Beyond Labels

The Columbine effect

Mother Jones has written an interesting series of articles about mass shootings, latest here, full data on shootings 1980-2015 here, another analysis here.

In at least 14 cases, the Columbine copycats aimed to attack on the anniversary of the original massacre. Individuals in 13 cases indicated that their goal was to outdo the Columbine body count. In at least 10 cases, the suspects and attackers referred to the pair who struck in 1999, Eric Harris and Dylan Klebold, as heroes, idols, martyrs, or God. And at least three plotters made pilgrimages to Columbine High School from other states.

As one longtime security specialist explains in our investigation into a growing national effort to stop mass shooters before they strike, “It’s a cult following unlike anything I’ve ever seen before.”

 

I don’t think we have a topic for this week, and if we want, we could (as with the IP topic) spend some time this week surveying the area, before going wherever our conversation takes us, and dig in to this topic next week.

It’s top of mind for me right now because a guy I used to work with started posting enthusiastically on Facebook about his experience using the new “auto pilot” software patch on his Tesla Model S.

For the past several years Tesla has fitted out its Model S cars with the sensors to collect the information that a self-driving car. Tesla’s cars have always-on wireless and for those years every one of those cars has been sending information back to the mother ship. The information has been of two kinds: far more granular information about roads than maps alone provide, and information about users driving habits and choices, along with sensor data.

Tesla has been building car-driving algorithms, and training with the data acquired from their customers. Now with the release of a software patch (something Tesla does regularly) the self-driving feature is usable, “in beta” for customers who pay an additional $2500 to turn it on.

My friend is over the moon and so are a lot of other users. Tesla’s recommends that you drive “with at least one hand on the wheel” and makes clear that they assume no liability. But the plan is simple: once there’s enough concrete evidence that such cars are safer than cars driven by (inattentive, inconsistent, unreliable) humans, they will start the process of gaining legal acceptance for self-driving cars.

Meanwhile with some 60,000 Teslas autopilot ready, and more and more of them turning it on, and probably every new Tesla S customer turning it on, and with the mass-market sub $40,000 Tesla only a few years away, there will be enormous popular support to make it legal to get in your car, dial in your destination and do something useful.

The is the start of something that is big and is going to happen fast. Tesla says: every driver is teaching their algorithms how to do a better job.

This is the promise and the challenge of the intelligent machine. The promise, which is already around us, and largely unseen are AI capabilities that  integrate with what we do and assist us, making us better and more efficient. I notice this with my use of Google Search. It used to be that if I had a complex search I’d have to think about how to keyword it, then type it in. Now I don’t have to think it through. As soon as I type enough of my half-formed thought, Google Search completes it for me.

Because AI at work.

Tesla’s smart car is another example. Rather than replacing human drivers (what Google’s car is capable of doing) Tesla’s car helps them–giving just as much help (with as much risk) as drivers choose.

And at the same time, and this is very, very important, becoming a better and better driver as it goes.

I’ll post some links on the topic later on today and during the week.

If anyone else finds things that might aid the discussion and has learned the baroque interface to this site, they can post.

Otherwise email to Scott and me, and whoever gets to them first, will  post.

My addition to Scott’s post on this week’s topic, gun control. A few things that I happened on this morning.

The image below is a heat map of gun deaths _per capita_ by county.  I found it attached to this post on G+.  I’ve added some explanation below the image.

This is a link to an interactive map, on which the post is ultimately based.

The guy who made the post, Yonatan Zunger, is an engineering director at Google.  He’s an incredibly smart guy, with a PhD in physics from Stanford, and knowledgeable about a broad range of subjects. He’s one of the people I follow on G+ and I always look forward to his posts.

What’s interesting (beyond Yonatan’s post and the underlying article) is the discussion in the comments section. Like a lot of good creators of Internet content, he’d attracted a following of smart people who comment his post and add links to other, related information.  You can get quite an education on gun control reading and following the links in the comment section.

To precis some of the more interesting comments:

1. This is per-capita not, total deaths. So most cities are relatively safe, though there may be areas within any city that are not.
2. This is all gun deaths. Nationwide the ratio between suicide and homicide gun deaths is 65-35 (per this article)
3. Suicide by gun is about 80-85% effective. Suicide by non-gun is about 5-8% effective.
4. This is a link to a map showing locations with “mass shootings” I didn’t see a definition of “mass shootings” but clicking on the smaller dots, it seems to be >= 4 involved, though not all killed. Before you look at the article, guess how many such shootings occurred between the first of the year and Oct 2, the date of the article. Were you close?

Finally, no discussion of gun control can be complete without reference to  Ben Carson’s support of a theory so often cited, it’s been given a name: “Nazi Gun Control Theory.”

This is an kind of (and I love this term ) “Reducto ad Hitlerium” argument.

Geeks like me who frequent online discussion groups are familiar with “Godwin’s Law” which states: “”As an online discussion grows longer, the probability of a comparison involving Nazis or Hitler approaches 1.”