Month: November 2022

Here are average retail gasoline prices over the past year:

In most of the country, the average price of a gallon of regular is $3.32. Here in the People's Republic of California it's $4.81. In both cases, if you adjust for inflation we're now below the prices of a year ago.

Over the past day I've read at least a dozen pieces explaining how stupid it would be for Apple to toss Twitter out of the App Store. But was Apple ever thinking of doing this? Elon Musk visited Cupertino to find out:

Good conversation. Among other things, we resolved the misunderstanding about Twitter potentially being removed from the App Store. Tim was clear that Apple never considered doing so.

— Elon Musk (@elonmusk) November 30, 2022

Musk has now gone full Trump. Two days ago he said that "Apple" had "threatened" to ban Twitter from the App Store, "but won’t tell us why." People wrote and wrote about this, but I couldn't figure out why since there was zero confirmation from anyone that it was true. Today, miribile dictu, Apple's CEO "resolved the misunderstanding." They had never even considered doing this.

Of course not. Musk invented the whole thing to get attention, and Tim Cook responded calmly because he's a normal adult who doesn't need the grief but also doesn't need to start World War III over some stupid shitposting from a guy who thinks of himself as the world's richest edgelord.

Cook has had years of experience with the Chinese, and I doubt that Musk is more than an irritating fruit fly by comparison. So he took a deep breath, invited Musk to visit, and placated him without launching any dangerous confrontations. It's the same way you'd deal with an unstable seven-year-old who had his hand on a nuclear trigger: Calm him down and divert his attention somewhere else. Then get back to your grown-up work.

Here's an interesting comparison of an artwork with itself in the Orange County Museum of Art. This is a normal exposure that shows what the piece looks like to the naked eye, and I initially took it only to make fun of it. A blank white canvas? Please. It's been done.

But then I noticed something that prompted me to re-edit the photo. I selected just the canvas itself and then performed a massive Photoshop exposure correction on it:

It's not a blank white canvas. It's formed of glass beads on acrylic, and under the right light you can see the pattern imprinted on it by artist Mary Corse.

You may or may not think much of this pattern. That's up to you. But why would the museum light it in such a way that you can't even see it?

Today Eurostat released its flash inflation estimate for November. Here is my usual presentation of the data: monthly core inflation with a trendline to smooth out the noise.

The headline inflation rate was 10% thanks to high food and energy prices, but this is due mostly to Vladimir Putin, not to underlying economic conditions. Monthly core inflation is the better measure, and it dropped to 2.4% in November. The trendline is around 5.5%, which is probably the best estimate of actual inflationary pressure in the EU area.

This is still too high, but it's not wildly out of control. It's higher than the US rate of roughly 5%, but only slightly.

What should we think of the railroad contract that's been rejected by several of the rail unions and raises the prospect of a nationwide strike that would sorely test the economy? Let's review.

First off, the railroad carriers have agreed to a 24% wage increase over five years (from 2020 to 2024). That sounds pretty good, but I will bore you as usual with my admonition that you should always account for inflation:

The solid line is actual inflation. The dashed line is an estimate of future inflation from the Congressional Budget Office. It amounts to 21% over the life of the contract, which means workers are getting a real increase of 3%. That's 0.6% per year. Both sides have agreed to this, but color me unimpressed anyway.¹

The other big issue is sick leave. This isn't an issue for workers with ordinary 40-hour-per-week jobs, but it's a big issue for workers in the field. These workers face schedules that are increasingly unpredictable as railroads squeeze their timetables to increase profits.

But it's a weird issue. Rail workers currently get 11+ personal days and the new contract gives them one more. These days can be used for anything at any time, including sick leave. What's more, rail workers already have sick days, but they're unlike the sick days you and I are used to. Each year, the first incidence of illness pays nothing for the first seven days while illnesses after that pay nothing for the first four days. After that, sick pay benefits kick in. Roughly speaking, it means workers get no time off for, say, having a cold, but do get paid for more serious illnesses that require a substantial amount of time off.

The reason for this—as well as management's unwillingness to provide more personal days—is not something they will talk about honestly. But as near as I can tell, management's beef is that they think workers routinely lie about being sick. They're willing to pay for longer illnesses, since those are fairly verifiable, but not for day-to-day stuff that workers can fake.

There are a few other issues in the contract, but apparently none of them are all that important. Basically, the unions negotiated a 3% pay raise over five years and one extra personal day. That's it.

This doesn't strike me as a huge windfall for workers. I mean, maybe they were never going to get the 15 sick days they asked for, but one day seems mighty stingy. The failure of this contract seems like it falls far more on management than on labor.

¹Workers also will receive a $1,000 lump sum payment for each year of the contract.

Here is my best estimate of the change in the murder rate over the past decade:

Data for the years 2010-19 comes from old-school FBI Uniform Crime Reporting. The years 2020-21 come from the FBI's new NIBRS reporting. The year 2022 comes from combining population weighted local reporting for big cities and FBI reporting for smaller cities (both through Q3). I used growth rates rather than absolute levels because the adoption of NIBRS makes it impossible to compare levels across the past four years. Conversely, growth rates are roughly comparable though not entirely reliable.

Two things are worth noting. First, 2020 is the only year of big growth. Second, although growth immediately slowed down after 2020, it only slowed a little bit. This means the absolute level of murders has stayed nearly the same. The murder rate skyrocketed in 2020 and we've stayed at that high level for the past couple of years.

I am in an abusive relationship with my tablet. As you might remember, I'm a Microsoft Surface Pro junkie. It's nice and big. The screen is sharp and bright. It runs vanilla Windows, which means I can sync it with my desktop computer and carry around a clone of my desktop everywhere I go. These are not features that appeal to everyone, but they appeal to me.

At the same time it's got a couple of big drawbacks. First, the battery life is terrible, maybe 5-6 hours in ordinary use (which, for me, means browsing, reading, Exceling, and video watching with WiFi always on). Second, the screen constantly fails after about a year or two. It fails differently every time, and no one else has reported this problem to me, but I have three or four old Surface Pros that are basically bricked because something is wrong with the screen.

There's not much I can do about the screen except scream into the void. But the battery is a different matter. Every time Microsoft releases a new model they swear that they've significantly improved battery life. Every time I fall for this con. And every time it turns out not to be true. It's always 5-6 hours.¹

A few weeks ago the Surface Pro 9 was released. Naturally Microsoft said the battery life was great thanks to new features in the latest Intel processors. This time I didn't believe them. Enough's enough.

But! Microsoft also released an ARM version of the Pro 9, and there's every reason to think that it should provide much longer battery life than an Intel processor. Unfortunately, there's also reason to believe that it's not compatible with 100% of all Windows apps. So how does it do on those scores?

I tried to figure out the battery life from reviews, but as usual they were all over the map. I literally read reviews that put the battery life at 10% more than the Intel model and others that put it at double.

Compatibility and performance were widely disparate too. Everyone agreed that there's a category of games that don't work, but I don't play games so I don't care. Beyond that, some reviewers said it was so slow they wanted to throw it in the trash while others said performance was all but identical to the Intel model.

After diddling around with this stuff for a week or so, Marian looked at me and said, "You know you're going to buy it anyway, so why don't you just go ahead?" She was right, of course, so I did.

So what's my review? It's pretty simple:

Performance: It's a little sluggish with non-native apps, but not enough to be seriously annoying. Let's face it, most of us could use a circa-2000 Pentium chip and our apps would run fine. We're not doing weather forecasts with our tablets.

Compatibility: In the end, I only ran into two problems. First, my anti-virus software wouldn't install. In fact, no anti-virus software would install. This means I have to go back to using the built-in Windows security software, but that's not a big deal since it's fairly good these days.

The second problem was a single feature in a single app. I don't like the on-screen keyboard that's built into Windows, so I use an app called Hot Virtual Keyboard. To bring up the keyboard you use a three-finger tap, but that doesn't work on the ARM tablet. This might sound a little trivial, but the three-finger tap is seriously great and I miss it.

Battery Life: Hallelujah! The ARM version of the Pro 9 actually gets better battery life. However, it doesn't get way better battery life. I'd peg it at about 30-40% better, which means I can generally use it all day without charging. Barely. It could still use another couple of hours, but this is genuine progress.

All in all, it's a bit of a compromise. The battery life is a little better but performance is a little worse. Compatibility is good, but there's no telling if there's some crucial app you rely on that won't work. These compromises all worked out for me, but it's a bit of a crapshoot whether they will for you.

¹Why is battery life so bad? Several reasons. First, Intel processors are power hogs. They're better than they used to be, but power hoggery is built into the architecture of the chip and there's a limited amount Intel can do about it.

Second, Windows was designed as a true multitasking operating system, and that fundamentally uses a lot of power. This is part of the deep architecture of Windows, and there's not much Microsoft can do about it short of rebuilding Windows from scratch.

Third, the screen is big and bright—and therefore sucks a lot of juice. This is obviously a place where different people would make different compromises, but I wouldn't mind if the screen were a little smaller and little lower resolution.

Of course, one easy solution to all this would be a model that had a bigger battery. This would make the Surface Pro bigger and heavier, but it's already so big and heavy that I don't think it would matter much. You're either willing to accept big and heavy or you aren't, and Surface Pro users are obviously already OK with it. An extra few ounces isn't going to kill the deal.

This is a kid descending from a tree in Square Louis-XVI, a small park just a few hundred feet from our apartment in Paris. He didn't do anything while he was up the tree—or perhaps he failed at whatever he was hoping to do—and eventually he shinnied down.

The Wall Street Journal points out today that our ongoing yield curve inversion is getting worse and worse:

Judging from this, we were already due for a natural recession to start sometime in the first half of 2023. Then, a few months ago, the Fed started jacking up interest rates, which will put downward pressure on the economy at about the same time. The Fed could respond to this double disaster by lowering interest rates—which it will—but of course it will be a year too late for that.

We are so fucked. I sure hope I'm wrong about all this.

NOTE: Normally investors demand higher interest rates for longer-term bonds in order to protect themselves against unexpected inflation and Fed rate increases. For example, if the yield on a 10-year bond is 5% and the yield on a 2-year bond is 3%, the yield curve is 5% - 3%, or +2%. This is normal

Sometimes, however, investors are so convinced that near-term prospects are poor—i.e., a recession is coming—that they expect the Fed to fight it by lowering short-term rates. To protect against this they demand super-high rates for short-term bonds. When this happens you get a situation like we have right now: 10-year yields of 3.68%, 2-year yields of 4.42%, and an inverted yield curve of -0.74%.

Traffic fatalities have increased 15% since 2013. But the increase hasn't been evenly spread:

Since 2013 urban fatalities have grown a whopping 46% more than rural fatalities even though urban miles driven are up only about 10% more than rural miles driven.

Why? The question here is not why there are more urban fatalities. In fact, that hasn't always been true. The question is why the growth of urban fatalities vs. rural fatalities was nearly zero from 2006-12 and then suddenly began to skyrocket starting in 2013.