When President Bush originally nominated Judge Samuel Alito to the Supreme Court, I commended him for picking a nominee whose beliefs and judicial philosophy were public knowlege. At the time, I wrote that this nomination would be about reading the tea leaves but rather about arguing over substantive issues. Little did I suspect that the Judge Alito and the White House would try to play down his conservative bona fides at every turn. The White House is actively pushing back against journalists who describe Alito as conservative. They probably should have stuck with stealth nominees. (Via Eric Umansky.)

Dahlia Lithwick is ringing a warning bell when it comes to Sam Alito and the war on terror:

I have written before that the arc of his Supreme Court nominations can best be explained by his desire to pack the courts for all of the Hamdan, Hamdi, and Padilla cases to be heard by the courts for years to come. Think about it: Roberts, Miers, and Alito each have a long track record of endorsing executive power. Each seems highly likely to strongly support the president’s claims to virtually limitless executive authority in wartime. The Bush administration saw that claim repudiated by a margin of 8-1 in Hamdi. And the president won’t let that happen again.

It won’t. How do I know? In his 15 years on the federal bench, Judge Samuel Alito has yet to rule on a case substantively involving the war on terror. But Alito’s votes in pending and future war on terror cases can be fairly accurately predicted. They lurk in dark alleys, near his decisions about criminal rights, immigration cases, and government power. Alito’s record in none of those areas bodes well for people who worry about the Bush administration’s push for unchecked war powers.

Since I started using a Powerbook regularly a couple of months ago, I’ve been trying to decide on which text editor to live in. For writing Java code, I continue to use Eclipse, but I needed to decide on something to use for HTML, PHP, Ruby, and other random stuff. First, a quick list of products I tried but decided not to stick with.

The first editor I tried was gvim. I use vim whenever I’m editing files within a console window, and under Windows I use gvim for everything but Java. On a Mac, though, gvim just doesn’t fit in. You don’t use a Mac every day to use an editor that’s a rough port of a command line application. I still use vim twenty times a day, but not as my GUI editor.

The next editor I tried was TextWrangler. About ten years ago I actually bought BBEdit for $100 bucks at MacWorld Boston, so I gave TextWrangler a shot. It was fine but I didn’t stick with it for very long, mainly because I became enchanted with SubEthaEdit. The idea of collaboratively editing documents with other people seemed really cool to me, so I started using SubEthaEdit. SubEthaEdit is a fine text editor, but I never got around to using it to collaborate with anyone. If I do need to use the collaboration features, I’ll go back to it immediately.

What I wound up with is TextMate, and for one reason only. In the Locomotive Ruby on Rails environment, there’s a menu item called Edit in TextMate for each of the applications it manages. When you use it, it automatically creates a TextMate project for your Rails applications and gets you started editing that project. From there it’s not hard to figure out how to set up projects for your other applications. This project-oriented approach is great for a developer. As I understand it, TextMate has many other cool features, but just the project view compelled me to buy a license. (The actual motivation for buying a license was the application no longer allowing me to click on the “Later” button when I launched it and it asked for a registration key.)

James Roberston has a good post summaring a debate between competing philosophies of language design. The subject is Java’s java.util.List versus Ruby’s Array class. Refactoring guru Martin Fowler argues that Ruby’s version is better because it includes bunches of convenience methods, and Elliotte Rusty Harold argues the converse, that fewer methods is better.

My contribution to this debate is to say that I was hacking on some Java code last night and found myself wishing that Java’s List class had methods for grabbing the first and last elements in the list. To me, the convenience methods make your code much more readable. In the Java world in particular, everybody uses editors that auto-complete method names anyway, so it’s not as though we need to memorize all of the public methods of every class we use in the first place.

Eric Umansky has posted the second entry in his torture policy glossary series, this time addressing the phrases “for the purpose of interrogation using torture” and “when we believe he will be tortured.”

Clive Thompson’s latest column on gaming for Wired News discusses a topic that I find persistently interesting, the uncanny valley, which he defines as well as anyone:

This paradoxical effect has a name: the “Uncanny Valley.” The concept comes from the Japanese roboticist Masahiro Mori, who argued that simulacra of humans seem lively and convincing so long as they’re relatively low-resolution. Think of history’s best comic strips: With only a few quick sketches on a page, Bill Watterson can create vivid emotions for the characters in Calvin and Hobbes. When an avatar is cartoonish, our brains fill in the gaps in the presentation to help them seem real.

I think that Blizzard, the creators of World of Warcraft, understand this effect very well. They took a lot of flack for the cartoonish avatars (like these or this guy), but doing so enabled them to avoid the uncanny valley effect almost completely. Compare those screen shots to this one EverQuest II, which was released at about the same time. The EverQuest II avatar is downright creepy — she looks sort of like a virtual Joan Rivers. I think that more and more game companies are going to choose to create characters that are lifelike rather than realistic. That approach has certainly served Pixar well.

News.com has an article on Wikipedia’s two incidents last week. The first was John Siegenthaler’s op-ed for USA Today explaining that someone had written a Wikipedia article about him that stated that he was at one time thought to be complicit in the assassinations of both John and Robert Kennedy which had gone uncorrected until he noticed it himself. The second was the flap over Adam Curry’s rewriting the article on podcasting to suit his own recollection, eliminating mention of some contributions made by others.

To me, in the Siegenthaler case, Wikipedia flat out failed. An anonymous user posted defamatory information about a public figure and nobody caught it for months. Had Siegenthaler himself not found the article, who knows when it would have been corrected, if ever?

In the Adam Curry case, I think the system worked pretty well. Curry went in and edited the article (anonymously as well). People caught the edits and furthermore identified Curry as the editor by tracking him down from the IP address associated with the edits. The record has been corrected and hopefully everyone involved has learned a lesson.

I’m not sure how to solve the problem that burned Siegenthaler. Wikipedia is going to prevent anonymous users from posting new articles as of this week, but that doesn’t change the fact that obscure articles are hard to fact check. Generally speaking, Wikipedia editors should probably remove accusations like the one in the Siegenthaler article unless they are accompanied by credible references.

Here’s Condoleeza Rice’s statement this morning on extraordinary rendition:

In conducting such renditions, it is the policy of the United States, and I presume of any other democracies who use this procedure, to comply with its laws and comply with its treaty obligations, including those under the Convention Against Torture. Torture is a term that is defined by law. We rely on our law to govern our operations. The United States does not permit, tolerate, or condone torture under any circumstances. Moreover, in accordance with the policy of this administration:

— The United States has respected — and will continue to respect — the sovereignty of other countries.

— The United States does not transport, and has not transported, detainees from one country to another for the purpose of interrogation using torture.

— The United States does not use the airspace or the airports of any country for the purpose of transporting a detainee to a country where he or she will be tortured.

— The United States has not transported anyone, and will not transport anyone, to a country when we believe he will be tortured. Where appropriate, the United States seeks assurances that transferred persons will not be tortured.

There’s not a whole lot of wiggle room in those statements, is there? So that only leaves one question. If all of those statements are true, why do prisoners who we subject to rendition keep getting tortured?

Update: Eric Umansky explains what “respects the sovereignty” means in context.

I’ve been following the ongoing debate over the One Laptop Per Child project with interest. I’ve generally been in the camp that says that if we want to help the world’s poorest people, our resources should go toward providing reliable sources of clean water, addressing preventable disease, and addressing conflict and famine. The technology divide is important, but aid dollars are scarce, so we should maximize our bang for the buck.

The thing is, though, that this project is not targetted at the world’s poorest people. If you read the FAQ, you see that the countries initially being targetted are Brazil, Egypt, Thailand, and China. Most of the critiques of this project point out the absurdity of devoting resources to giving laptops to people in places like Niger as opposed to addressing more pressing needs, but that’s not what the One Laptop Per Child Project is about, at least right now.

On the other hand, there are plenty of considerations that need to be taken into account, even with the project as proposed. The Fonly Institute has issued a critique that’s worth reading (and they’re not just rock throwers either).

To me, the biggest issue with the project is the proposed scope. While the idea of producing and distributing millions of laptops to every child in a particular country is grand and ambitious, I don’t think it makes the most sense, at least at first. It is undoubtable that there are going to be a number of flaws in the system that will have to be addressed, including corruption, technical issues, and cultural issues (most of which are discussed in the Fonly post). So why build and distribute millions of laptops before discovering and addressing those issues. Why not start producing the laptops and selling them at cost (even if the lower volume drives up the price) and seeing who’s interested and what they use them for? Why not distribute them to some kids in America to see how they can be used as an instructional tool?

I think that in countries where computer literacy can open a much wider variety of career options to you, providing more children with the option of becoming computer literate is a great idea. As I’ve said before, had I not had access to a computer when I was a kid, I probably wouldn’t have nearly as interesting a life as I do today. I just think that the project is more likely to succeed if they take a phased approach.

I’m thinking about trying out Dreamhost for Web hosting. I’m definitely not staying with TextDrive — I had to wait two hours to post this because the server crashed again. Unacceptable. I’ve heard good things about Dreamhost and did the requisite “dreamhost sucks” Google search to find contrary opinons. There are definitely some customers out there who have had bad experiences, but they seem to be in the minority.

The number one question for right now is, can I run my own install of Movable Type 3.2 that uses cgiwrap or suexec, so that I don’t have to make my directories world writable if I want scripts to be able to store files in them?

The other question is, of course, how reliable and responsive are they? They have an emergency status weblog which has few entries, either because they only report really big problems or don’t have that many problems. (You can contrast it with the TextDrive Status page or Pair.com’s system notices.)

By the way, you can see the difference between pair.com and TextDrive from the respective status pages. Pair.com has hundreds of servers, and yet when a server crashes (or even requires replacement of hardware), it’s rarely down more than 10 minutes. My server at TextDrive crashed this morning, and it was down for over two hours. Here’s a status update from pair.com:

eiga (www326) encountered an error and reset. While it was down, a new chassis was put into service. The new specifications are a Pentium 4 3.8GHz with 2GB of RAM. Downtime was under 15 minutes.

The obvious question you’d ask yourself is why I don’t just go back to pair.com? (I could easily do so. In fact, I didn’t even cancel my account, I just downgraded to the cheapest package.) The answer is that I had a rough time getting Movable Type working there, and if possible, I’d rather pay Dreamhost $8 a month than pay pair.com the roughly $33 per month that I used to pay.

I’m trying to do more due diligence this time, though, so if you hate Dreamhost, now’s your chance to get it off your chest.