.”

“Since UUIDv7 is timestamp-ordered, unlike the random UUIDv4, consider the impact on existing indexes and queries. It’s therefore recommended to test performance thoroughly with your specific workload.

“A few things to be aware of are that UUIDv7 relies on system clocks, requiring clock synchronization, like NTP, and that the timestamp precision is limited to the millisecond.

“Finally, it’s essential to update any foreign keys and external systems that depend on the specific UUID format to make sure nothing breaks.”

Fun

POP Phone (Native Union)

Pop Phone: Alarm Red

“Thoughtfully designed for more meaningful conversations, the POP phone helps you disconnect from distractions and reconnect with people. Its USB-C connection works effortlessly with your smartphone, laptop or tablet.”

• High-quality microphone and speaker
• No charging, no pairing, just plug and talk
• Optimized for video calls (Zoom, Teams, and FaceTime)
• Works with any USB-C device (Smartphones, Laptops, Tablets)
• Compatible with iPhone 15 and later (Not compatible with Lightning connector)
• Comfortable grip reduces hand strain during long calls
• Keeps your smartphone away from your face (reducing exposure to radiation)
• Built-in pick up and hang up button
• Made with recycled materials