Kevin Drum

The Washington Post has a story today about NewsGuard, a service that rates the reliability of news outlets. It's the usual thing: conservatives are upset because lots of conservative sites are flagged as unreliable. Imagine that.

As it turns out, NewsGuard doesn't publish its ratings. You have to pay for them. But the Post provides a sample:

National Review scores higher than the New York Times! Sadly, there's no rating for jabberwocking.com. I wonder how I'd do?

But this is not what I'm really interested in. It turns out that NewsGuard also grades the accuracy of AI chatbots, and they do release these ratings publicly. Sort of:

NewsGuard says they don't reveal the individual chatbot names because of the "systemic nature" of the accuracy problem. Boo! At the very least I'd like to know the identity of Chatbot 10, which had a perfect record. For that matter, I'd also like to know the identity of Chatbot 5, which seems worth avoiding.

If anyone at NewsGuard wants to leak the names of the chatbots, my email address is at the upper right.

This is the Thunder Mountain ride at Disneyland. Since I now have a camera that can do this, I played around with zooming the lens while exposing the image. This turned out to be trickier than I had supposed. You have to time everything just right; zoom the lens at the right speed; and make sure the subject is centered even though it's moving. I failed on that last one over and over. This was the closest I got.

I noted the 2023 violent crime rate from the NCVS survey back when it was released, but here it is again:

To recap: violent crime didn't go up during the pandemic. Only homicide did. In fact, overall violent crime rates haven't changed more than slightly for over a decade. But just to show how hard crime analysis is, here's a breakdown by race:

A couple of days ago I noted that Black victimization was much higher than white victimization. That was based on FBI summaries of local police reporting. But in the NCVS survey data there's virtually no difference.

So which is it? I personally find the FBI numbers more plausible, but who knows? Both of these sources are fairly reliable and they flatly disagree. We just don't know.

Just how many people work from home these days? This question has puzzled me a bit, but a team of St. Louis Fed researchers may have the answer. They combined data from several sources to come up with overall estimates for 2020 through 2024.

One thing they discovered is that a lot of the disparity between different sources was due to self-employed workers. The numbers tighten up if you consider only employees—which is generally what we care about anyway.

The charts below show the rates over time for (a) 100% remote workers and (b) the share of remote days for all workers:

In both cases, after peaking early in the pandemic, remote work rates have steadily fallen back. Fully remote work has gone from 7% to 11%—a big percentage gain but not really a lot in absolute terms.

Likewise, total remote days is considerably higher than it was before the pandemic, but even at that it's only about 22% and still declining. It's significant but not earthshaking.

This may be the answer to the weird mismatches I've noticed for things like office vacancy rates, commuting times, and reported WFH hours. Maybe working from home hasn't increased all that much and continues to decline as employers tire of it. It might still be a couple of years—and a bracing recession—before we really know how this will play out.

Brendan Karr is Donald Trump's pick to head the FCC. Naturally this means he has to write threatening letters to media outlets Trump has beef with—like ABC:

The letter, which was obtained by CNN, begins, “Dear Mr. Iger, Americans no longer trust the national news media to report fully, accurately, and fairly.”

....“My understanding is that ABC is attempting to extract onerous financial and operational concessions from local broadcast TV stations under the threat of terminating long-held affiliations.... “The fact that a massive trust divide has emerged between local news outlets and national programmers like ABC only increases the importance of retransmission consent revenues remaining available for local broadcast TV stations to invest in their local news operations and content that serves their communities,” Carr writes.

This is typical MAGA intimidation, but for about the millionth time I'd like to point out that Carr is wrong about declining trust in the media:

In the post-Fox era, Republicans have lost trust in the mainstream media. Among Democrats, trust has actually gone up a little.

Republicans have spent the past few decades setting their own hair on fire and then pointing out that it's getting a little warm in here. But those of us who stay away from the conservative gas can haven't been affected much. We just smell the burning scalps and wonder why Republicans do this to themselves.

UPDATE: Apparently some conservative site got hold of this and my Twitter feed is now stuffed with MAGA types claiming that I'm a no-good lying liberal asshat. This is because (a) I didn't include independents in the chart and (b) I also didn't include an age breakdown.

Very strange. I didn't include indies because it's well known that nearly all of them are actually partisans but Gallup didn't break that down. I didn't include the age breakdown because . . . why would I? The whole point here is to look at partisan differences.

In any case, as always, I've linked to the source so you can look at anything you want.

Today's big news is the House Ethics Committee's report into the relentless sleaziness of (former) Rep. Matt Gaetz. Naturally you want to see this in chart form:

According to the report, these are all the women Gaetz paid for sex. "Woman 1" isn't included because she was Gaetz's long-term girlfriend and therefore "some of the payments may have been of a legitimate nature." Likewise, "Woman 10" is presumably really "Victim A," the 17-year-old Gaetz statutorily raped in 2017.¹

As it happens, I don't think the report tells us anything we didn't already know. Obviously Gaetz is a sleazeball. On the other hand, the sex was apparently all consensual; the minor girl was 17 years 8 months old and never mentioned she was underage; the drugs were just the usual party pharmacopeia of the time; the alleged gift violation was a single trip to the Bahamas;² and he helped one of his girls get a passport.

I'm happy to see Gaetz gone, but honestly, I can't get too worked up about this stuff. I really don't care much about consensual sex and drugs, and the other stuff is only barely over the line. He's hardly worth wasting an indictment on.

¹Allegedly.

²In other words, about a twentieth of a Clarence.

Fatal occupational injuries were down by about 200 last year, representing a rate decline from 3.6 per 100,000 to 3.4.

As you can see, occupational deaths declined steadily through 2009 and then abruptly flatlined. There's been no progress since 2010. Here's how everything broke down in 2023:

Most of the deaths by violence were during robberies, but every year about a hundred people are killed either by a coworker or by their spouse showing up with a gun in their hand.

Sunrise over the Danube near Regensburg.

According to the Census Bureau, the median selling price of new homes plunged nearly $25,000 in November. The price of a new home is now exactly the same as its pre-pandemic average:

Over the longer term new home prices are still up considerably. Adjusted for inflation, a new home costs about $100,000 more than it did in 2000.

Tyler Cowen links without comment today to a paper that estimates whether immigration research is tainted by bias:

Our analysis exploits a rare opportunity where 158 researchers working independently in 71 research teams participated in an experiment. After being surveyed about their position on immigration policy, they used the same data to answer the same well-defined empirical question: Does immigration affect the level of public support for social welfare programs?... We find that research teams composed of pro-immigration researchers estimated more positive impacts of immigration on public support for social programs, while anti-immigration research teams reported more negative estimates.

Technically this description is true. But check out the amount of bias:

It's tiny! There are some smallish bumps in the tails, but overall the teams produced very similar results regardless of whether their personal sentiments were pro, anti, or neutral. This should increase our confidence in the honesty of immigration research.

Now, for what it's worth, I'm skeptical that this paper represents the real world very well. One of the co-authors, for example, is George Borjas, and I'm not sure I've ever seen his name on a paper that has anything good to say about immigration. Likewise, more lefty economists almost always conclude that immigration is beneficial in one way or another. They might all come to similar conclusions in an artificial environment where they know they're being graded, but real immigration research has always seemed considerably more biased to me.