Sunday Cutdown LVIII (19 January 2025)

Clearing my inbox of stories I want to document

---

1) How COVID will be killing for years to come

NYT: Falling Child Vaccinations — U.S. vaccination rates against once-common childhood diseases are falling.

As recently as five years ago, America was right on target for vaccination rates, with polio, whooping cough, and measles all registering around 95 percent for US kindergarteners.

Enter COVID and the anti-vax movement re-ignited by our government’s ultimately successful “warp speed” effort to develop a new vaccine, and just four years later that trio is all below 93%.

That varies significantly by state:

And yes, I am ashamed to see MN and WI so atop the list.

The breakdown by politics is disheartening, but this is hardly a sharp rural-urban or blue-red divide. Instead, it seems concentrated in schools where parents have come together on this issue — a sort of viral dynamic all its own (self-sorting).

Not an experiment we should want to run:

The decline in vaccination seems likely to have consequences. “Herd immunity,” in which unvaccinated children can’t easily spread measles because others are protected, requires about 92 percent inoculation. The further rates fall, the more likely an outbreak becomes. For children, measles and other once-common childhood diseases can lead to hospitalization and occasionally death.

“It’s trouble waiting to happen,” said Dr. William Schaffner, an infectious disease specialist at Vanderbilt.

I just watched a great new American Experience show on the polio crusade. Gives one a sense of the Before Time not to be re-lived, if at all possible.

Enter Robert F. Kennedy Jr.

2) The myth of nuclear proliferation

Jim Nuttle illustration from “America’s New Map”

VISUAL CAPITALIST: Global Nuclear Warhead Stockpiles (1945-2024)

Kind of looks like India, yes?

At first, America ruled. Then the Soviets did. Then arms control and the end of the Cold War contracted those totals considerably.

The great proliferation that we’ve all heard so much about? That requires a bit of magnification.

Here’s how we did it in the book:

The “All Others” get lost in the wash, even as their going nuclear similarly served the world’s purposes in capping interstate wars among fierce rivals.

From America’s New Map:

Following the first and only uses of nuclear weapons by the United States in 1945, six additional nuclear powers (the Union of Soviet Socialist Republics [USSR], United Kingdom [UK], France, China, Israel, and India) came into being over the next three decades. Since India joined the so-called nuclear club in 1974, there have been two new members reluctantly admitted: Pakistan (1990) and North Korea (2006). That record describes a profound slowing of nuclear proliferation over eight decades—from one power every four years to one every twenty. Despite those persistently scary predictions, we remain stuck in the single digits at nine club members.

What does that anything-but-uncontrollable dynamic tell us?

Nuclear arsenals only make sense for certain great powers—namely, those in antagonistic rivalries with other great powers capable of large-scale conventional and nuclear warfare. So, once America invented nukes, the Soviet Union needed them in reply. That pushed France and the UK to counter the USSR within Europe. China needed nukes against both the United States and the Soviets. Israel needed them to protect itself against the combined forces of the Arab world. India needed them against both China and its archenemy, Pakistan. Pakistan followed suit vis-à-vis India. Finally, North Korea made the reach out of fear of US-backed South Korea’s significant military capability— and our nuclear arsenal.

There is your history of “out of control” proliferation, which has resulted in zero strategic wars among that constellation of powers once their individual nuclear status was achieved and recognized by the rest.

Nuclear weapons killed great power-warfare — eight decades and counting. Their invention and their demonstrated use (by Truman) were — and remain — very good things.

You could call it a strategic vaccination of sorts.

3) Putin’s creative war accounting

MOSCOW TIMES: Russia's Hidden War Debt Creates a Looming Credit Crisis

The trick is a basic one:

Moscow has been quietly pursuing a two-pronged strategy to finance its escalating war costs. In addition to the publicly scrutinized defense budget, it has set up a system of state-directed, off-budget soft loans where the Kremlin badgers banks into making easy credits to defense-sector companies to unofficially fund its war machine.

So, Putin is basically putting much of the war’s costs on his bank “credit cards,” in effect forcing banks to load up on debt that one presumes is unsustainable as a pure one-off (i.e., not connecting to any wealth creation or productivity or profit per se but just funding a sinkhole expenditure).

So, an “off-budget” scheme that says, we’re not really spending government money if it’s not on the government books.

A reckoning is coming:

Now analysts warn that the amount of accumulated debt may begin to unravel, posing risks to Russia's financial stability. By maintaining its official defense budget at ostensibly sustainable levels, the Finance Ministry has misled observers and fooled them into significantly underestimating the strain the so-called special military operation is having on the corporate and banking sectors. The off-budget funding scheme is only fueling more inflation, pushing up interest rates, and weakening Russia's monetary transmission mechanism.

All this structural stress back home should make Putin suitably pliable for Trump’s desired fast-track resolution, if our new/old POTUS plays his cards right (a big if).

4) Explainable AI (XAI) badly needed

TECH CRUNCH: OpenAI’s AI reasoning model ‘thinks’ in Chinese sometimes and no one really knows why

Sometimes, it seems, OpenAI’s “reasoning model” starts speaking in tongues … other than English.

As the title says, “nobody really knows why.”

Abnormalities like this and “halluncinations” in general will keep presenting themselves, one assumes, as technologists keep pushing the boundaries of AI, always asking for more. But these bugs/features will remain the great brake — in my opinion — on AI’s spread throughout society, the economy, and political system, all of which will weigh in with judgments and grievances and complaints and declared liabilities — particularly in terms of bias.

My once-and-future colleague Stephen F. DeAngelis, CEO and Founder of Enterra Solutions and Massive Dynamics, describes the emergence of XAI as the journey from the black box to the glass box. In Steve’s description, poor user adoption is the telltale sign: you don’t adopt what you don’t trust and/or fully understand.

Steve has forgotten more about AI than I’’ ever know, which makes him an interesting read if you are so inclined.

I wrote extensively about my collaborations with Steve in my 2009 book, Great Powers: America and the World After Bush. Collaborating with both the Pentagon and the Kurdish Regional Government of Iraq, Steve and I worked on a concept called Development-in-a-Box — sort of a quick-and-dirty connectivity strategy for postwar and post-disaster situations.

Anyway, back to the story: what are the theories regarding why this is happening — this language jumping?

Several on X, including Hugging Face CEO Clément Delangue, alluded to the fact that reasoning models like o1 are trained on datasets containing a lot of Chinese characters. Ted Xiao, a researcher at Google DeepMind, claimed that companies including OpenAI use third-party Chinese data labeling services, and that o1 switching to Chinese is an example of “Chinese linguistic influence on reasoning.”

“[Labs like] OpenAI and Anthropic utilize [third-party] data labeling services for PhD-level reasoning data for science, math, and coding,” Xiao wrote in a post on X. “[F]or expert labor availability and cost reasons, many of these data providers are based in China.”

Labels, also known as tags or annotations, help models understand and interpret data during the training process. For example, labels to train an image recognition model might take the form of markings around objects or captions referring to each person, place, or object depicted in an image.

Studies have shown that biased labels can produce biased models. For example, the average annotator is more likely to label phrases in African-American Vernacular English (AAVE), the informal grammar used by some Black Americans, as toxic, leading AI toxicity detectors trained on the labels to see AAVE as disproportionately toxic.

Other experts don’t buy the o1 Chinese data labeling hypothesis, however. They point out that o1 is just as likely to switch to Hindi, Thai, or a language other than Chinese while teasing out a solution.

Rather, these experts say, o1 and other reasoning models might simply be using languages they find most efficient to achieve an objective (or hallucinating).

Clear enough?

But here it gets truly interesting:

“The model doesn’t know what language is, or that languages are different,” Matthew Guzdial, an AI researcher and assistant professor at the University of Alberta, told TechCrunch. “It’s all just text to it.”

I don’t know whether to laugh or gasp at that statement: it’s either supremely naive about the complexity and contextual nature of language or these guys have really cracked some universal translator code — or both.

But this is why all these explanations are labeled as theories.

Indeed, models don’t directly process words. They use tokens instead. Tokens can be words, such as “fantastic.” Or they can be syllables, like “fan,” “tas,” and “tic.” Or they can even be individual characters in words — e.g. “f,” “a,” “n,” “t,” “a,” “s,” “t,” “i,” “c.”

Like labeling, tokens can introduce biases. For example, many word-to-token translators assume a space in a sentence denotes a new word, despite the fact that not all languages use spaces to separate words.

So, maybe AI is code switching just like humans do:

Tiezhen Wang, a software engineer at AI startup Hugging Face, agrees with Guzdial that reasoning models’ language inconsistencies may be explained by associations the models made during training.

“By embracing every linguistic nuance, we expand the model’s worldview and allow it to learn from the full spectrum of human knowledge,” [Hugging Face software engineer Tiezhen] Wang wrote in a post on X. “For example, I prefer doing math in Chinese because each digit is just one syllable, which makes calculations crisp and efficient. But when it comes to topics like unconscious bias, I automatically switch to English, mainly because that’s where I first learned and absorbed those ideas.”

The ability to switch languages so as to optimize a particular effort … that is impressive, as in, beyond normal human capacity (outside of a few people who can master more than 2-3 languages, I would argue (as someone who has studied four non-English languages in French, Russia, Romanian, and German). I mean, I love rap in French, I like history in German, I like poetry in Russian, and I like general conversation in Romanian.

Can I effortlessly and at high speed code-switch between them? On a very limited basis.

If I could do that all the time? Say, in 30-40 languages? That would be wow.

So, I guess the uptake here is: this is probably a good problem to be encountering on the road to greater capabilities.

Still, the knowing of why and how AI works remains crucial. That gap in understanding can’t grow too large, otherwise the social and economic and political backlash when things go wrong (and things will go wrong) will be debilitating.

I look forward to working with Steve on such conundrums of explanation.

5) Unintended consequences

WAPO: Surge in Americans getting sterilizations given states’ abortion laws

The word:

Research shows a significant increase in vasectomies and tubal ligations in the months just before and after the Supreme Court ruled in Dobbs v. Jackson Women’s Health Organization, the 2022 decision that ended a constitutional right to abortion after nearly half a century.

Experts predicting a lot more of the same across a second Trump administration.

Planned Parenthood of Northern New England has 15 locations that log over 54,000 patient visits yearly. After Donald Trump’s reelection in November, it saw a 368 percent increase in consultations for vasectomies, chief executive Nicole Clegg said.

Hmm, that’s one way to opt out.

The kicker: it’s the young (18-30yo) doing this.

I, for one, am surprised:

“We know that young people, including young adults, are sensitive to major changes like the Dobbs decision, so it’s not totally surprising to me that we would see some type of shift,” said Julia Strasser, author of the latest study and an assistant research professor in the Department of Health Policy and Management at George Washington University. “I do think the magnitude of the change was somewhat surprising.”

It’s just the whole preemptive nature of this step:

With nearly 20 states now banning or significantly limiting abortion, many Americans are also worried about future access to birth control. Health experts predict the number of people looking for a permanent solution will continue to grow if restrictions expand further.

Docs are flabbergasted at the rising demand for these specialized services.

“In 43 years of doing this, I’ve seen maybe three or four couples who never had children or men who never had children,” he said. “But now we’re seeing a huge increase in the number of couples who think the world is too awful a place to bring children into.”

This doesn’t feel like America becoming great again, does it?

It almost feels like America is losing its brand appeal at home.

And we know what happens to stale brands …

6) A more practical example of the need for XAI

WAPO: Arrested by AI: Police ignore standards after facial recognition matches

Apparently they do all look alike to AI that’s been optimized for White faces.

Dangerous, scary stuff:

A Washington Post investigation into police use of facial recognition software found that law enforcement agencies across the nation are using the artificial intelligence tools in a way they were never intended to be used: as a shortcut to finding and arresting suspects without other evidence.

Most police departments are not required to report that they use facial recognition, and few keep records of their use of the technology. The Post reviewed documents from 23 police departments where detailed records about facial recognition use are available and found that 15 departments spanning 12 states arrested suspects identified through AI matches without any independent evidence connecting them to the crime — in most cases contradicting their own internal policies requiring officers to corroborate all leads found through AI.

First off there is the sheer laziness/under-resourced problem that drives some police to seek any and all shortcuts.

But, beyond that is the problem of these “100 percent matches” that simply are not true, thus begging for explanations of how this technology works (or doesn’t) and why it should be trusted in such weighty matters as law enforcement.

Finally, there’s the institutional racism of it all: technologies perfected by Whites for Whites and thus dangerously underperform with non-Whites, and, when they do, those failures are more easily swept under bureaucratic rugs.

In the WAPO story about 8 wrongful arrests, it is noted that 7 of the 8 cases involved African Americans.

This is an increasingly well-studied issue:

Federal testing in 2019 showed that Asian and Black people were up to 100 times as likely to be misidentified by some software as White men, potentially because the photos used to train some of the algorithms were initially skewed toward White men.

Now, the scariest part: the lack of transparency on usage.

These examples of questionable police work — gleaned through The Post’s analysis of rarely seen internal software records, arrest reports, court records, and interviews with police, prosecutors and defense lawyers — are probably a small sample of the problem. The total number of false arrests fueled by AI matches is impossible to know, because police and prosecutors rarely tell the public when they have used these tools and, in all but seven states, no laws explicitly require it to be disclosed.

As all this surveillance tech, abetted by AI, spreads throughout our society, such issues are going to skyrocket in their severity and frequency and blowback triggered, making XAI all the more politically urgent an issue.

7) I’m Thomas P.M. Barnett and I approved of this message

WAPO: Rubio details what Trump’s ‘America First’ foreign policy will entail

The guy could be good, if he can last, if only because he thinks like this:

“When they write the book about the 21st century, there’s going to be some chapters in there about [Russian President Vladimir] Putin. There’s going to be some chapters in there about some of these other places,” Rubio said. “But the bulk of that book about the 21st century will be not just about China but about the relationship between China and the United States, and what direction it went.”

Speaking strictly in state-level terms, it’s hard to argue against that logic, but I do believe that this century is the century in which global issues rule for the first time in human history.

Still, a good bit and a good sign.

8) Keep it simple, stupid! (KISS)

Pretty big for unmanned surface vehicle?

BREAKING DEFENSE: Navy admiral ‘skeptical’ of Large USV’s future amid congressional scrutiny

Story about Navy’s surface warfare requirements director speaking at surface vessel symposium and pushing the KISS principle:

A senior Navy officer overseeing the service’s investments into unmanned surface vehicles says he’s “skeptical” of fielding the Large Unmanned Surface Vessel.

“Instead of different large and medium designs, we need one craft that is affordable, non-exquisite, and can come off multiple production lines in an identical manner and go towards one of two payloads — either the envisioned magazine payload of the Large USV or the envisioned ISR-related Medium USV payloads,” Rear Adm. William Daly, the Navy’s surface warfare requirements director, told attendees this week at the Surface Navy Association’s annual symposium.

“We can build this craft in numbers at many shipyards. Designs already exist. We must not over-spec this,” he continued.

One assumes he’s talking about the 194-foot vehicle (USV Mariner) in photo above.

What got me was his description of wanting to keep it simple: affordable, non-exquisite (in other words, lay off the bells and whistles), replicable across various production lines, and built for just two payloads — all stuff that makes sense.

We shall see how much the Department of Navy schemes to junk this thing up with every dodad under the sun (meaning, to “over-spec” or over specify).

Conversely, I loved Daly’s criticism of the LUSV:

“The Large USV has a great purpose, but it has it has walked that path towards exquisite, expensive [and] unpalatable,” Daly told reporters on the sidelines of SNA. “I’m skeptical about that landing in the fleet.”

Right up there with my mantra of the many, the cheap, the disposable, and the unmanned!

If you want to read the late, great Allen Drury, check out his novel Pentagon because it’s all about a ship design that goes nuts and, in doing so, magnificently complicates things so much that it ends up entirely missing the actual war.

I know, funny right?

Old White guys sitting around a conference table. It don’t get any sexier than that!

Okay, there's a guy macking on a lady in the upper left corner.

Drury, who was long rumored to be gay, was noted for his employment of a gay character in his classic Advise and Consent (great movie by the always audacious Otto Preminger).

From Perplexity:

Homosexuality in Drury's Work

Drury's representation of homosexuality was progressive for its time. The character Brigham "Brig" Anderson, a senator with a secret homosexual past, is often cited as one of the first sympathetic gay characters in mainstream American fiction. His plight reflects the societal pressures and stigma surrounding homosexuality during the 1950s and 60s. The novel's portrayal of Brig's tragic suicide due to blackmail over his sexual orientation resonated with many readers and highlighted the severe consequences of societal homophobia. Drury himself was rumored to be gay, a fact that was acknowledged in private conversations among political figures such as Richard Nixon. Despite this speculation, Drury maintained a level of discretion about his personal life throughout his career. He lived in Tiburon, California, until his death in 1998

The irony?

After the Cold War ended, the DoD pursued an investigation to see if anybody gay had ever been turned solely on that basis by an enemy and the answer was no — not a single instance.

The real reason time and again? Money, which is totally American, is it not? That’s why, when you’re investigated for a clearance, the Watchmen ask your references if it has ever seemed like you’ve lived beyond your means — another totally American habit, da?

But the Lavender Scare image was a Cold War staple in movies and books: the evil, weak gay man who — of course — betrays his country because he lacks all morals!

So I guess we credit Drury for his portrayal, even if the only way he could get it in was through this socially acceptable negative stereotype with no basis in reality.

Wow, this entry started one way and then ended up in an entirely different direction: from unmanned to gay-manned!

9) Paging Star Trek IV: The Voyage Home!

YAHOO NEWS: Scientists sound alarm after documenting unprecedented behavior in humpback whales — here's why this matters

As I always note, climate change presents the following menu to all species: adapt, move, or die.

The trigger:

Each year, humpback whales make a lengthy migration from tropical breeding grounds to colder feeding grounds. But researchers recently logged the migration of one whale who had traveled far beyond the species' normal migration path, prompting officials to assess what this behavior might signal for humpback populations.

For regularly migrating species, the solution would seem to be both adapt behavior and redefine one’s sense of acceptable “move.”

The early analysis of this rogue whale suggests they’re traveling that much farther in order to reach sufficient krill concentrations. Krill are migrating with warming oceans, and that may put them increasingly out of reach for whales unwilling to go that extra mile, so to speak.

If [warmer ocean temperatures] is the main driver behind the whale's far-reaching migration, researchers warn that this could spell out a future where the food supply grows increasingly unstable, leading whales to die of exhaustion as they look for available sources. This is one example of how a warming planet is leading to a loss in biodiversity — which, in turn, disrupts the ability of species and ecosystems to function and survive normally.

Adaptation at high speed is, of course, dangerous for those so engaged.

Similar research has been conducted on other behavioral adaptations, some of which seem possibly beneficial and others which seem downright harmful. For example, certain species of insects produce duller colors to stay cool in summer, which has led to a drop in breeding rates.

On this subject of how climate change is stressing species the world over, we’re still in the phase of not even knowing how much there is that we don’t know.

If that ominous Star Trek probe ever does show up, we can’t say we weren’t warned.

10) A speech that will be remembered

WAPO: Biden warns of the rise of a new American ‘oligarchy’

The conflicting trajectories/desires here are clear enough:

• We’re in a race with China to network the world, earn all that Big Data, and achieve AI supremacy (if that’s even possible or makes sense) … oh, and win the perceived ideological race between authoritarianism and democracy (can’t forget that!).

• The Tech Bros are true oligarchs/robber barons and we’re living through a rerun of the Gilded Age in which the rest of us mere mortals both admire these titans and simultaneously want them terminated with extreme prejudice.

• But we need Big Tech and the Tech Bros to thrive so as to win that race with China.

• But there’s only so long that our society can remain angry and populist (since 2008) without serious progress — as in, progressivism unfolding once again as it did from McKinley through FDR.

• Something thus has to give.

So, yeah, we’ll look back at Biden’s warning like we do Ike’s bit about the military industrial complex.

And yes, we’ll eventually get around to breaking and busting-up these quasi/de facto monopolies and trusts … because eventually that’s best for all of us.

The question, of course, is one of timing, but eventually Biden will be viewed as suitably prophetic.

Just one note: it’s okay to both love and hate the Tech Bros, because they will inevitably wear out their welcome on both ends of the political spectrum (already with Dems, soon enough with GOP). That is both inevitable and natural and fine.

Why? Because we know that monopolies are great for building networks, but freer markets are so much better at running them.

One of the oldest lessons in our book.

11) Just once I’d like to read …

YAHOO NEWS: The end of the world as we know it? Theorist warns humanity is teetering between collapse and advancement

Just once I’d like to read about the strategic thinker who says, Actually, things are going okay and reasonable rates of progress appear in the offing. It’ll be scary here and there but we’ll muddle through as we always do.

Instead, we ALWAYS get this binary choice between success (techno-utopian — always revealed in Hollywood movies as hiding some godawful secret) and failure (cue the White House-exploding CGI shot… Why? Because that never gets old!).

“Independence” Day … get it?

Why do we accept such extremist analysis (here, our theorist is doubling down by claiming both ends of the spectrum)? That approach makes you, the theorist/futurist, that much more important before the public. Such experts want to warn but not really explain in a broadly framed/balanced/contextualized whole because, if you do really explain, then people can figure out the true reality on their own and walk away feeling better and more confident about the future.

But who wants that?

12) On the road again

VISUAL CAPITALIST: Visualizing the World’s Busiest Migration Corridors

If I understand the color coding correctly, then I would predict that this chart will be far red-brownish in tone in the future — as in, displaced by climate.

For now, the US is depicted as all-green inbound… I guess, signifying that everyone coming here is coming for economic opportunity reasons.

But that’s just the other side of the displaced coin, is it not?

I’m thinking we’ll be debating these categories and definition intensely in the years ahead as climate displaces people off the land across the Global South, pushing them into cities where their economic needs cannot be met, and then we’ll call them economic migrants once they head north for a better future.

I know for a fact that this is happening all the time right now all over my Middle Earth (30 degrees north and south of the equator). We’re just not seeing the causal chain in whole.

But such sight is coming.

---

---