A real world experiment in police enforcement

In the wake of the murders of two NYPD officers, arrests in NYC have plummeted.

Citations for traffic violations fell by 94 percent, from 10,069 to 587, during that time frame.

Summonses for low-level offenses like public drinking and urination also plunged 94 percent — from 4,831 to 300.

Even parking violations are way down, dropping by 92 percent, from 14,699 to 1,241.

Drug arrests by cops assigned to the NYPD’s Organized Crime Control Bureau — which are part of the overall number — dropped by 84 percent, from 382 to 63.

The latest official stats are here (PDF).

The police unions deny any coordinated work stoppage.

Mr. Bratton said on Monday that a “weeklong period of mourning” and demonstrations that were straining resources were contributing to the drop-off in arrests and summonses. But he said the slowdown should not concern New Yorkers. “I would point out it has not had an impact on the city’s safety at all,” Mr. Bratton said.

A top union official flatly denied that there was a job action and pointed to the orders to double up and the need to police demonstrations as the main reasons.

We rarely have such a stark change in public policy with which to analyze the effects in the real world. This is a case where a crude A/B test jump-started itself in the real world (it would be more useful if it were only specific precincts within NYC that saw this decline in arrests rather than all of them, but we can still look at the effects across cities). It will be interesting to look back in a few weeks and see if there are any new conclusions to be drawn about the broken windows policy.

You shall not pass

Having spent the first morning of 2015 surfing links on my iPad, it's striking to me how much better the user experience of so-called “new” media sites is on mobile than it is for “old” media sites.

It's almost entirely because the newer sites don't resort to full-screen takeover ads which, on mobile, often make it impossible to find the close or skip button because the ad unit isn't centered properly in the in-app or mobile browser. I don't know how many times I've tapped on a link in Twitter to read an article, only to have it hidden behind a giant black blob, with no idea which direction to scroll to find the close button. I wonder how many such sites even bother to test the experience of trying to read their content the way most people find it nowadays, through links posted on social media, on mobile devices.

Let's hope in 2015 we see more native ad units as on most social networks or new media sites, and fewer of these obtrusive non-native ads like banners and full-screen takeovers. Native ads got a bad name somewhere along the line, but their user experience is far superior.

Love is all around

2014 ended as if America was fraying at the seams, but Bryan Caplan points out all is not gloomy.

Superficial observers will see further evidence that economists can't shut up about selfishness.  But on reflection, the logic of collective action is compelling evidence for the power of altruism.  How so?  Because actual human beings often engage in collective action despite the strong selfish case for inaction!  Many people give blood without the slightest recompense.  Many people voluntarily join the army when they see their country in danger, despite high risk and low wages.  Many people donate to charity even though eligibility for charity has nothing to do with their donation history.  If altruism is not their motive, what is?

Sure, true believers in ubiquitous selfishness can grasp at straws to protect their dogma.  Perhaps people donate blood for the free cookie, join the army because they might run for office one day, or give to charity in order to make business connections.  Or maybe millions of average joes are clueless enough to believe that the blood supply, the safety of the free world, and the availability of charity hinge on whatever they personally choose to do.  

Anything is possible, but that doesn't mean that anything is plausible.  Once you grasp the logic of collective action, basic economics strongly supports a conclusion that economists rarely advertise: Genuine altruism is all around us.  Benevolence doesn't explain why bakers bake bread for paying customers, but it does explain why blood donors give blood to strangers for free.

Better than the Turing test

Is the Turing test really the best way to screen for artificial intelligence? One alternative sounds more promising: Winograd schemas.

The test would take the form of a multiple-choice quiz of reading comprehension. But the text itself would have some very specific features. It would consist of Winograd schemas: pairs of sentences whose intended meaning can be flipped by changing just one word. They generally involve unclear pronouns or possessives. A famous example comes from Stanford computer scientist Terry Winograd:

"The city councilmen refused the demonstrators a permit because they feared violence. Who feared violence?"
1) The city councilmen
2) The demonstrators

And:

"The city councilmen refused the demonstrators a permit because they advocated violence. Who advocated violence?"
1) The city councilmen
2) The demonstrators

Most human beings can easily answer these questions. We use our common sense to figure out what "they" is supposed to be referring to in each case. And that common sense basically involves a combination of extensive cultural background knowledge with analytical skills. (In the first question, we can deduce that the city councilmen feared violence. In the second, the demonstrators advocated violence.)

For computers, however, these questions can be quite difficult. From a grammatical standpoint, the "they" in the sentences is technically unclear. In both questions, "they" could be either the councilmen or the demonstrators.

Is this truly a superior test of AI? We may start collecting some data points soon. In 2015, Nuance Communications is sponsoring the first of what will be an annual Winograd Schema Challenge.

I continue to insist that the next great milestone in AI is when Waze stops trying to send me to unprotected left turns as shortcuts. UPS already figured this out.