Category: Uncategorized

The American economy gained 199,000 jobs last month. We need 90,000 new jobs just to keep up with population growth, which means that net job growth clocked in at a moderate 109,000 jobs. The headline unemployment rate ticked down to 3.7%.

The number of employed was way up and the number of unemployed was way down, so this is a nice jobs report. Things are still trending down, but at least for now the employment picture is holding together.

One note, however: several news outlets have taken November's number as evidence that we're approaching a "soft landing." That's absurd. Employment is trending downward, and we have no idea what that means for the future. Typically, employment is OK up until the very month when it isn't and we slip into recession. The current trend is simply no indication one way or the other of what will happen six months from now or how bad it will be.

Charlie Stross wrote today about an experiment with Google's Bard AI assistant. "Tell me five fun facts about Charles Stross," he commanded it, and for the first five it did pretty well. But then he asked for five more, and then five more. And at that point Bard just went bonkers, making stuff up out of whole cloth.

That sounded like fun! So I headed over to ChatGPT-4 and asked it for fun facts about Kevin Drum. It was very disappointing. It basically got everything right, though it did so by being pretty careful and sticking closely to Wikipedia.

But maybe ChatGPT-4 is just better than Bard? Sure enough! I won't bore you with the entire conversation, but here are the final five fun facts Bard offered up about me:

5 More Fun Facts About Kevin Drum:

Language enthusiast: Kevin Drum is fascinated by languages and enjoys learning new ones. He is fluent in English and Spanish and is currently learning French. He believes that learning languages opens up new perspectives and fosters understanding between cultures.

This is laughably wrong. I speak only English and am famously bad at learning new languages.

Gardening hobbyist: Despite living in an apartment, Kevin manages to cultivate a small but thriving balcony garden. He enjoys growing vegetables, herbs, and flowers, and finds gardening a relaxing and rewarding activity.

Also laughably wrong. In fact, when we bought our current house I made sure to tell Marian to pick one with a yard small enough that she was comfortable taking care of it herself. I wanted nothing to do with it. Also, I don't live in an apartment.

Travel enthusiast: Although not as much as he'd like, Kevin enjoys traveling and exploring new places. He has visited several countries around the world and is always eager to discover new cultures and experiences.

This is more or less right, though I'm not really all that passionate about discovering new cultures. Mostly I like taking pictures.

Amateur astronomer: Kevin has a passion for astronomy and enjoys gazing at the stars. He owns a telescope and occasionally attends stargazing events. He finds the vastness of the universe awe-inspiring and loves learning about the cosmos.

Half credit? I do have a telescope, but I've never attended a stargazing event and I use it for photography, not gazing.

Animal lover: Besides cats, Kevin has a deep love for all animals. He supports animal welfare organizations and advocates for the humane treatment of animals.

I do like most animals, but I've never supported an animal welfare organization.

Bonus Fact: Kevin Drum is a big fan of coffee and can often be found sipping on a cup while writing or reading. He considers coffee to be his essential fuel for thought.

Again, laughably wrong. I hate coffee.

Oddly enough, Bard also told Charlie Stross that he was a language enthusiast. Does it just make that up about everyone when it starts to run out of things to say?

I got into a Twitter altercation last night with Nate Silver, and I need to get it off my chest because it's just so damn weird.

Background: As you know, shortly after the pandemic started a team of researchers published an article titled "The proximal origin of SARS-CoV-2." It concluded that the COVID virus was almost certainly natural, not a lab leak. The article was highly influential and read by millions.

Then, earlier this year, a large cache of internal emails and Slack messages among the researchers was leaked. This started a huge controversy: did the scientists privately believe a lab leak was likely while publicly saying it wasn't? After a while I finally got curious enough to read the entire archive of messages, along with multiple stories outlining the argument that they had lied. In the end, I came to the conclusion that there was nothing there. The case for lying was based on cherry-picked quotes taken out of context; timelines that hopped back and forth; and deliberately bad-faith interpretations. In reality, all that happened is that the researchers communicated openly with each other about doubts and differences of opinion along the way, with those opinions changing as new evidence came in. My summary, which I presented in a clear timeline with all the relevant context, is here and here.

That's where the argument with Nate starts. Here it is:

NS: People in the scientific community have to stop being wimps and call shit like this out. It is *incredibly obvious* that Proximal Origin was a bogus, unscientific political exercise. This has been an Emperor Has No Clothes Moment for the scientific community. It's depressing.

KD: There is no evidence whatsoever for that. You've gone off your nut on this subject.

NS: Kevin I respect you but that article I linked to contains 10k words of exceptionally detailed and persuasive reporting. The authors of the paper are literally saying they don't believe the conclusion. You have a real blind spot here that I'm having trouble understanding. I would ask you to take a fresh look at the topic. The evidence is not ambiguous.

KD: I agree! The evidence isn't ambiguous. The article you linked to is the usual cherry picking and deliberate misinterpretation. If you're willing to read a few thousand more words, here's my take after reading every last email.

NS: Yeah I've read those and I think you are just self-evidently, insisting-the-sky-is-orange wrong here. It's OK I have lots of friends who have some obviously wrong beliefs and we're still friends. It's probably time to agree to disagree here.

Nate is a smart and careful guy who, as far as I know, had no special axe to grind when he began looking at this issue.

But so am I. And yet, we don't just disagree on what should be a fairly simple factual dispute, we completely, totally, 180 degrees disagree. Nate thinks I'm sky-is-orange wrong and I think the same of him. It's mind blowing to me that we could end up here.

Keep in mind that we aren't disagreeing about a subtle scientific issue. This isn't a question of whether the natural origin theory of COVID is true or not.¹ It's a simple look at an archive of messages to see if a team of human beings said the same thing in private that they said in public. You need to have a reasonable grasp of the science to understand what they're talking about, but that's all.

And yet we are where we are. How is this possible?

¹In the years since "Proximal Origin" was published, I'd say the evidence for a natural origin of COVID has become almost insurmountable. But that doesn't matter for our present purposes. All that matters is what the researchers knew in March 2020 and how they went about presenting it.

This is a follow-up to the previous post. It shows federal taxes and spending separately for three different categories: Medicare, Social Security, and everything else.

The "everything else" category includes ordinary discretionary spending, Medicaid, and other social welfare programs. Social Security includes both retirement and disability. Medicare is Medicare. Here they are:

I chose the years carefully: the charts start at 2003, after increased spending for the war on terror had settled in, and end in 2019, before pandemic spending surged. This provides a good long-term look at things without being fooled by transient spikes.

As you can see, spending in the "everything" category goes up and down with recessions but has basically stayed stable since 2003. It was 11.9% of GDP in 2003 and 11.7% of GDP in 2019. It's currently running a deficit of about 1% of GDP because taxes have fallen since 2015.

Social Security outlays have gradually increased and the program is now running a deficit of about 1% of GDP, which is being made up by cashing in bonds from the the trust fund. In a decade this will stabilize at about 1.5% of GDP.

Medicare has gone up steadily and is now running a deficit of about 2% of GDP. Its growth rate has slowed recently but it's still likely to run bigger deficits over time.

This is everything except interest on the debt, which isn't under anyone's control.¹ Roughly speaking, the deficit is all about Social Security and Medicare, with a little bit due to recent declines in corporate tax receipts. That's it.

¹And anyway, interest expenses will go down mechanically if primary spending and taxes are brought into alignment.

There is apparently some chatter among the chattering classes about the need for a debt commission to examine the federal debt and recommend ways forward. This is unbelievably tedious and dumb. The federal budget is not complicated, and everyone serious already knows what we have to do if we want to balance the budget.¹

Along those lines, today I got pointed to a piece by Paul Winfree explaining why the federal debt is about to explode:

In the next three years, the elements of the perfect debt storm will come together. At the end of 2025, (1) most of the provisions of the 2017 tax reform that lowered taxes for American workers, families, and small businesses will expire. A year later, (2) $350 billion in funding that states and local governments have become addicted to (and that was part of the $1.9 trillion Biden spending bonanza) will also expire, (3) as will $400 billion in contract authority from the bipartisan “infrastructure” bill. Meanwhile, (4) hundreds of billions of dollars in unpopular provisions used to pay for the Inflation Reduction Act will kick in (5) along with unanticipated costs associated with the new green energy tax subsidies.

Unless I'm missing something embarrassingly simple, these items are (1) a tax increase, (2) a funding cut, (3) another funding cut, (4) another tax increase, and (5) a vague and unspecified spending increase that probably won't happen.

In other words, they're all things that will decrease the budget deficit, not increase it. Am I an idiot? What am I missing here?

¹You need to address spending/funding of Medicare, Medicaid, and Social Security. Everything else combined is a relative pittance.

This is the sun setting behind a pair of bridges over the Atchafalaya River in Morgan City, Louisiana.

Google takes a victory lap for its latest AI model:

With a score of 90.0%, Gemini Ultra is the first model to outperform human experts on MMLU (massive multitask language understanding), which uses a combination of 57 subjects such as math, physics, history, law, medicine and ethics for testing both world knowledge and problem-solving abilities.

Keep in mind the old saying: There are lies, damn lies, and benchmarks. That said, this is impressive stuff. What it really shows, though, is something no one wants to admit: modern AI compares favorably to human reasoning primarily because human reasoning is no great shakes. The vast majority of what passes for human thought is just memorization and sophisticated pattern matching. That's what LLM models like Gemini and ChatGPT do, and we know they follow fairly simple rules because we wrote the rules. If they can now compete with ordinary people, it's because ordinary people also rely on fairly simple rules.

But what about the Shakespeares and Einsteins? The people who produce genuinely innovative thinking? Face facts: Einstein's brain wasn't all that different from yours or mine. It used the same basic connections and rules. Whatever the difference was, we'll probably get there before much longer—maybe with LLMs, maybe with something else.

This is why I'm so confident that AI will reach human level in the near future. First, because I don't think "human level" is all that extraordinary. At heart, we are overclocked apes.

Second, AI is following precisely the learning curve you'd expect from steadily increasing compute power. The fundamental advantage humans have always had is simply that our brains are phenomenally powerful. As recently as five years ago, ChatGPT couldn't have existed because processing power simply wasn't adequate outside of supercomputer labs. Today it's close. In a few years it will be there.

From Reuters:

The main lobbying group for U.S. retailers retracted its claim that "organized retail crime" accounted for nearly half of all inventory losses in 2021 after finding that incorrect data was used for its analysis.

....The NRF's claim that organized retail crime accounted for "nearly half" of inventory losses was repeated in multiple media reports on the issue. The NRF has cited growing rates of crime in calls for Congress to pass new laws, including proposed legislation that would broaden the scope of offenses considered “organized” crime and increase potential penalties.

This was always laughably ridiculous. The NRF's own reports have consistently reported that all shoplifting is responsible for only about a third of inventory shrink. Organized gangs are responsible for a fraction of that. At a guess, they account for 2-3% of inventory shrink, a number that hasn't changed much over the past decade.

America is prone to periodic media-driven panics. Preschool satanic rituals. Frivolous lawsuits. Video games. Social media. CRT. Whenever one of them takes off, there's suddenly an explosion of evidence for it—and it's the same for shoplifting gangs. News networks play the same few video clips over and over. CEOs use it as a handy excuse for weak earnings reports. Experts are trotted out to "explain" what's happening. We're constantly reminded that most shoplifting isn't reported, as if that hasn't always been the case. And lobbying groups like the NRF cite transparently idiotic "facts" about the skyrocketing scope of the problem.

Eventually time passes and we all realize we overreacted. But while the panic is in high gear we're utterly convinced by the manufactured evidence tossed in our faces.

Shoplifting is a problem. Does anyone deny it? But it's been a problem for a long time and isn't much different today than it was a decade ago.

Apologies, but I couldn't really keep focused on the Republican debate tonight. Instead I spent my time shopping Time's Cat of the Year cover—which, honestly, was probably a better use of energy than watching the TV.

Basically, there were two things going on tonight. On one side you had Chris Christie and Vivek Ramaswamy, who know they have no chance of anything and can therefore say whatever they feel like. In Christie's case this mostly meant taking weak shots at Donald Trump. In Ramaswamy's case it meant being a childish asshole.

Then you have Nikki Haley and Ron DeSantis, both of whom have minimal but real chances of winning the Republican nomination. This meant spending lots of time calling each other liars. DeSantis accused Haley of being insufficiently dedicated to the degradation of trans people. Haley said he was lying. Haley said DeSantis was insufficiently dedicated to bathroom exclusions for trans people. DeSantis said she was lying. Very uplifting stuff.

On substance there was precious little daylight between the two. They love Israel. They hate university presidents. They love Taiwan. They hate China. They love parents. They hate trans people. They hate illegal immigrants. They hate the Deep State. They hate the FBI. And they are going to CLEAN HOUSE when they're elected president.

Yawn. It was a nightlong insult fest with little else going for it. My guess is that DeSantis made his points better than Haley, but his affect is so unlikeable that Haley might have won on points.

The one thing that continues to perplex me is Chris Christie's attacks on Donald Trump. Tonight he went after Trump more persistently than he's done in the past, but his actual criticisms of Trump are astonishingly weak. "He's unfit for the job" is fine as a summation, but only after you've reeled off all the reasons he's unfit for the job. Christie never does that. He talked tonight about Trump being all about retribution and revenge, which is fine, but hardly something that's going to change anyone's mind. I wonder why he doesn't go after Trump more thoroughly?

For obvious reasons, I approve of Time's choice for Mammal of the Year.