Sacking Some Statistics
Thanksgiving and Football: Why you should always go for it on 4th and short
[I]f teams that decide to kick when they are on their opponents’ 30 yard line make the field goal an average of 33% of the time, then the benefit of kicking is assigned a point value of 1 (since a field goal is worth 3 points, and 33 percent of 3 points is 1 point). Since teams that only have one yard to go when they are on the 30 yard line convert for a first down 64% of the time, and teams that are inside the 30 yard line score a touchdown about 40% of the time, the benefit of going for a first down is assigned a value of 1.8 (0.64 x 0.40 = 0.24, or a 24% chance of scoring a touchdown by going for it on 4th and 1, and 0.24 x 7 points = 1.8 points). This means that “going for it” should result in scoring almost twice as many points than kicking
One problem is that the chance of making a 47-yard field goal is about twice the value used here. It varies from year to year, of course, but field goals from 50+ yards are made at about a 50-50 clip (almost 53% last year), and 40-49 yards is north of 60% (73% last year). Which makes the expected gain from a field goal attempt from that distance about the same as from going for the first down, and perhaps slightly higher.
Also, the distribution of the abilities of offenses and defenses play a role. The average chance of scoring might be 40%, but I’ll bet that the e.g. Colts, Patriots and Saints are higher than that, and teams trying this against the Ravens are lower. And add to that situational details, such as whether a field goal increases your lead to more than one score, or it gives you a lead or ties the game.
Statistics are all well and good, but there’s a problem with looking at them without context, and not understanding what they mean. It’s also not a good idea to just retrieve a number from a dark place in order to make your conclusion look good, when the correct numbers are available.
Those Wheels are Hot
Commentary from Gizmodo
What you’re watching is a piece by artist Chris Burden called Metropolis II, in which every hour 100,000 cars pass through the city of wood block, tiles, Legos and Lincoln Logs. It’s a follow up to Burden’s Metropolis I, which was built on a similar concept but employed only 80 cars. Metropolis II is currently being constructed at the Los Angeles County Museum of Art.
More TSA Nonsense
Security and Terrorism Expert Bruce Schneier: TSA Scans “Won’t Catch Anybody”
And why would they? Zero bombers made it through (or were caught by) the old system.
Errata Security: I was just detained by the TSA
Today, I was detained by the TSA for about 30 minutes for taking pictures while going through security. Taking pictures is perfectly legal.
…
TSA: How would you like it if somebody came to your work and disrupted your procedures? How would you like it if people took pictures of you at your work?
Me: I don’t work for the government. Government agencies need to be accountable to the public, and therefore suffer disruptions like this.
TSA: Not all parts of the government are accountable to the public, especially the TSA.
Me: Wow. No, ALL parts of the government are accountable to the people, especially the TSA. I’m not sure what type of country you think we live in.
The TSA and America’s Turning Point
If America has a single founding principle, it is this: no government has any authority to take any action without the consent of the governed. Our Founding Fathers did not object to the principle of paying taxes per se; they objected strongly to the idea of being forced to pay taxes to a government where they had no input. Freedom’s cry was not “No taxation” then, and it isn’t now; it was “No taxation without representation.” The same goes for any other intrusive regulation.
George Will takes another tactic: The T.S. of A takes control
What the TSA is doing is mostly security theater, a pageant to reassure passengers that flying is safe. Reassurance is necessary if commerce is going to flourish and if we are going to get to grandma’s house on Thursday to give thanks for the Pilgrims and for freedom. If grandma is coming to our house, she may be wanded while barefoot at the airport because democracy – or the equal protection clause of the 14th Amendment; anyway, something – requires the amiable nonsense of pretending that no one has the foggiest idea what an actual potential terrorist might look like.
I have to disagree here, because I don’t think that it’s amiable nonsense at all. Here he falls into the trap that the TSA has: implementing a protocol to stop a specific threat. It’s inefficient and ignores the thousands of other ways a successful attack could be carried out. If you start profiling, then all it would take is someone who doesn’t fit the profile. People, Juan Williams being a prominent example, have expressed the feeling that they are nervous about flyers who are “in Muslim garb.” All a terrorist has to do, then, is dress to blend in, to allay that fear. But it doesn’t get rid of the threat. People in Boston famously freaked out over some flashing LEDs (twice) in recent memory, because “that’s what bombs look like.” Which is nonsense. Bombs “look like” pretty much whatever you want them to. The same goes for terrorists. At the very least they could wear a mask.
More commentary and links at Uncertain Principles: Invasive Searches and Underage Drinking in which Chad makes the comparison to other situations which seem to fall under the rubric of “We must be seen as doing something about the situation. This is something.” Which is nothing but a CYA move. “Don’t blame me, I did something!”
The bottom line is that in all things there is always, and will always be, risk. 100% safety is unattainable, and it’s dishonest to imply otherwise. It’s dishonest to manufacture fear in order to justify actions restricting our freedoms.
Update (11/26): Roger “Carlos the Jackal” Ebert: Where I draw the line
Gaining an Edge(rton)
I recall seeing some stories recently, trumpeting investigations into how cats drink. Some of the data they could have used has been out there for 70 years.
The Physics of Pie Crust
Pie crust phase equilibria: update
As anyone with baking experience knows, the stability region for pie crust is a relatively small area on the wet side of the two phase flour + dough field. This field is generally approached by adding water to a flour/butter mixture
Thermodynamics is a Good Idea
Toyota has a new commercial series out, going by the moniker “Ideas for Good.” The gist of it is using technology that they have invented, or at least use, and point it to new applications. The first one I saw was crash modeling and using it to analyze football collisions to help reduce concussions. Great.
But the next one was taking regenerative braking and putting it on roller coasters, so that we could “create the world’s first self-sustaining amusement park.” Which sounds suspiciously like perpetual motion. You can’t do it. You will always have losses of your useful energy (heat), and can’t recover all of the mechanical energy to use it again. Maybe they meant something else, but if they did, the execution was off.
Here is a link to the commercial, in case you want to watch it.
I Know What You Did Last Friday
200 students admit cheating after professor’s online rant
“I don’t want to have to explain to your parents why you didn’t graduate, so I went to the Dean and I made a deal. The deal is you can either wait it out and hope that we don’t identify you, or you can identify yourself to your lab instructor and you can complete the rest of the course and the grade you get in the course is the grade you earned in the course.”
…
So far more than 200 students have admitted to cheating.
UCF cheating scandal makes national news, on ‘Good Morning America’
It’s a 15-minute video, so the short version: the professor noticed a bimodal score distribution (rather than the expected normal distribution), with the average a grade and a half higher than usual. And then an anonymous student dropped off a copy of the entire exam bank, saying that it was in wide distribution.
I think the professor bluffed a bit when he told the students that he could hand in a list with 95% probability that all the cheaters were on it, which is probably true, and would shortly be able to hand in a list that only contained the cheaters, which is probably false. The list with all the cheaters would just be a list with all the high grades, which deviated from a normal distribution — the high-end part of the bimodal distribution. (You might not incorporate a few at the low end, but I’m guessing there isn’t much of a worry about a student who couldn’t pull better than a “C” while having the test questions ahead of time.) The problem is in identifying the cheaters with no false positives. That means not including anyone who legitimately got a high grade, and I don’t see how you can conclusively do that. But tossing the test makes a lot of that moot, since nobody gains an advantage from the cheating.
An interesting and scary scenario is what happens if the university decides someone who didn’t come forward is a cheater. What do you do if you’re that student, and you didn’t cheat?
The part about the incident not appearing on the transcript might have been a smokescreen as well. Would you hire a business grad from UCF who took this class in Fall 2010 and also had the four-hour ethics class on their transcript?
However, some information is missing, and I’m not entirely convinced this is cheating. It’s an advantage to know what questions might be asked, but whether that’s an unfair advantage (i.e cheating) depends on how you came across that information. The course had been given for several years and presumably at other institutions, so it’s possible questions were re-used. How many of them were “in the wild?”. Gathering up old exams to be used as a study guide is perfectly legitimate as far as I’m concerned; professors have to be profoundly naive to think that wouldn’t happen (and is why exams were treated as restricted material when I was teaching in the navy). The real issue here is how the students came to have the bank of exam questions and where you draw the line of coming by that information legitimately.
No, Mr. Bond, I Expect You to Diet!
The Virtuosi: Problem of the Week #3
The minions of Dr. Cherrycoke have placed Mr. Bond in an elaborate and unnecessary death contraption that, if triggered, would drop our hero into a pit of ravenous killer koalas. In other words, certain death!
Trust Me
Bruce Schneier has an extensive collection of links regarding the TSA and current screening procedures.
One thing that seems to get overlooked in all of the stories I’ve read, in which some government official insists that the ever-more-invasive security protocols are needed, in order to prevent attacks like the shoe bomber and Christmas-day bomber, is this: these protocols never would have stopped either of those attempts, because neither passenger boarded a flight originating in the US. They are being used as excuses.
The government says, “Trust us. We need to do this for your safety.” The problem is that the government has no credibility. There’s no incident of a bombing which could have been prevented by these scanners to which they can point, and no statistic of risk they can cite which they could improve upon. They promise that the scanner images aren’t retained, but then we discover that’s false. How could it be true, anyways? The government isn’t going to hang on to potential evidence in case a passenger needs a followup pat-down, or there’s a subsequent problem on a flight? All they have is a manufactured fear they keep promoting.