Thursday’s Tabs
Steven L. Taylor
·
Thursday, January 22, 2026
·
41 comments
About Steven L. Taylor
Steven L. Taylor is a Professor Emeritus of Political Science and former College of Arts and Sciences Dean. His main areas of expertise include parties, elections, and the institutional design of democracies. His most recent book is the co-authored
A Different Democracy: American Government in a 31-Country Perspective. He earned his Ph.D. from the University of Texas and his BA from the University of California, Irvine. He has been blogging since 2003 (originally at the now defunct Poliblog).
Follow Steven on
Twitter and/or
BlueSky.
We are heading back to the US at the end of the month. our visas are hung up in red tape and our Schengen time is running low. So, back to the States for February and March, during which time my wife has book tour for a couple weeks, then back to Portugal for most of April, and back the States for May, and back to Portugal for June and July.
I have very mixed feeling about going back to the US. We’re taking the first week of February to survey the housing market in LA. And there are one or two meetings regarding possible TV stuff. Then back to Vegas, where at least Amazon works and the pizza is better.
Query: I am setting up a substack as a place to write about my/our books, especially around the 25th anniversary of Animorphs and the possible TV series. I am helpless with anything remotely technological. I need help getting things up and running, and possibly more long-term help – I literally don’t know. I wonder if anyone here is in that sort of work. This would be a paying gig, I’m not looking for freebees.
Here’s my email with spaces that should be removed and a symbol that should be inserted. michael reynolds grant at g mail.
Shouldn’t this be “forum”? Header says “Tabs” but there are no tabs.
Why AI is a bubble:
“Link”
“Link”
“Link”
This really is as stupid as tulips. The difference is the believers are supposedly smart giant technology corporations and their plutocrat major shareholders.
Was just noodling around the internet and this AI summary popped up:
Apparently, this is the reference:
US citizens seek millions in damages after violent ICE arrests
I wonder what the number is now.
Your tax dollars at work.
My legal brain fart of the day: If a claim against the US government is not settled in 90 days, the claim is presumed valid and paid out.
I don’t know what is going on politically in the rest of the country but in Texas Islamaphobia and Sharia Law screeching is in full swing. The Republican bigotry is pushed by Gov Abbot, Sen Cornyn, Republican Senate candidates. It truly is disgusting.
North Texas school district cancels Islamic Games for alleged ties to group governor calls terroristic
My own school district:
Gov. Greg Abbott demands Cy-Fair ISD cancel Islamic Games event: ‘Not welcome in Texas’
@Michael Reynolds: Mind if I pass this question on to my Discord group? There’s good variety of tech skills and employment needs there, including several folks who grew up with the Animorphs. (Probably would need more specific info about what you’re looking to do near-term and long-term with it.)
New Englanders can chime in.
Has Trump, ICE ensured Susan Collins’ defeat?
Trump administration starts immigration operation in Maine
@Michael Reynolds:
I’m passing this along to my wife. Her position was eliminated in October and since then she’s been freelancing on stuff just like this.
@ptfe:
I’d appreciate that.
ETA: Ditto, Neil.
My daily read.
Heather Cox Richardson’s Letters from an American
@Scott:
Unfortunately, that brain fart seems to be more myth than fact. A claimant has to notify the government in writing of a claim, and the feds then have 6 months to accept or deny the claim. If the feds deny the claim a lawsuit can be filed under the Federal Tort Claims Act (FTCA). If the feds don’t respond to the notice of a claim, the claim is deemed denied, and then you can file your FTCA lawsuit. Under FTCA, the law of the state where the incident occurred is applied. E.g., Renee Good’s civil case would be decided under Minnesota law as applied by a Federal district court judge.
That being said, DHS will certainly be forking out millions of dollars for the damages created by its poorly trained, poorly supervised and overzealous goons.
@Charley in Cleveland:
I’m reposting this AP link that shows ICE agents are wrongly rather than only poorly trained.
TL;DR: they’re told they can break into private homes without a warrant.
To the surprise of absolutely no one: Venezuela’s Delcy Rodríguez assured US of cooperation before Maduro’s capture
TL;DR: Machado gave up her Nobel Prize medal to someone who’d already stabbed her in the back.
@Kathy: *sigh* Reich Chancellor Miller has convinced the 79 year old dotard that he can ignore both Article 1 and the Bill of Rights. Miller is the de facto head of DHS, so when he screams at guys like Bovino to ramp up the nasty, guys like Bovino are happy to oblige. When federal judges cite the Constitution, Miller goes on Fox and says they are “rogue leftists,” and Pastor Mike Johnson says impeachment is on the table for judges who dare to say no to Trump administration unconstitutional power grabs. And SCOTUS sits on its hands. It’s 5 o’clock everywhere while the reign of error continues, so please pass the bourbon.
@Charley in Cleveland: “That being said, DHS will certainly be forking out millions of dollars for the damages created by its poorly trained, poorly supervised and overzealous goons.”
I’m sure that many in the Einsatzgruppen were poorly trained, but that was not the problem
We’re supposed to get 1 to 2 feet of snow here in Mass. on Sunday and Monday. Yikes.
“Public Notice”
…
I don’t mind no Tabs, but there’s no cat pic!!
Tom Sullivan at Digby’s place posts a Xit from a Danish economist,
I would note that Congress, the Court, and much of the media are in Republican hands. But this Dane sees little pushback from the American public.
Sullivan points out, “But he’s mistaken to think that “there is no real action from the US public … about this insanity.” There is just not enough of it.” and adds that,
This has long been a pet peeve of mine. Protests against Trump seem to get very little coverage, even that Minnesotans are protesting outside in subzero temps. It seemed in 2008 and 09 like any dozen Tea Partiers could get press attention and draw an NYT Cletus safari. Maybe we need to get drums and tricorne hats. But it wouldn’t work in a press “curated for him by the people in Trump’s billionaire class to minimize the public seeing pushback”.
A lot of the chattering class decided that Biden having a terrible debate and a Black woman losing after a six month campaign signaled a big rightward vibe shift. If enough of us take to the streets maybe they’ll question themselves.
@charontwo: I’ve been reading a few smart economists/business analysts about this. Here’s a couple of points:
1. Established firms such as Google, Apple, Microsoft, Facebook are determined to not be disrupted by AI technologies. So they are investing heavily, more than would otherwise be wise, just to prevent that.
2. What is coming next is less like a bursting bubble, and more like the wildfire that burns all the undergrowth and smaller trees, providing fertilizer for the more healthy things. In this case “fertilizer” amounts to “people who have some idea how to do this” and also “server farms already set up and going”. Maybe a few more organizational ideas.
3. We will see a big pullback, I am sure. AND I have seen enough really useful applications of large language models to conclude that it isn’t ever going away. I hope, I really hope, that some of the more crap will go away.
Just yesterday, I watched most of a YouTube video that appeared to be a voice track of a lecture by Feynman until he said something that was anachronistic. Then I looked and yes, they noted that it wasn’t really Feynman, but “in the spirit” of Feynman with an AI-generated voice reading a script that they had written.
Bugger that.
@Jay L. Gischer:
I’d like to know of any. In particular something that would 1) be helpful for regular people in everyday life, and 2) not require more work than doing it yourself.
@gVOR10:
Reagrding Christensens’s
Thge thing, is most people outside a country tend to regard said country as a “balck box” when it comes to foreign policy.
They mostly neither know, nor care, about internal debates or diffrent admisnitrtions or whatever.
All that really matters is the output.
So Trump’s policies, to most non-Americans = US policy; and the administations’s inanities = the American view.
This is why the damage done to perceptions of the US, and changed politics of relations with the US, will outlast Trump, even if a Democrat won the next presidential election. There will always be a increased sense of caution, and hedging of bets, going forward.
And these things have their own dynamics.
As Christensen might have said: electing Trump once might have been considered an aberration.
His getting elected twice, and the complicity and cowardice of the Republican Party establishment, says something important about the nature of US politics, that other countries simply cannot overlook.
@Scott:
I’ve been wondering the same thing. If ICE behaves as they have in mpls, then it’ll wound her. Maine is an old state by average age and here like elsewhere, home health aids and other healthcare aids are significantly immigrants. Those folks disappearing will piss a lot of elders off and those are voters that the Senator from Concern relies on.
@charontwo:
@Jay L. Gischer:
This distinction between LLM’s and AI approaches based on “world models” and “reasoning” is vital.
Hence the current thing for “we need to enhance LLM’s with reality models and reasoning”
The issue being, the LLM’s were essentially an attempt to “brute force” a way past the problem that we can’t seem to actually construct reasoning or reality models that actually work.
And the LLM “shortcut” seems to be turning out to be a dead-end.
Maybe we can use neural-networks to generate a solution.
After all, it only took evolution a few billion years to get there.
The generalised versions of LLM types, pattern recognition models, seem to have massive potential for data sorting and matching. Very useful in medical diagnostics, chemistry and biochemistry, etc etc.
But expecting “AGI” to pop out of this particular hat seems rather optimistic.
Back to the drawing board, folks.
The worst of the worst.
Today America is safe from dangerous 5 year old boys coming home from school.
America can no longer be regarded as a civilized country.
@Kathy:
And it’s going to get a whole lot worse before it gets better. They’re eventually going to shoot a kid point blank in the face.
@Jay L. Gischer:
I seriously doubt this and it comes across as motivated reasoning (not your’s, the “experts”). How often does something that inflates a bubble like this calmly burn away only the bad undergrowth.
First, it’s quite obvious that, at a minimum, OpenAI, Microsoft, Nvidia, and Oracle are cooking the books on a scale that will make Enron seem quaint. Personally, I’ve set my expectations at them blowing out the economy at least as bad as 08. We’ve learned nothing and on this I fully and honestly blame Obama.
Second, it’s becoming obvious that not only are these AI idiots starting to get the flop sweats, but they’re all cooking their own brains on rightwing bullshit. They’ve ALL bought into the idea that they are the apex of history and the fact that they’re rich assholes means they’re touched by god.
The second point is going to make the first point worse. There’s absolutely no quiet backdown for these idiots. I think it’s better to assume they’ve all inflated their egos like Musk. The only difference is they’ve managed to keep their drug use manageable.
@Kathy: Well, I’m not sure what constitutes “regular people”.
However, I have a friend who was a real-estate appraiser. Appraisers write reports. There is the important stuff, which is going to properties, taking photos, looking at comparables, and so on.
Then there is the report. About 10 percent of the report is important, the rest is pretty much boilerplate. An AI assistant could write that boilerplate, and then the appraiser does an editing pass. This is a time-saver. It’s like having a junior assistant.
Likewise in programming, one can train an AI to be a junior assistant and do all the repetitive boring stuff.
Now the problem this poses is in training and development. Junior programmers (humans) learned to be senior programmers by doing this stuff and if we use AI, how are we going to develop more senior programmers. But maybe we will figure that out, and the job will shift around.
But maybe you don’t think of programmers as ‘normal people’. I’m not sure I do, and I is one. 😉
I don’t see LLMs as eliminating the need for programmers. Seeing a non-programmer use an AI to create a program provokes wild laughter among programmers. Also, it doesn’t work beyond the most primitive things. Because, remember, the AI doesn’t actually know anything beyond which word comes after which. They are like a precocious child.
@Jay L. Gischer:
As a very amatuerish spreadsheet wrangler, I’ve found ChatGPT sometimes useful for working out the gnarlier bits of VBA syntax for a sub-routine, so long as you know what that task is supposed to do in the first place and give it some pseudo-code to chew on.
But even then, it sometimes comes up with stuff even I can see is either wrong or inefficient.
It’s bit like the old Excel contextual help on steroids.
That is, before Microsoft junked that rather useful on-machine approach, and went for “online help”, which is frequently utterly useless.
@Jay L. Gischer:
Except, that’s the entire point of AI/LLMs. The thing that Silicon Valley seems to despise the most is its senior programmers and managers. They want workers that are too scared to fight back when the owners abuse them.
@Jay L. Gischer:
I tried using an LLM to draft emails. It didn’t work out well. I’ll discount it sounded nothing like me, because I don’t expect the LLM to know my style. The result might not even have been the LLM’s fault. there are a bunch of assumptions for emails within the company for, say, requesting pricing a list of products with a maximum divergence in package/can sizes of +/-10%, and specific brand requirements. these are things you can’t expect the LLM to know.
So I got two results: 1) the prompt was pretty much the email, 2) it got things so wrong that I spent more time correcting the draft than I would have just writing the email myself.
I’ve not tried things I could, like asking it to re-write a scene. I want my works to be mine.
I tried some more involved methods than mere prompts for work related stuff. It seemed to be doing fine, but there was a need to check its work. For instance, asking it to list the technical requirements in a request for proposals. Checking that involves going through the request and identifying all the technical requirements, at which point you’ve done the work you asked the LLM to do, and then comparing them with what the LLM came up with.
Whether it got it right is irrelevant, if I have to do the work myself anyway.
Other work related experiments were even worse. When I asked it to make the format for an electronic proposal in Excel based on a word file, I estimate it got the format about 90% right. I also estimate that fixing the LLM’s attempt would have required more work than doing the format myself.
So, color me underwhelmed.
Menwhile in the Med:
Did the Grinch steal Christmas? Voila, Macron steals The Grinch! lol
France demonstrates that they also can interdict Russian “shadow fleet” tankers.
@Beth: There are people in management who are exactly how you describe.
AND, there are other people here who are not at all like that. They work quietly, and don’t get the press, though.
In 1981 or so, the then chairman of the Stanford CS Department, a guy who worked in AI, gave a talk where he claimed that in 5 years, we wouldn’t need programmers because AI.
It was hype then (the field is notorious for hype). It is still hype now. I mean, some fool probably believes it, but no. Look at what JohnSF says above. I think with a lot more training, an LLM assistant would be quite useful to him. But it won’t eliminate the need to have a person doing what he does.
@Kathy:
A friend of mine has been using one for coding, and reports that it’s like a junior engineer that works very quickly, requires a lot of supervision and never learns to be more independent. Sounds like hell, but for a lot of applications, that’s all you need.
I expect they will be excellent at denying preauthorizations at health insurance companies, providing customer service — really any activity where the goal is just to add friction, as the corporation benefits more from things not happening than happening. Corporations are people too, my friend.
Similarly, they are excellent at disrupting online organizing with a flood of disinformation.
And for home use, there are the various chatbots. A quick survey shows that you can chat with a desperate homeless elf with a two year old daughter, your broken friend (or toy) with low self-esteem, a homeless girl who has hit rock bottom, or your elder sister who has found your ai chat history, your absentee father, or one of the pilots from Voltron*.
Many, many fine uses for LLMs. And then there are the image generators!
*: I skimmed the list of popular characters from a few chat bot sites, and I don’t feel good anymore…
@gVOR10:
Having long and vocally been in the ‘the problem is the American electorate’ camp (especially those left and right still finding excuses not to vote for the opposition), this American agrees.
But this American also sees little pushback from Europeans. E.g., they enable Trump by refusing Ukraine adequate financial and military support. Europeans have had four years to ramp up weapons and energy production accordingly. And yet.
Until the EU recognizes its fecklessness towards Putin emboldens thugs like Trump, Xi, Khamenei, Netanyahu, et al then Europeans are part of the problem too. Tho, yes, not as bad as we Americans — racists still too weak to call Trump’s Jan 6 terror attack “a terror attack.” Instead we coddle the (white) working class with undeserved romanticizing and scold educated woke whites. Backwards af
@Beth:
We may be seeing a mix of share value supremacy and a surplus of cash reserves in the big tech companies. Share value is kind of not entirely real, and it can be manipulated in many ways. So you get things like OpenAI buying NVidia chips with Nvidia’s money. It’s also clear that Open AI is burning money as fast as it gets “invested.”
I can understand, to a point, Obama’s actions to bail out the banks and other financial companies, but there was no excuse for letting the executives responsible for the mess off the hook. That was the big mistake.
I won’t claim companies decide to do whatever they want, while knowing they’ll be bailed out if they mess up. But in taking huge risks, this is how it feels they’re acting now.
Also, the money the US government is paying AI companies constitutes a preemptive bailout. I don’t think there’s been enough money involved for that, but things ight just be starting. There was the announcement last year of a half trillion investment as well…
Since all things superficial rule this American admin, one must acknowledge that Mark Carney is slim trim handsome and intelligent. Trump is dumpy wheezy orange jowly and stupid.
Stating the obvious.
This again: Is listening to an audiobook as good as reading?
I’ll tell you one thing: it’s a lot better than not reading. Given my work schedule, I’d have to forego some entertainment in order to read, and I’d be lucky to average one book every ten days or so. As is, listening while driving and cooking, I managed over 55 books (including Great Courses lecture series) in 2024, the last year I kept count, mostly non-fiction.
@Jay L. Gischer:
On Beth’s comment, consider that a lot of employees across all industries are underpaid relative to inflation, and economic and productivity growth. Few of these are worth replacing with AI.
Programmers, on the other hand, are highly paid. The math does itself.
Executives, too, are highly paid. If I were one, I’d consider this fact before I decided to replace anyone with an LLM…
Random thoughts:
I’m still hopeful I can make the trip to Spain in 2027 to see the Eclipse, but I don’t think I’ll be taking the train.
Granted big accidents tend to put a smaller spotlight on other incidents. but the collapse of a retaining wall and crashing with a crane are not minor incidents.
The USSR lasted for about 74 years. During most of that time, material conditions for the vast majority of people were bad, and social and political conditions were terrible.
America, and to some extent the rest of the western world, has been under supply side, trickle down economics for 46 years. During that time material conditions have gotten much worse, and now social and political conditions are literally falling off a cliff.
How long will you put up with this shit?
Last: if there’s an infinite number of parallel/alternate universes that can vary in any and every detail, like say natural laws, mass, and size to name a few, would it be possible an infinite number of universes* have a crazy high magnitude strong nuclear force, the mass of 20 stars, and is smaller than a subatomic particle? It seems to me it’s inevitable.
* subsets of an infinite set are themselves infinite. Proof: the set of real numbers is infinite. the subset of negative real numbers is infinite, as is the subset of positive real numbers.
This would also mean there are an infinite number of parallel universes identical to our own down to the last detail. I find this thought depressing and a strong argument against infinities.
@charontwo:
More on AI:
“Gary Marcus”
…
@gVOR10:
Something like what’s been rolling around in my head of late: The US public, through a combination of apathy and poor-education induced partisan-propaganda assisted Dunning-Kruger, have become a population unsuited for a republic. Congress? Most American don’t even know their congressman’s name. They only remember the letter behind the name. Haven’t a clue what their congress-critter has been doing, what he or she has voted for or against, let alone understand the whys for either, which requires reading beyond the headlines.
They now seek “something be done”. As always, the poor want to be saved from their condition and the rich want to be protected from the poor, but neither has a clear idea how that should be done, particularly the former.
With Congress subject to endless partisan dead-lock, a disorder common with committees, it’s only a matter of time before such a population seeks to have one person with the power to make things happen running things. The worse things are or are perceived to be, the sooner they seek, nay…insist they have that.
Nobody complains that our party nomination processes have been abandoned to populism, and the same people can be driven to a rage by a suggestion that most any other issue is not what the “Founding Fathers” wanted, and nothing is clearer in the history of the founding than they didn’t want that.