31 December 2008

Recent favorite things

Neal Stephenson's Anathem. I was a little wary after The Baroque Cycle, but I really enjoyed this— it was a real page-turner. Stephenson has created such a rich and lush world for this novel that you just don't want it to end. And there are dozens of fascinating ideas scattered throughout the story.

World of Goo (WiiWare; $15). "If I had played this game as a kid I might be a civil engineer now." World of Goo is an addictive physics-based puzzle game. You connect elastic "goo balls" to build structures to reach a goal while avoiding various obstacles. It's not as easy as it sounds because the structures sag under their own weight. The game physics are astoundingly realistic—so when your tower of goo balls collapses, you get the sense that it was your own darned fault. However, the levels are really well-designed and many of them require these critical epiphanies. The interesting thing about World of Goo is that it is surprisingly deep. As you play more and more, you hone your intuition for the game physics. So after you reach the primary goal in each level, you can come back later and play it again for the "OCD" point, which requires you to reach the goal in an even more efficient manner. World of Goo was written by, like, three guys in a garage, and yet it is very polished. Highly recommended.

I'm in the middle of Brian Greene's The Fabric of the Cosmos. Greene has a gift for developing accessible (and faithful) analogies that expose the core ideas—but not the math—behind physical phenomena, such as relativity and quantum mechanics. Highly recommended for laypeople.

28 December 2008

The inexorable march of Moore's law

I think this is the first time I've ever noticed something measured in "terabytes" and thought it too small.

I noticed that the ext3 filesystem has a maximum filesystem size of 16TB. You can buy 1.5 TB hard drives at Fry's today, so you can probably have a 16TB disk array under your desk in the near future.

Fortunately, ext4, which was just marked stable, will support a maximum filesystem size of 1 EB (1024 PB, or 1,048,576 TB). Whew!

(This blog just turned one year old, and this is its 75th post!)

27 December 2008

Where is your humanity?

The AP reports:

PHILADELPHIA (AP) — A man enraged by a noisy family sitting near him in a movie theater on Christmas night shot the father of the family in the arm, police said.

21 December 2008

SCU psychologist replicates Milgram's famous experiment

The San Jose Mercury News, among others, reports:

Replicating one of the most controversial behavioral experiments in history, a Santa Clara University psychologist has found that people will follow orders from an authority figure to administer what they believe are painful electric shocks.

The setup was essentially the same as in Stanley Milgram's 1974 experiment. The conclusions are the same, too.

Additionally, the volunteers confronted a novel situation—having never before been in such a setting, they had no idea of how they were supposed to act, he said.

Finally, they had been told that they should not feel responsible for inflicting pain; rather, the "instructor" was accountable. "Lack of feeling responsible can lead people to act in ways that they might otherwise not, said Burger.

"When we see people acting out of character, the first thing we should ask is: 'What's going on in this situation?'"

Milgram's assessment of his own experiment seems no less true today:

Ordinary people, simply doing their jobs, and without any particular hostility on their part, can become agents in a terrible destructive process. Moreover, even when the destructive effects of their work become patently clear, and they are asked to carry out actions incompatible with fundamental standards of morality, relatively few people have the resources needed to resist authority.

19 December 2008

A gross misallocation of resources

Paul Krugman argues that the nonsensical accounting and compensation systems in the financial "services" industry allowed money managers to keep reaping big bonuses even as their clients lost everything. He suggests that the primary difference between Bernard Madoff and his colleagues was that

...while Mr. Madoff was apparently a self-conscious fraud, many people on Wall Street believed their own hype.

Well, shame on us for entrusting our money to people who were not actually held accountable and to a totally opaque system that allowed the managers to spin their numbers however they wanted. Not a winning plan.

But the really tragic part, in my view—the part where our losses may not be so easily computed—is the that the growth of the financial services industry produced a distorted view of what is really valuable:

[How] much has our nation's future been damaged by the magnetic pull of quick personal wealth, which for years has drawn many of our best and brightest young people into investment banking, at the expense of science, public service and just about everything else?

05 December 2008

Token-based authentication coming to credit cards

Credit card numbers, like all numbers, are just too easy to steal. And it doesn't matter if you have to steal 19 digits instead of 16.

Visa is going to test out credit cards with a tiny keypad and display built in. You have to enter your PIN to get a one-time password to get your online transaction to go through. This is pretty nifty.

The root problem (people copying your card) still exists for physical magstripe cards, though. This is a huge advantage that RFID has over magstripe and I wonder why people haven't made a bigger deal out of it. Once you've read the magstripe off a card, you have all that there is to know about the card. But an RFID chip can contain active machinery inside. It's a black box that you can only interact with in certain ways, unless you reverse-engineer the hardware to figure out what the silicon is doing (this is not super easy). So you can implement a challenge-response mechanism which resists some kinds of man-in-the-middle attacks (remember the hubbub about fake ATMs?).

04 December 2008

Mathematical humor

The Ham sandwich theorem states that:

Given n "objects" in n-dimensional space, it is possible to divide all of them in half in volume with a single (n - 1)-dimensional hyperplane.

The theorem gets its name because in the case n = 3, it implies that if I have a ham sandwich consisting of arbitrarily shaped pieces of ham, cheese, and bread (hah!), and an obsessive-compulsive friend, I can cut the sandwich in half with one swipe so that each of us gets exactly half the ham, half the cheese, and half the bread.

Cute, huh?