The prediction market thread got very relevant very quickly.

Whether you're talking about the temptation to rig a traffic light or find a way to effectively sell government secrets, the potential to misuse these markets is huge.  

In case you were wondering, Polymarket had yet another spate of likely inside traders betting that the US would strike Iran by February 28. Per the due diligence investigation service Bubblemaps, the wallets used were created 24 hours earlier. The Pentagon Pizza Index has been replaced.

[image or embed]

— Matthew Sheffield (@matthew.flux.community) February 28, 2026 at 5:26 PM

“.. this is more or less offering a proxy market on assassination,” Amanda Fischer, a former chief of staff at the Securities and Exchange Commission, wrote on X .. @wsj.com www.wsj.com/world/middle...

[image or embed]

— Carl Quintanilla (@carlquintanilla.bsky.social) March 1, 2026 at 10:15 AM

It’s insane this is legal. People around Trump are profiting off war and death. I’m introducing legislation ASAP to ban this.

[image or embed]

— Chris Murphy (@chrismurphyct.bsky.social) February 28, 2026 at 6:09 PM

Emily Nicolle writing for Bloomberg:

As US and Israeli bombs fell on Iran this weekend, bettors on Polymarket — where $529 million was traded on contracts tied to the timing of the strikes — were cashing in. Almost immediately, blockchain sleuths began hunting for unusual patterns in recent bets.

Six accounts on Polymarket made around $1 million in profit by betting on the US to strike Iran by Feb. 28, according to analytics firm Bubblemaps SA. The accounts were all freshly created in February and had only ever placed bets on when US strikes might occur. Some of their shares were purchased, in some cases at roughly a dime apiece, hours before the first explosions were reported in Tehran.

These are the hallmarks that blockchain analysts associate with insider trading in prediction markets, an industry without widespread oversight and no agreed-upon methodology for distinguishing luck from leaks — and they’re far from conclusive on their own. Similar patterns suggested that an insider made a big profit betting on the ouster of Venezuela’s Nicolás Maduro in January, and have also been used to identify several other cases of alleged insider trading.

...

Kalshi Inc., a Commodity Futures Trading Commission-regulated rival, said Saturday it does not offer markets that settle on death. In the event of Khamenei’s death, it said it would resolve its contract based on the last price offered. Kalshi’s CEO Tarek Mansour later said on X that the platform would reimburse all trading fees from such bets.

Polymarket’s main trading platform is situated offshore and does not accept US-based customers, placing it outside the CFTC’s oversight. The company has argued that its contracts provide valuable data because they crowdsource information in volatile situations and help the public gauge risk, especially when conventional reporting lags.

Sidenote:

The Supreme Leader insisted on not taking special security measures even though he knew the attacks were about to start and was killed in his home. He wanted to die this way. The concept of martyrdom is an extremely potent, galvanizing force in Shia and Iranian culture.

[image or embed]

— Ali Ahmadi (@aliahmadi.bsky.social) March 1, 2026 at 2:42 AM

‪Ahmadi‬ continues.

There were def people in Washington who advised Trump not to do this, basically saying, "hes 90, hes had cancer twice, hes going to die soon anyway. Dont make him a martyr for the cause". He didnt listen. 

The very fact that this article calls to mind a Roald Dahl short story is probably a red flag

 Victor Tangermann writing for Futurism.

 New Platform Lets You Gamble on CCTV Footage

Thanks to the rise of sports betting and prediction markets, gambling has turned from an activity sequestered to casinos and scratch tickets into something that practically anybody with an internet connection can dive into headfirst — something that has experts warning of a surge in gambling addiction, especially among young, impressionable minds.

Platforms like Polymarket mean that gamblers are no longer limited to betting their hard-earned cash on red or black during a game of roulette or mindlessly pulling the lever of a slot machine — now they can bet on whether Jesus Christ will return before the long-awaited release of the video game “GTA VI,” or by what date the United States will strike Iran.

And now, a new gambling game called Rush Hour CCTV on the crypto casino platform Roobet is taking the phenomenon to an even more ludicrous conclusion. As casino publication Win.gg points out, gamblers there are betting on how many cars, trucks, buses, motorcycles, or pedestrians are crossing a specific point within a predetermined time period on a street in live, licensed CCTV footage being streamed from big cities, including Tokyo, Bangkok, New York, and London.

It may sound banal — and it many ways it is — but instead of relying on traditional random number generator mechanics (RNG) that determine the outcome in slot machines, the new game relies on the real world instead.

The game is deceptively simple. Each round kicks off with a simple message: “How many vehicles?”

Gamblers can then bet on how many vehicles will cross a point within the next 55 seconds. They can also give a range as an answer, which will give a significantly lower payout than guessing the exact number.

 

 The rest of the article focuses appropriately on the rise of gambling and gambling addiction, particularly among young men, but there’s another issue worth noting. (For more on that, check out this USA Today article.)

Whenever you can bet on something, there’s a temptation to try to rig the outcome. The best-known example is the almost universally condemned practice of fixing sporting events. As far as I can tell, even among the wackiest libertarians, there’s no great push to legalize point-shaving or to allow jockeys to bet against their own horses. Even compared to other forms of cheating, throwing a game—or even simply making sure your team doesn’t cover the spread—is seen as especially unacceptable.

But just to play devil’s advocate, isn’t this level of social opprobrium a bit excessive given the actual social harm? Sports, despite all the mythology we build up around them, are fundamentally trivial. Historically, the primary victims of things like point-shaving schemes have been people who were engaging in the generally illegal activity of sports betting. Yes, the winners were even less sympathetic—professional gamblers and organized crime—but we’re not exactly talking about bilking widows and orphans out of their life savings here.

We can certainly be impressed by the skill and dedication of a performer. We can appreciate the aesthetics of a great athlete (Muhammad Ali, Wayne Gretzky, my personal choice, fourth-quarter Joe Montana). But we could make a similar case for almost any form of entertainment. It’s not unheard of for an actor who is pissed off at the producers of a film to deliver a bad performance, but I’ve never heard of anyone suggesting they should be banned for life from the movies.

When, however, you start betting on real-life events, the potential consequences of rigging the outcomes can get very big very quickly. I can think of lots of ways to guarantee low traffic at a given intersection at a given time. None of them are things we would like people doing.

Roald Dahl explored this idea—and the possibility of its unintended consequences—in his classic short story “A Dip in the Pool.” It’s well worth checking out. You can find it in many anthologies online, including the Internet Archive, or you can check out one of the adaptations. It was filmed for Alfred Hitchcock Presents, starring Keenan Wynn (though that version does not appear to be streaming). If you don’t mind ’80s videotape cinematography, I’ve embedded a version from Tales of the Unexpected starring the fine character actor Jack Weston.

Good analogy, though under the circumstances, I'm a bit surprised he didn't go with Zelig.

Ed Niedermeyer, the journalist who wrote what is still the definitive book on Tesla specifically and on any Musk enterprise in general, and who is doing the ugly but necessary job of digging into the Epstein files to see what they tell us about the world’s richest man in an ongoing thread. 

Musk-Epstein: Year One
February 10, 2026

The rolling release of the Epstein Files is nothing short of a seismic event, throwing open the very ground we walk on and opening staggering new vistas into the ways in which our world actually works. Though much of work on the files so far has focused on Epstein’s trafficking of women and girls, for good reason given the horrors his victims endured and their inability to obtain justice thus far, the files illuminate far more about our fallen world. A kind of Forrest Gump of our hideously venal and corrupt elite, Epstein’s dealings are a skeleton key for an almost impossible number of terrible yet difficult-to-explain phenomena at the highest levels of wealth and power.

My little corner of that world centers on Elon Musk, whose empire of deception and greed has become something of a fascination for a decade now, starting as a simple automotive story. With the benefit of so much context, I’ve found a lot of material in the most recent releases of the Epstein files that make sense to me, but which might not mean anything to people who have managed to live their lives free of such sordid obsessions. With the first rounds of media and congressional reporting managing to avoid a lot of the Musk connections, I feel called to share what I’ve learned in my digging.

First, though, an important caveat: what follows is a limited first look at a massive corpus of evidence, which itself only offers a limited glimpse of events, and which has obviously been compromised by agenda-driven redaction. This is my best effort at putting elements from the Epstein Files into order and context, but should not be viewed as a comprehensive or final accounting of all the facts. I offer it here to help the public make the most of this unique moment of transparency, without malice toward anyone referenced within it. Though I personally believe that Jeffery Epstein and Elon Musk are among the worst individuals humanity has recently produced, they have the same right to the truth as everyone else. If anything, my time with the Epstein Files have only further convinced me that the worst of us deserve the truth most of all.

One of these days, there will be a serious, definitive biography of Jeffrey Epstein and one of Elon Musk. We are considerably closer to the first than the second. With Epstein dead and disgraced, and with most of his life now in the public record, it’s just a matter of letting the dust settle and finding a biographer who is up to the enormous task.

Musk is currently at the peak of his wealth and political power, and an alarming number of influential people still somehow buy into the personal mythology on which he built his empire, though the cracks in the façade are growing rapidly.

When the two books are finally available, you probably won’t see a lot of Epstein in the Musk biography, and vice versa, except for their mutual association with Donald Trump (who will be all over both books). This does not mean those overlaps will not be important. In Epstein’s story, Musk tells us a great deal about how one courted the rich and well-connected, while Epstein’s role in Elon’s story tells us a lot about Musk’s businesses, his tendency toward revisionism, and perhaps most of all, the underreported role that Kimbal Musk played.

In an age full of Fredos, Kimbal may be today’s definitive example—weak, ineffectual, easily influenced, and a general source of embarrassment. In his defense, he has been more loyal than the original Fredo Corleone (though that really hasn’t been tested so far). As a member of the handpicked boards of both Tesla, Inc. and SpaceX, he’s been a reliable rubber stamp whenever his brother decided to loot either company.

Kimbal’s Fredo tendencies have recently come into high relief thanks to the release of the Epstein files, largely due to the diligent work of Ed Niedermeyer, the journalist who wrote what is still the definitive book on Tesla specifically and on any Musk enterprise in general, and who is doing the ugly but necessary job of digging into the Epstein files to see what they tell us about the world’s richest man.

Epstein seems to have immediately zeroed in on Elon’s younger brother as the most vulnerable point of attack.

The origins of the Musk-Epstein relationship are one of the most important ways to understand who Jeffery Epstein was and how he operated. Though widely depicted as a sub-literate lecher, and certainly not the kind of intellectual he liked to court, Jeffery Epstein was not stupid. His casual, text-y (probably dyslexia-inflected) email style makes him easy to dismiss at first glance, but after reading enough of his email it eventually becomes clear that Epstein was in fact a remarkably sophisticated operator. Learning to appreciate Epstein’s actual abilities–the ability to spot opportunity, insinuate, flatter, hype, build intimacy, and connect–leads to a far more important lesson: these, and not any actual intellectual abilities, are the tools that our elite run on.

...

On September 18 [2012], Boris Nikolic (a medical doctor who worked with the Gates Foundation at the time) formally made the first connection between the Musks and Epstein in the files, in one of the more disturbing emails I’ve personally read. Nikolic’s entire email [PDF] reads as follows:

Hope you are having fun and getting adjusted to living a simple life [wink emoji inserted here -- MP]

Just talked to Kimbal. His actual bday is on Thursday night (although
his party is on Saturday).

Kimbal, Elon and few of their closest friends will go out that night.
I told him that I am going w Mette [Ed:likely the crown princess of Norway] for that Gala and that after we
wil join them somewhere — or even easiest that they should come to
Boom Boom Room.

I told him that you will join us as well. Also I told him that you are
coming with [victim redacted] and that he might want to ditch his ex/or current
to be. He said yes and is looking so much forward.

So please prepare [victim redacted]—;)
She might like Elon as well.

The overall image of Epstein that emerges in the files is of a man who understands connections in a world that runs on them, and already we have a good picture of how he operated. His connections allowed him to understand the preoccupations of his fellow elites, understand who they wanted to meet, work his own relationships to get into a room with them, and then use women as currency to win them over. In the very first reference to Epstein’s path crossing Musk’s we see him leveraging Nikolic’s relationship with Kimbal to reach Elon, and “preparing” a woman as the bait for his hook. In the meantime he had already informed [PDF] Jes Staley, head of private wealth management at Deutsche Bank, who Epstein appears to have served as a client funnel for, of the Musk’s presence in New York.

Kimbal, it seems, was easily hooked. A recent Guardian piece fills out the story: the woman was in fact a victim, who has said through a lawyer that she was trapped, coerced, and abused by Epstein. As the email makes clear Kimbal was both aware that Epstein wanted to meet him, and willing to “ditch his ex/or current to be” for a woman to be provided by the man who had been convicted of procuring a child for prostitution four years earlier. That initial hangout appears to have been jovial enough for Kimbal to email Nikolic on the 21st: “Fun time last night! Let Jeffrey and his friends know they are invited tomorrow night.”

 

 

 One essential bit of context that needs to be kept in mind when reading anything about connections between Epstein and the Musk family is that this all starts in 2012, four years after Jeffrey Epstein’s first arrest.

One of these days, when we have more distance from recent events, we should probably have a discussion about what the ethical or honorable thing to do is when we learn that a friend or someone we had admired has done something horrible. It is not a straightforward question; it is a complex one. But it’s not a question that applies here. Elon and Kimbal Musk knew they were dealing with a sex offender before they had any exchanges with Epstein.

Along similar lines, we should probably revisit the old debate (going back at least as far as Shaw's Major Barbara) about the ethics of accepting charitable donations from reprehensible people and, more to the point, about how far we should go in allowing these people to launder their reputations through such donations.

Once again, though, none of this applies to the Musk brothers, who, as far as I can tell, never showed any interest in charitable causes during their interactions with Epstein.

The files also give us additional insight into (or at least confirmation of) Elon Musk’s almost pathological arrogance and dishonesty, possibly even toward himself, regarding his decision to aggressively push for opening the files during his bitter, albeit brief, feud with Donald Trump a few months ago.

I previously suggested that Musk was so blind with rage over having someone in the White House drop the dime on his drug use and bizarre behavior in that The New York Times story that he simply didn’t think things through, particularly given his long history of evading consequences for his actions.

I’m still certain that’s at least part of the explanation, but having seen the relevant emails, I suspect that he had convinced himself that they weren’t all that damning. We are all revisionists when it comes to personal memories, something that heavy drug use can only exacerbate. It’s true that Musk was sometimes brusque with Epstein and had more than once rejected his invitations. For someone with a lifetime of seeing events through the most self-serving lens possible, it’s not difficult to see how he imagined himself getting out from under this one.

By early November Musk was “looking forward to” visiting Epstein’s island [PDF], by the end of the month he was infamously asking [PDF] about “the wildest party on your island,” and by December it was just a matter of discussing logistics [PDF]. It appears that Musk may not have actually made it to Little St James that winter, despite Epstein’s urging to clear customs in St Thomas [PDF], but it didn’t prevent Epstein from bragging to the Norwegian Crown Princess [PDF] that he would in fact be having lunch with Musk. One of the major impressions one gets of Epstein from his emails is that, like Musk himself, the gap between perception and reality is irrelevant. What seemed to matter to him was that people believed he was close to Musk, whether they were actually meeting or not.

Clearly Musk and Epstein were getting closer, as just months later the two mens staffs were coordinating a visit by Epstein to SpaceX, with three of “his girls” in tow. In the meantime, a series of puzzling emails [PDF] hint at a possible motivation on Epstein’s part for pursuing Musk, beyond a generically valuable social connection and/or a promising lead for Deutsche Bank’s wealth management department. This exchange, between Epstein and a redacted party that has been widely identified through redaction errors as Svetlana (“Lana”) Pozhidaeava, is one of many in the files fitting a pattern of women complaining to Epstein about his treatment of them. “I thought you went to califomia and want to be with elon now,” Epstein seems to tell her. “You have told me as much. So you confuse me.”

This cryptic exchange is made more compelling by an email from roughly three months earlier [PDF], in which Pozhidaeva recounts a long exchange between herself and one Joshua Fink, who “asked 6 times if Elon has given me anything? and what did elon give me?” Public photos of Joshua Fink, son of Blackrock CEO Larry Fink, with Svetlana Pozhidaeva at a 2011 gala provide a very real context for one of the more surreal references to Musk in the files. Reporting linking Pozhidaeva and other young Russian women in Epstein’s orbit with Putinist organizations and potentially Russian intelligence, provides yet another layer of context for a woman who the files suggests became close with Elon Musk as well as the son of one of Wall Street’s largest financial institutions.

...

The day after the SpaceX tour [February 27, 2013 -- MP]  Epstein emailed Musk [PDF], writing “thanks for the tour , you would have had fun at xmas,” to which Musk replied “I see [smiley emoji]”. In the weeks that followed, a pattern would emerge in the two mens relationship, with Epstein gently but persistently insinuating himself into Musk’s good graces, and Musk responding in a brief but friendly manner, always referencing his workload. One particularly solicitous March 2013 exchange [PDF] even sees Epstein suggesting “nuvigil” (Armodafinil, a relative to Modafinil, a potent non-stimulant eugeroic) as a way to offset his lack of sleep, a rare exception to Epstein’s horror of drug use.

...

A couple of months later, Epstein attempted to reconnect [PDF] with Musk on the same level, asking “will you come to caribean [sic] this xmas? woody allen with me, you might enjoy.” “Yes,” Musk replied. By mid-December the strategy had worked, and now, at long last, Musk was the one emailing Jeffery Epstein [PDF]. “Will be in the BVI/St Bart’s area over the holidays,” he wrote the convicted sex offender. “Is there a good time to visit?” 

 

 

Of course, this story is still in the process of being told, with journalists only beginning to dig through these documents and who knows how many still waiting to be released. Aspects such as Russia using connections with Jeffrey Epstein for spying and influence have only begun to be explored.

The Epstein files are also opening up considerable insights into the role that countries like Saudi Arabia and Dubai have been playing behind the scenes for decades. That is a topic that Niedermeyer is currently exploring later in the thread.

 

More takes on Citrini (I'm not the only one using scare quotes around "report")

 Yesterday we discussed this:

From CNN:

Fears of AI disruption continue to weigh on markets. Citrini Research on Sunday published a report on Substack laying out hypothetical scenarios for how developments in AI could disrupt certain parts of the economy. Stocks that were mentioned in the report tumbled on Monday.

American Express shares (AXP) sank 7.2% and had their worst day since April. Shares of DoorDash (DASH) and private equity firm KKR (KKR), two other companies named in the post, sank 6.6% and 8.89%, respectively.

 In a post with the somewhat overlong title:

 Traders who have ignored threats to Fed independence, erratic trade policy, and the deportation of much of the American labor force just panicked over a citation-free piece of fan fiction.

Now two of my favorite financial journalists have weighed in (both from newsletters so no links).

Allison Morrow of CNN: 

I put “report” in quotes because this 7,000-word screed amounted to little more than AI fan fiction — a dystopian thought experiment imagining a scenario in which AI is so successful it actually contracts economic growth and drives US unemployment rate to more than 10% by 2028. It went viral in a similar way as Matt Shumer’s similarly long-winded “Something Big Is Happening” blog post earlier this month, with people who are incentivized to make AI scary sharing it in “see I told you so” posts as if they were Prometheus bringing fire to the people. (I’ll get into this a bit more later this week, but suffice to say memos of this genre tend to have some blindspots, both substantive and stylistic). 

 

I could have spent my day debunking or otherwise making sense of the Citrini report but I — and I can’t stress this enough — did not want to. Instead, I’ll share some of the reactions from people much smarter than I am. 

 Matt Levine of Bloomberg:

I was writing specifically about a tiny company that had pivoted from karaoke to AI logistics and announced a disruptive AI logistics thing. (“I would probably be more inclined to be skeptical that this particular company is gonna be the one to disrupt the industry,” said an analyst, but added that someone probably will.) But of course you don’t even have to run the company that announces the disruptive thing. At this point, simply saying, publicly, “hey I think AI will disrupt _____,” for some company or industry or whatever, has a decent chance of driving down the price of _____. The market is really jumpy!

Obviously in all of these things it helps for your announcement to be well-written, well-reasoned and generally jazzy. But I have never seen a market where it has been so easy for an activist short to have a big impact. Like I feel like you could go on financial television today and say a company’s name, pause meaningfully, say “AI,” pause meaningfully, and walk off, and the company’s stock would drop 10%. Try it!  “DoorDash. AI. [grim nod].”

 

 

Traders who have ignored threats to Fed independence, erratic trade policy, and the deportation of much of the American labor force just panicked over a citation-free piece of fan fiction.

  

From CNN:

Fears of AI disruption continue to weigh on markets. Citrini Research on Sunday published a report on Substack laying out hypothetical scenarios for how developments in AI could disrupt certain parts of the economy. Stocks that were mentioned in the report tumbled on Monday.

American Express shares (AXP) sank 7.2% and had their worst day since April. Shares of DoorDash (DASH) and private equity firm KKR (KKR), two other companies named in the post, sank 6.6% and 8.89%, respectively.

Calling this a report is really stretching things. It is another one of those bloated “letter from the future” fanfics that are alarmingly common in the tech visionary world. Despite its absurd length, there is virtually nothing of substance here. I’m not going to attempt any kind of comprehensive takedown (fortunately, Ed Zitron has taken care of that in an annotated version I’ll be quoting from).

I tried to read the original first, but I only made it about a third of the way through—I don’t get paid for doing this, but if I did, you would not be paying me enough to do that—but with the snarky comments from Zitron reminding me that I was the one who was crazy, I managed to make it to the end.

There is not a single page here that doesn’t say something worth criticizing, so I’ll limit myself to the passage that caused so much damage in the financial services sector today.

 

 

 

 [Friction going to zero is a favorite incantation of the singularity crowd, but we'll have to come back to that in a future post.]

It should go without saying that a rational market would not rush to dump financial services companies because someone claims (with no support whatsoever) that AI is about to replace credit cards with crypto.

There is nothing of value in the Citrini "report," nothing useful to be learned, but the fact that people are listening to it tells us a great deal, none of it good.
 

A Tale of Two Logics (This is what you get when an old math and English teacher weighs in)

In his classic 1946 short story A Logic Named Joe, Murray Leinster told how, through some tiny malfunction in the manufacturing process, a small networked computer (called by Leinster a “logic”) developed sapience and rewired what we would call the internet to be all-knowing and willing to answer any question.

Imagine, if you will, a kind of mashup of a Leinster logic story and an Isaac Asimov robot tale, where we have two logics powered by two different types of AI but both capable of a high deqree of natural language processing. Simply by talking with them, we need to figure out which is which.

The first is what we might call a Pólya logic, capable of the kind of curiosity, heuristics, and intuitive thinking described by the great mathematician George Pólya in How to Solve It and his other books on mathematical reasoning.

The second is what we might call a dumb logic. It has the ability to search through massive data sets and find basic relationships, overlaid with algorithmic capacity for “reasoning,” such as generalization and synthesis, but on such a basic, limited, and crude level that the scare quotes aren’t just justified but required.

The dumb logic does, however, have one big advantage. While the Polya logic has access to a great deal of data—think everything you might find in a small local neighborhood library—the dumb logic has access to an unimaginable collection of data, virtually everything that has been digitized and put on a publicly accessible site.

Trying to distinguish between the two, we quickly get into the harder-problem fallacy (if you’re trying to distinguish between different students’ levels of understanding, particularly in situations where knowledge can possibly substitute for reasoning and comprehension, simply making questions harder will seldom help and will often do just the opposite).

“Achievements” such as passing the bar exam or acing a graduate-level test are largely meaningless. the dumb logic might just be regurgitating and paraphrasing from the huge collection of old tests and study guides. 

In fact, right answers in general tell us almost nothing. It is the wrong answers that are potentially informative, because that’s where we get insights into the underlying thought processes. If one of our logics gives a correct response and the other gives an incorrect one that nonetheless shows originality, insight, and a grasp of the underlying problem, the kind of answer a bright student new to a subject might give, then I would say the first was the dumb one. 

Perhaps the big ending of our Leinster/Asimov story might be the researcher announcing that the logic with the lower score was actually the more intelligent.

We have previously discussed the “Anna hurt Anna self” example (something I heard a concerned toddler say about her sister(. With language, as with everything else, young children constantly make mistakes as they try to master the world around them, but those mistakes tend to reflect underlying reasoning and comprehension.

By comparison, the mistakes made by LLMs frequently seem so bizarre because, while they are presented with a level of language mastery that we would usually associate with normal adult intelligence, they are saying things that reflect no grasp whatsoever of what they’re describing. 

The drive to reduce the error rates is understandable, but doing so with more data, more training, and particularly more post-training  may be making it more difficult to see the capacity and limits of the technology, which is especially troubling since part of the pitch we're betting the economy on is that these logics, like Joe, will not only continue to advance in the near future without significant new data or post-training, but will do so on an exponential curve. 

What Jamie Dimon meant when he talked about "the next DOJ"

 Back in November, Dimon pointed out that some of his peers weren't thinking about how things like suspicious looking contributions were going to be treated by future administrations, but people are starting to realize consequences may start sooner than that.

Talking Points Memo had a characteristically sharp post Thursday (Josh Marshall has been on a roll recently) about the risks corporations are taking by not only publicly cozying up to the Donald Trump administration, but in some cases crossing the line into bribery- and kickback-territory. Marshall reminds us that it is not good citizenship but sound business logic that leads companies to avoid appearing aligned with one party or the other.

And yet now, as Hoffman points out, lots of corporations are starting to realize that these moves are probably not going to age well at all in 2026. There’s a major public backlash brewing, and there’s a very good chance that at least the House will fall to Democrats and the Senate is now more than the theoretical possibility it seemed a year ago. This is of course why corporate America, particularly the big diversified and largely de-personalized mega corporations, has always tried to steer clear of being too identified with either party. There’s political giving. But what we’ve seen over the last year goes far, far beyond that. Lots of big corporations have aggressively competed to be part of Trump’s corporation, in some ways that are not precisely illegal and in many that clearly are … as soon as there’s a functioning Justice Department back on the beat.

Hoffman links to an important overnight tweet from Sen. Ruben Gallego (D-AZ) in which he says this: “All the mergers approved under the Trump administration need to be undone. Big business has to understand there are consequences when they team up with corrupt government.” In its own way, what Gallego is calling for and committing to is almost more important than criminal accountability for those most guilty of criminal conduct in the administration itself. There’s a reason corporations generally steer clear of aggressive and visible political involvement. Because the tide turns. It’s better to be on at least reasonably good terms with both parties than to seek extreme advantage when one or the other is in power. 2025 went well beyond that with numerous corporations diving into the Trumpian corruption, allowing the president to build subservient media empires and more. Behind the scenes I have had explained to me the kind of transparent cash transactions that titans of corporate America are now ponying up to stay on Trump’s good side. Only public, visible and lasting consequences can reestablish deterrence against participation in these kinds of plots against the American republic. It’s not just creating consequences for criminality in public office. That criminality can only flourish if other major societal stakeholders go along with it, try to participate in it. And corporate America has done so — because of a mix of carrots and sticks — in a way that is almost unprecedented in American history.

 

A number of high-value stocks (Palantir Technologies, for instance) are priced at a huge premium that can almost certainly be attributed to the company’s relationship with the Trump White House. It’s possible that the magic of the market is telling us that Democrats won’t flip the House or the Senate, but I think it’s more likely that these traders simply haven’t priced in the possibility of aggressive congressional oversight.

Another factor is that public opinion appears to be solidifying against both the current administration and against big tech. This could play out in any number of interesting ways. As Disney learned earlier this year, while standing up to Trump certainly brings risks, so does publicly caving. Companies like Apple Inc. are already in danger of serious brand damage (and brand is the most valuable asset that particular company has).

The desire to go after companies like Meta are likely to be even more intense with large polling numbers supporting the effort. Under normal circumstances, Democrats would probably be reluctant to be seen too publicly going after political enemies, but given the current popularity of Elon Musk and Mark Zuckerberg, it’s difficult to imagine much of a backlash.

 

Aerodromes and LLMs

 What happens when, faced with a breakthrough in a transformative technology, we decide to pour all of our imaginable resources into the front-runner? I’m sure we could all think of some contemporary cases, but a historical example might be more informative.

Let’s take heavier-than-air flying machines shortly before 1900. Though we’ve seen a great deal of revisionism around this (much of it coming from management-consultant TED Talks), the most advanced technology in the field was very probably being developed by Samuel Pierpont Langley in the area of steam-powered aircraft. 

https://en.wikipedia.org/wiki/Samuel_Langley#Aviation_work 

His first success came on May 6, 1896, when his Number 5 unpiloted model weighing 25 pounds (11 kg) made two flights – 2,300 ft (700 m) and 3,300 ft (1,000 m) – after a catapult launch from a boat on the Potomac River.^[11][12] The distance was ten times longer than any previous experiment with a heavier-than-air flying machine,^[13] demonstrating that stability and sufficient lift could be achieved in such craft. 

On November 11 that year his Number 6 model flew more than 5,000 feet (1,500 m). 

Despite these early successes, Langley himself saw that internal combustion was the future of flight and by 1901 had commissioned what would turn out to be the revolutionary Manly-Balzer engine for his first attempt at manned flight. 

Nonetheless, in 1896, steam-powered aircraft was the state of the art. What if we had put everything into developing this technology and building an industry around it, maxing out financial and research resources and effectively crowding out work in competing and adjacent fields?

Eventually, internal combustion would have won out (the advantages of the technology were overwhelming), but we would probably have ended up with a lost generation of progress not just in aviation but in all sorts of related areas. It would have been a huge waste of resources with tremendous opportunity costs.

It also would have done the exact opposite of what it was intended to do.

Right now, with respect to AI—and certainly with respect to natural language processing—large language models are the clear front-runner, and we are in the process of spending unprecedented sums trying to get them to the next level. This decision has been driven largely by hype and bubble mania, with developers being given both almost unlimited resources and permission to break whatever laws get in their way, ranging from intellectual property to environmental regulations.

It seems like the burden of proof should be on the boosters to argue that we aren’t betting much of the world’s economy on yet another steam-powered airplane.

Imagine telling people twenty years ago that two of the most important stories in political media would revolve around late night network talk shows.

 

It was already a dinosaur of a genre in 2006, largely trapped in amber since the late sixties. It had even reached the deconstruction phase which normally indicates the death knell. Somehow, though, the format has not only survived but has become more relevant than the journalistic institutions that were supposed to be the guardians of democracy. The shows still had silly sketches and often inane interviews, but Colbert and Kimmel rose to the moment, something that very few editors at the New York Times et al. could claim. 

Extraordinary television. 

Why CBS Didn't Broadcast Stephen Colbert's Interview With James Talarico



Rep. James Talarico On Confronting Christian Nationalism, And Strange Days In The Texas Legislature



Why Everyone's Talking About Stephen Colbert, CBS, The FCC And James Talarico

We haven't complained about television for a while

 One of the biggest stories that has received next to no coverage over the past 10 years has been how television—which for well over half a century was a huge cash cow—has become, at best, profit-neutral. It’s a familiar story: an industry that learned all the wrong lessons from the dot-com bubble, forgot everything it knew about running a business, and credulously bought into tech-visionary narratives and the idea of the hype economy.

Let’s look at the case of Good Cop/Bad Cop.

Good Cop/Bad Cop impressed critics to the tune of a 91% fresh rating on Rotten Tomatoes, but the love of reviewers was seemingly not enough to save the Leighton Meester-led show from getting the axe after season 1.

There will apparently be no Good Cop/Bad Cop season 2, after star Luke Cook went on TikTok and announced the series’ cancellation (via TVLine). The news has not yet been confirmed by The CW.

"Sad to let you know this, but we will not be doing a Season 2 of 'Good Cop/Bad Cop,” Cook told his fans in a video. "I loved making that show. I loved making friends with everybody on that show, the cast and crew, and John [Quaintance], who wrote the show, they're all my great friends.”

Centered on a brother-and-sister detective team on a small town police force in the Pacific Northwest, Good Cop/Bad Cop received plenty of positive feedback from reviewers, who praised the show for its deft mix of comedy and drama, and for the performances of its main cast, particularly Gossip Girl alum Meester.

A 92% Popcornmeter rating indicates audiences have been as enthusiastic about the show as reviewers. Good Cop/Bad Cop indeed seemed set up for long-term success when it made its Prime Video debut last July, rising quickly to place on the streamer’s top 10 series chart.

This is a charming little no-budget show, definitely worth checking out if you have any interest in the genres represented or affection for the leads (including fan favorite, Clancy Brown). It’s not difficult to see how it won the critics over. More impressively, this is an example of that long-thought-extinct species: the word-of-mouth hit. As far as I can tell, there was no real time or money spent promoting this show initially. People found it on their own when it popped up on Prime.

To put this into context, streamers routinely spend multiples of the series’s budget on shows that bring in far lower numbers with worse reviews. On top of that, since this show ran over the air first, it had already brought in advertising revenue, possibly breaking even before it hit Amazon. Unfortunately, one of the most consequential lessons executives have managed to unlearn is the value of multiple revenue streams.

By any reasonable business criteria, Good Cop/Bad Cop should have been renewed the moment those Prime numbers came in—but those are not the rules television executives are playing by in 2025. They remain convinced that the future lies with streaming originals and star vehicles with big production budgets, often accompanied by comparable PR and marketing budgets. That is the model that fits the narrative and generates the hype. True, it often leads to, let’s say, questionable financial outcomes—like Peacock recently announcing that it had lost $552 million in its last quarter—but what's a few billion between friends. 

The harder-problem fallacy (which is about to become relevant again)

If you’re trying to distinguish between different students’ levels of understanding—particularly in situations where knowledge can possibly substitute for reasoning and comprehension—simply making questions harder will seldom help and will often do just the opposite. For example, if a Math Olympiad style test switched from geometry questions to trigonometry questions, the exam would mainly be good at identifying which students had taken pre-cal. 

In these cases, a well-designed test will find a way of leveling the playing field so that additional information and training will not give one person an advantage over another. One of the best examples of this is the old SAT reasoning test, before David Coleman—The New York Times darling—“fixed” it.

An old English professor of mine (who, not entirely coincidentally, introduced me to Raymond Smullyan) accurately described it as the toughest ninth-grade math test you’ll ever take. In terms of knowledge, it didn’t require anything beyond Algebra I and a few really basic geometry concepts that were helpfully provided on the first page of the test. On top of that, forms of notation were invented so that the student who hadn’t taken a math course for a year or two was on a more or less equal playing field with the kid who was well into the first semester of calculus. 

Back in 2014, we talked about how the SAT worked around the harder-problem fallacy (though not by that name) and about how the reporters covering the test (which was at the time out of fashion with the NYT et al. before shifting again) kept missing the point.

As you have guessed, we'll be connecting this to our AI thread in a few days.  

Perhaps we should add "opaque" to the list of journalists' vocabulary questions  

Last week, Andrew Gelman criticized Todd Balf for picking words and phrases for their emotional connotation rather than for their actual meaning in his New York Times Magazine article on the changes in the SAT. 'Jeffersonian' was the specific term that Gelman choked on. I'd add 'opaque' to the list though the blame here mainly goes to David Coleman, president of the College Board and quite possibly the most powerful figure in the education reform movement:

For the College Board to be a great institution, [Coleman] thought at the time, it had to own up to its vulnerabilities. ... “It is a problem that it’s opaque to students what’s on the exam."

There's a double irony here. First because Coleman has been a long-standing champion of some very opaque processes, notably including those involving standardized tests, and second because test makers who routinely publish their old tests and who try to keep those tests as consistent as possible from year to year are, by definition, being transparent.

This leads to yet another irony: though the contents of the tests are readily available, almost none of the countless articles on the SAT specifically mention anything on the test. The one exception I can think of is the recent piece by Jennifer Finney Boylan, and it's worth noting that the specific topic she mentioned isn't actually on the test.

Being just a lowly blogger, I am allowed a little leeway with journalistic standards, so I'm going to break with tradition and talk about what's actually on the math section of the SAT.

Before we get to the questions, I want to make a quick point about geometry on the SAT. I've heard people argue that high school geometry is a prerequisite for the SAT. I don't buy that. Taking the course certainly doesn't hurt, but the kind of questions you'll see on the exam are based on very basic geometry concepts which students should have encountered before they got to high school. With one or two extremely intuitive exceptions, all the formulas you need for the test are given in a small box at the top of the first page.

As you are going through these questions, keep in mind that you don't have to score all that high. 75% is a good score. 90% is a great one.

You'll hear a lot about trick questions on the SAT. Most of this comes from the test's deliberate avoidance of straightforward algorithm questions. Algorithm mastery is always merely an intermediary step -- we care about it only because it's often a necessary step in problem solving (and as George Pólya observed, if you understand the problem you can always find someone to do the math) -- but when students are used to being told to factor this and simplify that, being instead asked to solve a problem, even when the algorithms involved are very simple, can seem tricky and even unfair.

There are some other aspects of the test that contribute to the reputation for trickiness:

Questions are written to be read in their entirety. One common form breaks the question into two parts where the first part uses a variable in an equation and the second asks the value of a term based on that variable. It's a simple change but it does a good job distinguishing those who understand the problem from those who are merely doing Pavlovian mathematics where the stimulus is a word or symbol and the response is the corresponding algorithm;

Word problems are also extensively used. Sometimes the two-part form mentioned above is stated as a word problem;

One technique that very probably would strike most people as 'tricky' actually serves to increase the fairness of the test, the use of newly-minted notation. In the example below, use of standard function notation would give an unfair advantage to students who had taken more advanced math courses.

One thing that jumps out when us math types is how simple the algebraic concepts used are. The only polynomial factoring you are ever likely to see on the SAT is the difference between two squares.

A basic understanding of the properties of real numbers is required to answer many of the problems.

A good grasp of exponents will also be required for a perfect score.

There will be a few problems in basic statistics and probability:

I've thrown in a few more to make it a more representative sample.

We can and should have lots of discussions about the particulars here -- I'm definitely planning a post on Pavlovian mathematics (simple stimulus/algorithmic response) -- but for now I just want to squeeze in one quick point:

Whatever the SAT's faults may be, opaqueness is not among them. Unlike most of the instruments used in our metric-crazed education system, both this test and the process that generates it are highly transparent. That's a standard that we ought to start extending to other tests as well.

 

If you want to get a feel for the post–Musk Silicon Valley, you really should watch these videos.

It’s all here: the transparent scams, the intellectual laziness, the stunningly bad engineering, the messianic delusions of grandeur, the painfully overused tropes and buzzwords—all backed by Y Combinator and reported on by a credulous press.

The team at Common Sense Skeptic has done their usual well-researched job of pointing out the absurdities and fallacies here, though they miss probably a good 30 percent of what deserved mockery. Though, in their defense, each video is only 25 minutes long.

 

Pay close attention when the narrator talks about the Kardashev scale. This figures prominently in the Elon-verse.

Pleasantly surprised to see the NYT actually connect the dots here

 See, I do occasionally give the paper of record credit.

  Gallup Will No Longer Track Presidential Approval Ratings by Ruth Igielnik 

 After nearly 90 years, the Gallup Organization will no longer track presidential approval ratings, which served as a steady way to measure Americans’ views of their elected leaders.

The polling firm, which has been tracking presidential approval since Franklin D. Roosevelt, said that the decision was based on a shift in corporate strategy, intended to focus more on issues and policy polling.

“We’re focused on providing analytics that inform and drive meaningful change,” Justin McCarthy, a spokesman for Gallup, said.

...

Gallup’s decision comes as President Trump has escalated his threats against the press, and sued at least one respected pollster, J. Ann Selzer. He accused Ms. Selzer and The Des Moines Register of election interference after publishing a poll just before the general election showing Kamala Harris leading in Iowa. Mr. Trump won the state by 13 percentage points.

Gallup’s last presidential approval rating, in December 2025, put Mr. Trump’s rating at 36 percent, among his lowest ratings as president.

 

 

The Voss Chronicles

 

I was watching some videos to unwind before bedtime last night, and the following ad popped up. I normally would have skipped it, but something about it piqued my curiosity and, since it was less than 3 minutes, I decided to let it play.

I was not disappointed.

The simple, man-of-the-soil pitchbot explained that he had been working for an inexplicably wealthy farmer who claimed that the secret of his success was a system that had allowed him to win the lottery multiple times using AI. The implausibility of the pitch would have been enough to hold my interest, but it got better. The genius behind this amazing AI was a scientist named Dr. Leonard Voss.

Readers of the blog and students of generative AI esoterica will remember Elara Voss:

What’s so odd about this is that--for a name now so common across the megaplatforms--before 2023, “Elara Voss” did not exist. There is no person named Elara Voss in the United States. No birth certificate has ever been issued under that name; if you search for it in public records databases, you’ll turn up no results. There aren’t even any characters named “Elara Voss” in any book published before 2023. Until two years ago, the two words didn’t ever appear next to each other even by accident.

But if you direct almost any L.L.M. to generate a sci-fi story or narrative for you, it will name the main character “Elara Voss”--or a similar variation like “Elara Vex,” “Elena Voss,” or “Elias Vance”--with an alarming degree of frequency.

I can't say for certain that Dr. Leonard Voss is an ancestor of Elena's, but a quick Google search did show him popping up in similar roles. Perhaps some future LLM-generated novel will reveal that the Voss family dynasty was built on Powerball winnings.

The AI and the Super Bowl Commercial Bubble Curse

 

 

One of these days, I need to do a serious thread on my take on the capacity and likely impact of LLMs (tldr version: big advance in natural language processing; major productivity boost for coding and a few other fields; but loads of ugly unintended consequences; probably a Langley steam airplane). For now, though, I'm keeping the focus on the AI bubble.

For those of us old enough to remember the dot-com bubble, there are some disturbing parallels. I am always reluctant to make too much of historical precedent—there are certainly some differences here—but the differences themselves aren’t all that reassuring.

The ads of the dot-com Super Bowl, at least those that have lodged in memory, were mainly about establishing brand awareness, be it the notorious sock puppet or the dancing chimpanzee followed by the tagline “we just wasted 2 million dollars.”

The ads of this weekend’s Super Bowl, at least those I got around to watching, seem to be about convincing potential customers that this is a product they would like to use. The difference in tone reflects the fact that while the excesses of the dot-com boom were widely mocked, people didn’t hate the underlying products the way surveys indicate they do AI.

One exception was the Anthropic ad, which seemed to want to start a pissing war with OpenAI. It’s the sort of competitive back-and-forth you might expect in a mature industry, with a company trying to gain an edge over its competition—think about those “I’m a Mac” and “I’m a PC” ads from back in the day. Anthropic and OpenAI, however, are two money-imagining startups that are still raising cash primarily on the assumption that some product they will come up with in the future will be the killer app.

One key similarity between a quarter of a century ago and today is that I suspect both sets of ads are primarily intended to convince investors that not only is this the next big thing, but that the payoff is just around the corner. That corner turned out to be quite a few years later than promised for the dot-com companies. We’ll see how things work out for this generation.

Chris Isidore writing for CNN:

 

If there was an overriding message throughout the Super Bowl ads last night, it’s that artificial intelligence is your friend.

...

Americans have widespread concerns about how AI is changing society, from jobs to social relationships, surveys show. But AI companies, flush with cash from surging stocks, used the Super Bowl not to sell specific products but instead sell a vision of a kinder, gentler AI future.

The ads were about selling peace of mind – the kind people will need to adopt more AI tools and splash out on AI-enabled devices, subscriptions and other fare. Without that buy-in, AI companies could struggle to turn a profit. After hundreds of billions of dollars in investments, that’s a concern that’s already starting to spook markets.

Two-thirds of respondents in a September Marist poll said they “believe AI will eliminate more jobs than it creates.” Members of Gen Z and women were especially likely to feel that way. And about four in 10 people polled said they “rarely or never” use AI tools.

...

In real life, though, some experts worry there could be widespread job loss from AI in the years ahead. Rapid data center buildouts are also prompting concerns that rising electricity and water use will push up Americans’ bills.

These were not the first Super Bowl ads promoting AI services. But they were an overwhelming presence Sunday night, prompting some social media backlash.

...

But the AI companies apparently believed that the reported $8 million they spent for 30 second spots was worth it: In the streaming age, the Super Bowl is the one event a year that not only brings in top viewership but includes viewers willing to watch the ads rather than skip past them.

And the AI industry needs the help winning Americans over.

“Americans are much more concerned than excited about the increased use of AI in daily life, with a majority saying they want more control over how AI is used in their lives,” said Pew Research Center, summing up a recent survey on the topic. “Far larger shares say AI will erode than improve people’s ability to think creatively and form meaningful relationships.”

More thoughts from The Internet of Bugs