According to the FBI, here’s what crime rates look like in Chicago over the past few years:

Between 2015 and 2016, property crime was basically flat. Among violent crimes, robbery and aggravated assault were up some, but homicide skyrocketed. Why? In late 2015, the Chicago Police Department reached an agreement to drastically cut back its use of stop-and-frisk, which has led some to call the murder epidemic a result of the “ACLU effect.” Is that fair?

Paul Cassell and Richard Fowles of the University of Utah think it is. After all, homicide spiked upward almost immediately after the stop-and-frisk agreement took effect. But what about New York City? They also reined in their stop-and-frisk program, but their murder rate declined. The authors try to show that New York City is so different from Chicago that the comparison is invalid, but their case is fairly unconvincing. There’s also this:

As Professor Franklin Zimring has noted in questioning whether NYPD’s success in lower crime rates could be directly transported to other cities, “New York’s success may have been assisted by its low rates of civil handgun ownership. Even when there were a substantial number of handguns on New York’s streets, the number in homes was much smaller than in other big cities.” This means that removing guns from New York’s streets may have been an especially powerful tactic there, because handguns were harder to replace than in other cities.

I don’t think this makes the point the authors want to make. If stop-and-frisk was an especially powerful tactic in New York City, then cutting back on stop-and-frisk should have led to an especially powerful rebound in homicide. It didn’t. Likewise, if replacing a gun in Chicago was easy, then stop-and-frisk should have been pointless, and ending it should have produced only a minor effect.

The other evidence they present is similar. Most other big cities, they say, didn’t see a big increase in murder rates in 2016. True enough. But six of them did, a fact they try to handwave away. If a third of the 20 biggest cities all saw big homicide increases, then there are obviously lots of reasons for homicide to increase. This means there’s no special reason to insist that the end of stop-and-frisk is the only possible explanation in Chicago.

What’s more, as the chart above shows, other crimes didn’t follow the path of homicide in Chicago. As the authors say, this could be because homicide is uniquely tied to guns, so it’s the most strongly affected by stop-and-frisk. That’s certainly possible. At the same time, common sense suggests that a generalized change in policing that supposedly reduces inhibitions on criminal activity would have an effect on lots of different kinds of criminal activity. That’s sort of the case in Chicago, but not entirely.

I dunno. We’ve got broken windows. We’ve got community policing. We’ve got stop-and-frisk. We’ve got targeted drug raids. We’ve got the Ferguson effect. And now we’ve got the ACLU effect. We have a long history of trying to blame crime rates on specific police tactics—or the lack of them—and in the long run they never seem to hold up. In the case of Chicago, we’ve got one data point, namely that homicides started to increase at the same time that stop-and-frisk ended. Despite the length of this latest paper, that’s really all the authors have—and even that’s belied by the fact that murder rates had already increased 15 percent the year before. Maybe Cassell and Fowles are right, but I’d keep an open mind about this until and unless we get a whole lot more evidence.

Twentieth Century Fox

I had no idea this was happening until I read about it last night:

Atlanta’s municipal government has been brought to its knees since Thursday morning by a ransomware attack — one of the most sustained and consequential cyberattacks ever mounted against a major American city….Threat researchers at Dell SecureWorks, the Atlanta-based security firm helping the city respond to the ransomware attack, identified the assailants as the SamSam hacking crew, one of the more prevalent and meticulous of the dozens of active ransomware attack groups.

….In Atlanta, where officials said the ransom demand amounted to about $51,000, the group left parts of the city’s network tied in knots. Some major systems were not affected, including those for 911 calls and control of wastewater treatment. But other arms of city government have been scrambled for days. The Atlanta Municipal Court has been unable to validate warrants. Police officers have been writing reports by hand. The city has stopped taking employment applications.

[Keisha Lance] Bottoms, the mayor, has not said whether the city would pay the ransom. The SamSam group has been one of the more successful ransomware rings, experts said. It is believed to have extorted more than $1 million from some 30 target organizations in 2018 alone.

This ransomware group is tightly managed. A ransom of $50,000 is enough to be worthwhile but low enough that it’s a rounding error to lots of large organizations. And apparently SamSam refrains from bollixing up infrastructure that might get people killed—which probably helps them keep a low profile.

But this strategy only works if breaking into systems is so easy that $50,000 represents a substantial profit vs. the time spent doing it. Out of tens of thousands of organizations big enough to be worth going after, I suppose it’s no surprise that several dozen are easy pickings. But it’s still kind of shocking. It makes the second Die Hard movie seem a little less ridiculous, doesn’t it?

Reihan Salam points today to an interesting paper written a couple of years ago. The question at hand is: how do we get more low-income students to go to college? Part of the answer is to get more low-income students to apply to college. And the first step to accomplishing that is for them to take either the SAT or ACT college entrance exams.

About a decade ago several states changed the way these exams worked. Instead of requiring students to pay up-front for them and then show up on a weekend to take the test, they prepaid for every student to take the test and then administered it during school hours. In Michigan, this resulted in nearly every high-school student taking the ACT. So how many kids who otherwise wouldn’t have taken the ACT ended up scoring as college-ready? Joshua Hyman applied a bit of arithmetic to the before-and-after results and came up with this:

Hyman uses a score of 20 as the cutoff for college-ready. The area under the blue line is the number of test-takers who scored 20 or higher. The area under the green line is the number of students who otherwise wouldn’t have bothered with the test but ended up scoring 20 or higher when they did. The green area is almost half the size of the blue area. These are all students who qualified for college but never would have known it.

This means that among low-income students, the supply of college-ready grads increased nearly 50 percent simply by having everyone take the college entrance exam. The bottom line, it turns out, is that there are plenty of good students among the poor who never take the ACT because (a) college just hasn’t occurred to either them or their high-school counselor or (b) even the small cost of taking the exam is a barrier to trying.

ACT (or SAT) for all seems like a pretty efficient, low-cost way of ensuring that, at the very least, everyone who is college-ready is identified as college-ready. It’s a good start to getting more low-income students into college.

Come on, talk to each other!

Over at Vox, indefatigable interviewer Sean Illing speaks with Sherry Turkle, a technology skeptic (for lack of a better word) who worries that all our tech toys are doing more damage than we think:

Illing: Her most recent book, Reclaiming Conversation, is a warning about the consequences of living in a world where face-to-face interaction is less and less frequent.

Turkle: Mobile technology means we’re always on, always plugged in, always stimulated, always in a constant state of self-presentation….I watched kids grow up, spent time in classrooms, and saw how these changes were impacting their development….This is why I became so interested in the themes I’m exploring now — the flight from conversation, the flight from solitude, the flight from silence, the flight from boredom, all of these things that are so important to our development and to our ability to be with other people.

Illing: You’ve called face-to-face conversations “the most human thing we do.” What are the consequences of living in a world where we do this less and less?

Turkle: Well, I’m not so sure we’re going to continue doing it less and less….There are certain kinds of communication that can’t be done via texting or video messages or whatever, and I think people are starting to see that. If you want to be a true friend or partner or lover or colleague and you want to really connect, then you have to look at the person you’re engaged with; you have to actually be with them. That’s how progress is made. I think enough people are beginning to understand this.

I want to put aside the question of whether smartphones are harming culture or merely changing it, because I don’t know and I don’t think anyone else knows either. What I’m curious about is the foundational assertion that underlies this entire conversation: namely that face-to-face interaction is on the decline. Is that actually true? I’m not aware of any serious research on this question, so I’m just going to toss out some blue-sky thoughts:

• There’s a huge part of our days that are spent the same way as always. If you’re a student, you’re in school. If you’re older, you’re at work. I don’t think face-to-face conversation has declined in either of these places. Things like email and texts have probably cut down on phone calls—among introverts, anyway—but that’s all.
  Rejoinder: Kids these days will text someone a few offices away instead of just going over and talking. I’ll buy that as a tentative theory. But is it actually true?
• When we’re at home texting with someone, that’s not a replacement for face-to-face conversation. It’s in addition to it. We don’t go out for drinks less than we used to and we don’t go out on dates less than we used to. Nor do we throw fewer parties than we used to.
  Rejoinder: But what about family talk? Please. Before smartphones the kids holed up in their rooms or chatted with their friends on the phone. Husbands and wives watched television “together.”
• I don’t have hundreds of young friends or anything, but the ones I know mostly seem pretty centered and perfectly capable of holding a conversation. In fact, my experience is that today’s kids are more comfortable talking to adults than kids of my generation.
  Rejoinder: Maybe in Irvine. But lots of less privileged kids aren’t doing so well. OK, but…

…that’s the whole question, isn’t it? My guess is that time spent texting/Facebooking/Instagramming is mostly replacing time spent reading or watching TV or playing videogames. We get together in the flesh with people just as much as we used to, and the resulting conversations are much the same as always. It’s true, of course, that dinosaurs like me can get annoyed when our conversation partners are looking at their phones regularly and texting while we’re talking. But as near as I can tell, most people who do this aren’t being rude and aren’t ignoring you. They just have different cultural mores, ones that young people are comfortable with and old people aren’t.¹

Anyway, my primary question is this: do young people, on average, spend less time in face-to-face conversation than either (a) older people or (b) young people of past generations? Also, is there any evidence that people are just “beginning” to understand that real relationships require plenty of physical presence? I have no idea how you could measure either of these things, but that’s what they give people PhDs for. In the meantime, unless there’s some really persuasive evidence of this thesis, put me down as skeptical that it’s happening.

¹It’s a separate question to ask if addiction to smartphones in general is harming people. I suspect it is, but that’s just my instinct, not some hill I’m ready to die on.

These are low-lying clouds on Angeles Crest Highway leading to Mt. Wilson. At about 5,000 feet, the low-lying clouds became car-level clouds and the fog was pretty thick, though only in patches. It’s a very pretty drive.

Is this the future of self-driving cars?

David Roberts writes that he doesn’t share the current disillusionment that autonomous vehicles might never become safe enough to navigate city streets. But there’s a second problem that he is worried about:

The other form of disillusionment has to do with a growing concern among urbanists that AVs will, by making personal-vehicle travel so much more convenient, induce more of it. They worry that AVs will increase vehicle miles traveled (VMT), further clogging America’s already congested city streets….In fact, I don’t think people are worrying about it enough. There are reasons to believe that any private autonomous vehicle industry will not just increase VMT, but will pursue more VMT aggressively.

Roberts spins this out into a Space Merchants-esque dystopia where cars are free but we’re forced to watch advertising at all times to pay for them. I dunno. If taxis could be paid for solely by advertising, I figure it would have happened by now. So color me skeptical about this scenario.

However, I don’t there’s any doubt that self-driving cars will increase the total amount of driving. Of course they will. There are plenty of trips I don’t take because I don’t feel like fighting the traffic or I just don’t feel like driving for an hour or two. But with my new robot car, why not? All I have to do is settle back into the La-Z-Boy installed in my vehicle and do whatever I was going to do anyway. Take a nap. Read a book. Surf the web. Write a blog post. Of course I’d drive more.

There’s a limit to this. It will still cost money to drive, and there’s a limit to how many places we all want to go, no matter how easy it is. Still, car usage will go up.

Self-driving cars will be coming to a garage near you within a few years, no matter how many people gloat whenever there’s a blip of bad news. So how do we keep travel from skyrocketing? The increased traffic from self-driving cars is a classic externality, and the usual solution to an externality is to charge for it. So what we probably need is a graduated mileage tax on autonomous vehicles. We have to do something to keep those robots in their place, after all.

Chris Owens Asp Inc/ASP via ZUMA

Here’s a random bit of news from the “Huh” files. Lewis Hamilton almost won the Australian Grand Prix this weekend, but he ended up losing to Sebastian Vettel by a few seconds. The reason was pit stops. As you may know, race drivers prefer to take their pit stops when the cars are running under a caution flag. Since everyone is moving more slowly, the other cars eat into your lead a little less during the few seconds you’re off the track.

Of course, this isn’t always possible. You can delay your pit stops to try to catch a caution flag, but only for so long. If you need new tires or more fuel, then you have to pit. That’s what Hamilton did in the late stages of the race, losing the lead to Vettel but assuming he’d get it back when Vettel pitted, even if Vettel pitted under a caution. But it didn’t turn out that way: Vettel got his caution, pitted, and came out of the pit ahead of Hamilton. What happened?

Toto Wolff blamed a glitch in Mercedes’ software for the situation that cost Lewis Hamilton the Australian Grand Prix on Sunday….Team principal Wolff said Hamilton’s shortfall was down to a miscalculation by the systems used by Mercedes to calculate the gaps between their drivers and those from other teams…. “We thought we had about three seconds’ margin. I don’t know what happened to them, we need to ask the computers and that’s what we are doing at the moment. Whether we had a software problem somewhere, we need to fix it. I think the problem is within our systems. I think we have a bug somewhere that said 15 seconds is what you need, we had 12, it should have been enough but it wasn’t.”

Damn. They use software to figure out the gaps between cars so they know if it’s safe to pit? I had no idea.

Personally, I think the next step beyond this is robot drivers. Think about it: this is the perfect application. It’s a closed course. There are no pedestrians or obstacles or any of that. You just pilot the car around a track really, really fast. But that’s no problem: even at 200 mph, computers have far better reflexes than any human. They can gauge closing distances and gaps perfectly, and the rules of the road are trivial compared to what a robot car has to do in the outside world.

This seems like it would be a great PR stunt for, say, Google and NASCAR. Google could outfit a car that would compete in the entire regular season and everyone would be fascinated. I’ll bet the coverage would be wall-to-wall. Naturally the robot car would exit the field during the season-ending playoffs, since no one wants to run the risk of it winning. But it would be a lot of fun until then.

We have good news and bad news this morning:

• The good news: Democrats managed to get some additional funding for the 2020 census so that it can actually be done accurately.
• The bad news: Republicans are insisting that the census add a question about whether you are a US citizen.

The relentless drive of Republicans to maintain their gerrymandering edge in elections is truly a wonder. In Pennsylvania they’re outraged that a judge has made district maps only mildly pro-Republican instead of comically pro-Republican. In North Carolina, they’d allow blacks to cast only three-fifths of a vote if they could get away with it. On a national level they’ve tried to sabotage the census, because the people who are hardest to count are mostly Democrats. They’ve finally sort of caved on that—admitting they need more money and giving up on trying to appoint a Census director who thought that North Carolina’s approach to redistricting was just peachy—so now they’ve moved on their latest wheeze: a citizenship question. Their hope is that this will scare people away from filling out the census form, thus reducing the recorded population of Hispanic-heavy states. There’s also a very subtle effect of this on the state level that might help Republicans keep their gerrymandering edge in 2020.

They just never quit. Some of their scams are obvious and some are subtle, but the goal is always the same: to reduce the voting power of anyone who’s likely to lean Democratic. Ladies and gentlemen, this is your 21st century Republican Party.

The folks at March For Our Lives have released their five-point list of fanatical, half-baked, 2nd-Amendment-busting demands:

1. Fund more gun violence research. We actually made a step in this direction when President Trump signed the 2018 budget, which clarifies that the 1996 Dickey Amendment doesn’t prohibit the CDC from conducting gun research.
2. Unleash the ATF. Let them store their background-check records on a computer, for example.
3. Universal background checks. In theory, everyone is in favor of this. In theory.
4. High-capacity magazine ban. This has long been my favorite. MFOL is calling for a 10-round limit. I’d make it six, myself.
5. Assault weapons ban. The gun folks are right when they say it’s tricky to define “assault weapon,” but it’s not actually impossible.

This stuff is just plum crazy. Those kids have gone off their—

Wait. That’s it? That’s really…very reasonable, isn’t it? It’s also politically plausible. And legal too, since the Supreme Court has already ruled on all these things. They’ve really done their homework, haven’t they? Maybe we should listen to them.

I was noodling away this morning and came across a Bob Somerby post responding to a New York Times column that was fact-checking Donald Trump’s claim that the budget he signed on Friday provided the military with “the largest pay increase for our incredible people in over a decade.” As it turns out, it’s actually the largest in the last eight years, not the largest in over a decade. Somerby thinks it should therefore have been labeled “wrong,” “incorrect” or “false,” not “imprecise and requires more context.”

Fine. But the real reason I’m writing this post is because I was eventually led to a report from the Congressional Research Service that lays out how military pay increases work. It turns out that pay increases are based on a formula that’s similar to the inflation rate. Congress and the president are involved only if they want to change the formula. Here’s what this looks like since the current formula was put in place:

The Pentagon has lately been trying to reduce the growth of compensation costs following a decade of substantial increases, so they’ve requested pay raises lower than the formula for the past five years. This year, President Trump went along with that. He did nothing to try to increase pay for the troops. In the end, though, the military got a raise this year that matched the formula, which came to 2.4 percent. This is thanks to Congress, not President Trump.

But that’s nowhere near the most important point. Whether a pay increase is large or not depends on the inflation rate. A 10 percent pay increase in 1980 would have been terrible. A 3 percent pay increase in 2009 would have been pretty good. Here’s the growth in military pay since 2000, adjusted for inflation:

In the only terms that actually matter to real people, this year’s pay increase is the largest since…2016. It’s the fourth-lowest of the past decade.¹ It’s nothing to write home about.

Now, I don’t seriously expect politicians to refrain from using whichever statistics make them look the best. That’s life. But for the rest of us, why can’t we simply agree to always use inflation-adjusted figures in cases like this and dispense with all the “context” and “imprecision” crap? With only very narrow exceptions, a series of dollar figures over time should be displayed primarily in real terms and news consumers should become accustomed to this. If you feel the need to show actual nominal figures as well, do it in a footnote or something. If you don’t know how to convert nominal to real dollars, then you should learn before you write about stuff like this. It only takes two or three minutes for someone to show you how.

¹Based on a consensus inflation forecast of 2.3 percent for 2018.