Jerrod Ankenman's Journal

Username:	Password:
Forgot your password?	Remember Me

Games and Art
David Sirlin writes about Brenda Brathwaite, an artist/game designer who is very concerned with meaning and blurs the lines between games and art. I found this very interesting, and wonder what kind of experience it would be to play Train cold, without knowing what was going on.

http://www.sirlin.net/blog/2009/11/23/migs-brenda-brathwaite.html

This is cool.
Story about a guy who tried to vanish as a contest.

http://www.wired.com/vanish/2009/11/ff_vanish2/

Fish.
Our fish died today. It was a betta that we've had for three years or so.

That fish was awesome. It lived through two weeks with the heater at 50 in the winter. It lived through murky tank water, probably not being fed for several days in a row. It evaded the cats. It was pronounced dead by one of us at least ten times, but always came back to full health, swimming around and flaring its gills when you would look at it funny. The fish was a warrior.

Also, check this video out:
http://espn.go.com/video/clip?id=4628040&categoryid=2378529

Things that bring a smile to your face
I just couldn't help smiling at an email I got today from the Records office at Columbia:

Hi Jerrod,

Would you like your Middle name spelled out on your Diploma?

We currently have I.

*Name changes will not be permitted after the diplomas are printed.

Regards,

(etc)

PhD applications
I'm working on my personal statement for my PhD applications, and I wrote something like this:

In the past, my work on poker has been primarily focused on “toy” poker games, games that are either abstracted or limited in some way to make them analytically tractable. Solving such games can provide insight into some particular concept and incorporating that insight into play in the “real” game leads to stronger play. However, the next steps of poker research for me are largely in the direction of solving more complex and much more computationally intensive games. Other research interests of mine include modeling of events; I have worked in the past on stochastic prediction models for sports outcomes and for predicting election outcomes. At SIG, I worked on [redacted] models for [redacted]. We applied a [redacted] to tick-by-tick [redacted] and obtained [redacted] results. We then tried [redacted] the model by [redacted] the [redacted] parameter and that [redacted] our results.

Normed vector spaces...
Sign seen at a church on the side of the road:

GOD'S GRACE IS GREATER THAN YOUR SINS

jerrod: what kind of ordering on a space containing both God's grace and my sins is implied by that sign?

bill: well it seems they mean magnitude, so it must be some kind of normed vector space.

OpenOffice Base
Wow. I posted about this before, but then I thought it was just a usability issue. It's not.

OpenOffice Base is incredibly bad.

When you modify a database table, the database should be updated. This isn't a word processing document or a spreadsheet, where you have to "save changes" in order to save them. If you claim to be developing a database, but don't automatically write changes to disk, and don't provide any way to make sure that changes are written to disk other than, say, exiting the program, you DID IT WRONG. On top of that, the front end is really awful.

Our power went out today, and when Michelle's computer came up with OO document recovery, it recovered back to a database with entries through 9/22. So we lost two weeks worth of data on what Mara ate, which basically explodes the modeling I was going to start on soon and puts us back to square one on really understanding the dynamics of her phe intake vs lab results.

Going to go buy Access now, I guess. Argh.

Messing up intuition
OK, so I just learned something. Maybe I should have known it before, but it was pretty counter-intuitive to me, and kind of neat.

Consider the following process:

You are at zero on a number line, which goes off in either direction. Every second, you flip a fair coin; if it's heads you move toward +, if it's tails you move toward -. This process is called a random walk. One of the things we math people are interested in is whether the random walk is recurrent. A random walk is recurrent if the following is true: if we run the random walk forever, we will return to any particular place on the number line an infinite number of times.

It turns out that this one-dimensional random walk is in fact recurrent; that is, if we start at zero and run the process forever, we will revisit zero an infinite number of times.

Now consider a two-dimensional analogue: we are on a vast plane, starting at (0,0). The coin flip is now a 4-sided die, and we have an equal probability of moving north, south, east, or west. This two-dimensional random walk is likewise recurrent; we will revisit the origin an infinite number of times if we just run the thing forever.

Third, consider a three-dimensional analogue: we are in a vast space, starting at (0,0,0). Now we have six directions we can go in: NSEW, up, and down.

Here's the counter-intuitive thing: this three-dimensional random walk (and walks of higher dimensions) is not recurrent. So if we start at (0,0,0) and run the process forever, we will only return to (0,0,0) a finite number of times. Can you see why?

captaino can use this as a PotD if he wants.

miscellany
One of the best books that I own is called Lend Me Your Ears. It's a collection of great historical speeches edited by William Safire.

RIP.

Also, great news about Roman Polanski getting arrested! If you are at all unsure that this is a good thing, please read the transcript of the testimony about the incident.

More racism!
OK, so Jimmy Carter was talking about Joe Wilson, the guy who called Obama a liar during his health care speech. He said that Wilson's act was "based in racism" and rooted in the fear of a black president.

http://news.yahoo.com/s/ap/20090916/ap_on_go_co/us_health_care_heckling_carter

I would create a poll, but there doesn't seem to be any way to link the answers between one question and another, which is the whole point of this exercise.

Q1: Do you think that Wilson's outburst was because he is a racist or because of racist views he holds, or that it was particularly rooted in the fear of a black president.

Strongly agree/Agree/Neutral/Disagree/Strongly disagree

Q2: Where are you on the following political spectrum (one media example is given for clarity of what I mean)?

Far left (The Nation)
Left (Slate)
Center-left (CNN)
Center-right (The Economist)
Right (WSJ)
Far Right (Rush Limbaugh)

My answers: Disagree, although there could be more to the story that I don't know about. Center-right.

RIP Norman Borlaug
Died Saturday...

One of the greatest humanitarians of the 20th century.

tongue twister
Try this: the things to do to your teeth before going to bed.

"Brush, floss, rinse."

Go on, try it!

Protesting spending...
http://www.washingtonpost.com/wp-dyn/content/article/2009/09/12/AR2009091200971_pf.html

"Tens of thousands of conservative protesters, many complaining that the nation is racing toward socialism, massed outside the U.S. Capitol on Saturday, angrily denouncing President Obama's health-care plan and other initiatives as threats to the Constitution."

Uh, good idea and all, guys, but where were you while the Bush administration was spending like drunken sailors?

Just to mention...
I would just like to comment that OpenOffice Base is absolutely stunningly bad, compared to the rest of OpenOffice.

I cannot explain how genius this is
So if you don't know what I'm talking about, go read Ender's Game.

XKCD, of course.

political mail
I get DNC mailings now because I gave money to defeating California's Prop 8. And I get RNC mailings because I changed my voter registration from independent to Republican so that I could vote for Ron Paul.

Suckers. I hope both parties spend all their resources on mailing crap to me that I will just throw away.

High-speed rail
Harvard economist Edward Glaesar has posted a series of articles examining the Obama administration's high-speed rail proposal in the NYT Economix blog.

Example (you can get to the other articles through the links at bottom of page):
http://economix.blogs.nytimes.com/2009/08/18/what-would-high-speed-rail-do-to-suburban-sprawl/?em

Executive summary: Even with very favorable assumptions, it's really bad and basically cannot ever come close to paying for itself. Wonder if this will matter; I thought we were bringing science back into decision-making? Or is that only for stem cells, I can never remember what pie-crust promises* attach to what issues.

*Mary Poppins ftw!

oddly prescient
So while I was googling something because scottro apparently doesn't have google available to him, I ran across this page:

http://discuss.joelonsoftware.com/default.asp?biz.5.512660.13

about ram prices.

What struck me was a comment midway down where they were talking about inflation making the price differences between memory in 1980 and in 2009 even more remarkable.

Someone claimed that the discount value over those years was 160% or so. Then someone else responds. "House prices are up 400% over that time. Something's wrong."

I think in the context of the page, they were disputing the inflation number...but yes, something was wrong.

Competition in health care
I'm not even going to pretend to be informed about what the latest health care plans are, since it seems to change every few days and noone seems to know what is in the current one.

But I've heard from both sides stuff about a government-run plan that competes with insurers, like in a free market. Also both sides say that insurers won't be able to compete with this government-run plan. I had some thoughts about this and since it's my blog, I'm going to write them down.

People point to Medicare as a great cost controller and seem to argue from this that a non-profit government plan could outperform insurers and therefore would be more efficient. But I think this is wrong. There are two main reasons that Medicare outperforms costwise. 1) There is an incredible amount of fraud in Medicare (I think $60B a year, when I looked around last time?) that is counted as "care delivered." 2) Medicare's cost control mechanism is to say to doctors, "We will pay you less, and if you refuse our patients because we pay you less, then you must refuse all of them." Because Medicare has a virtual monopoly on care for patients who are eligible, doctors suck it up and take the patients. They make up the difference by charging people who have insurance more, and by charging people who do not even more than that. This is marginally sustainable because a lot of people have insurance, and so it evens out. However, if Medicare were all of healthcare, we would just have fewer and less qualified doctors over time, as the best and brightest migrated away from medicine. It's nice to have that kind of pricing power, but it's not a basis for cost controlling health care.

Suppose that we did have a government-run insurance company that was able to kick all the other insurance companies' asses by not needing a profit and being awesome because the government suddenly becomes incredible at running businesses. The idea here is that a government-run company would win because it would be a non-profit running as lean and mean as the for-profit companies and they just wouldn't be able to compete because of their cost of capital. This would be great. I'd totally submerge any "proper role of government" objection I might have if this were the case. This is one side of the debate about a government plan: call it the "left" side: that insurance companies wouldn't be able to compete with a government plan because the government plan would have inherent advantages that the insurers couldn't overcome.

The other side of this argument (call it the "right" side -- right here is directional, not correctness) is this: The insurance companies wouldn't be able to compete with a government plan because Congress would have too much invested to watch the government plan fail, and they would be unable to keep from meddling with it, expanding it, giving tax advantages or whatever to people who choose it, causing it to swell into a de facto single-payer system. In fact, if I were in favor of a single-payer system I would be lobbying hard to establish this government option, since that's the vehicle by which my vision could be realized.

So now to me: if insurers can't compete because the government is more efficient, that's great. If insurers can't compete because the government just keeps slowly forcing them out of business via the rent-seeking behavior of everyone, that's bad. I personally think the latter is more likely.