Mind Poison

Alex calls me out on some sloppy google research. I shot myself in the foot with the old Kb vs. KB thing and granted Bluetooth 8 times the bandwidth it actually has.

How embarassing.

(And in front of so many people).

That does pretty much sink the wireless idea. However, I'd just like to say, I wouldn't have thought the camera would transfer the full, uncompressed image, but rather the JPEG. Given that most of my camera's photos come out at around 600K, I guess you can see why I thought Bluetooth would be feasible, given my misunderstanding its performance.

Also, a while ago, I heard some rumor of some new WiFi standard being worked on to provide a high-speed personal wireless network, like Bluetooth, but better suited to something like this. It was like 802.something, as I recall, but I might well be hallucinating it. I can't seem to find anything about this anymore, since I have no idea what it would have been called. Ah, those vague, convenient rumors.

Finally, I'd like to point out that the wirelessness was not the main point of the post. If they could figure out a way to jam a more universal port onto the iPod (USB2 or Firewire), so that people could connect up the camera, it would still work about the same; you'd still have your iPod in your pocket. The main idea was about what the iPod could become, and how it could better integrate with all sorts of other electronic devices to provide storage for them.

And, I want it to be a cell phone, damnit.

Apple just announced their newest iPods, the iPod U2 Special Edition and the iPod Photo. Personally, neither of these is particularly appealing to me. The U2 iPod looks alright, but I think the back looks funny still being silver, it doesn't quite work with the front as well as the white does. They probably didn't want to mess with it too much, though.

The iPod Photo similarly doesn't do anything for me. It was probably inevitable that they'd stick a color screen in there; it was going to happen sooner or later, whether it was them or their competitors leading the way. However, the idea of being able to view my photo collection on a little 2 inch diagonal screen doesn't really excite me that much (neither does its increased size and weight).

Still, I think this could be a good first step towards what the iPod can become: a personal portable hard drive. Obviously, the real usage scenario for iPod Photo is plugging it into your TV or computer to show the photos in a way that someone might actually want to see them. That's fine, but what I really want to be able to do is have an iPod in my pocket and have my digital camera automatically send my photos to it, like it's an outside-the-camera 40GB roll of film. Since I don't want to have to carry around a cable for every given type of device I might want to connect to, I'd also like it if the iPod would work wirelessly. Given its bandwidth and the proximity the user will usually have to the device being used, Bluetooth is probably a great match for the iPod.

Of course, then that begs the question of what else you can do if everyone is carrying around a wireless hard drive in their pocket. For one thing, it strikes me that my cellphone (a small Nokia) would fit entirely inside my iPod, if you took the case off. Most of that smaller piece is a battery and the screen. In a few generations, could the chips used in the iPod be small enough to include mobile phone chips and an antenna, so I don't have to haul around both my cell phone and iPod (obviously, the battery and screen would be shared)? Let us hope. A small redesign to the iPod headphones could include one of those dangling microphones on the wire. Or, Apple could use a Bluetooth headset, similar to what is currently available for Bluetooth phones. I'm sure Apple could work out a nice interface for dialing with the scrollwheel (although I find that I'm usually dialing out of my contacts list on my cellphone, which is what I'd prefer if I could). Plus, it could stop me from missing cell phone calls while I'm walking around listening to music, which must be amusing to bystanders every time I walk by oblivious to the ringing cell phone in my pocket.

I hate speculation pieces like this, because the writer always assumes that R&D and production are free. Of course I want a cheaper, faster iPod with more storage and a higher-res screen that does the work of any electronic appliance you could think of! That's not what I'm trying to say, because I understand that's not helpful. However, I think there's a very real opening here for Apple to create and own a potentially huge new market in the, well, personal hard drive/universal connector mark...the "digital pod" market. As a matter of fact, I think it's either very forward-thinking or fortunate that Apple gave the iPod such a vague name, when it could have called it something more music-centered.

Lead the way, Apple! Earn that $600!

As I've said before, I don't have Flash installed on Firefox, because nothing good ever comes of Flash. It only leads to obnoxious movie previews and commercials making noise in your browser window while you're trying to read, or gigantic ads that obscure the page until you dismiss them. Well, almost nothing good comes of Flash. Occasionally, you want to check out Strongbad Email, or something like that. Then, it's too much trouble to bother, since you have to open IE, and deal with that whole unpleasant experience.

Well, the answer is Flash Click to View. This extension prevents every Flash object from loading, and instead replaces it with a play button. If you want to see a particular Flash movie, you just click the button. Otherwise, your browsing experience remains just as quiet and pleasant as if you didn't have Flash installed.

This was actually the first time I'd ever installed a Firefox extension, and I was pleasantly surprised. Firefox itself takes you to their extension repository, and a simple click on "Install Now" does all the work. It just tells you that you need to restart Firefox before it can take effect.

Similarly, installing Flash was easier than on IE. I really like the Firefox tells you that you don't have a viewer for a certain page object (with the yellow bar across the top of the page and the button that lets you get the viewer), but today I got to tell Firefox to load the Flash plugin. You never have to deal with Macromedia's site or an installer, or anything. Firefox hides the whole thing from you. All you need to do is agree to the license, watch the progress bar, and restart Firefox.

Incredible.

I had to let her know that the reason she'd never heard of me was because I was famous
I'd hate for you to think that I read Slashdot interviews (I don't), but I do read what Neal Stephenson writes. This put me in a kind of difficult situation when I saw a Slashdot interview with Neal Stephenson, but ties go to Neal.

Good Bad Attitude
OK, don't think that I read Slashdot... This is really sort of a self-serving essay: us hackers, we're so amazing, and the world just doesn't get us because they're squares. That would turn me off, if it weren't for the fact that I think he's absolutely right about the subversive attitude he's talking about.

In other news, I got an email from Jamie Zawinski today, as part of an email he'd sent out to a bunch of the major Gnome and GTK bigwigs. He was linking to an article he'd just put up on his website, discussing why XScreensaver doesn't use a widely used toolkit. I can't imagine how I got on this list along with Nat Friedman, Miguel de Icaza, Jeff Waugh, Davyd Madeley, and others, but it was definitely a manually-entered list and not some mailing list forward. I know I've never emailed Jamie before, so I thought that was pretty funny.

Surprisingly, I have found a library that works almost exactly as I would like. It's called ding, after the author. Despite some style differences, his library is almost perfect. The matrix class is parameterized on type, but not on dimension, which is what I was looking for. It doesn't seem to do any heavy-weight sparse-matrix stuff either. It has nicely defined operators, and some nice output facilities.

Looking around at his site, I see why it probably matches my needs so closely. Yongling Ding is a Finance PhD, and he bills his library as a quantitative finance library. It doesn't seem very complete for that specific task, but even on the basics it's obvious that the main thrust of what we want to do with it is similar.

I wish I'd found it sooner, so I didn't have to implement my own versions of all those functions. So, in an attempt to prevent this horrible fate from befalling another, I am mobilizing my 4 regular readers, and attempting to give him some googlejuice.

What I want to be doing right now, instead of writing this post, is writing some C++ code. Well, let me rephrase that. I want to be writing some high-speed numerical code. I honestly gave Java and C# a really honest chance last night, but it quickly became apparent that there just aren't any decent matrix math libraries for them. That's how much I wanted to avoid C++, I was willing to deal with the performance of Java for this problem. If you're about to snicker at the fact that I would consider those solutions to be "high performance," shut up until you read the rest of this. Besides, even Java would still be a major improvement over Matlab.

There's some Matlab code we have to do a Geweke-Hajivassiliou-Keane (GHK) simulation to find the probability that an event that is distributed multivariate normally is within a given set of bounds. This is actually a pretty neat thing to do. Because the pdf of a normal distribution cannot be integrated analytically, calculating the cdf requires numerical integration. Normally, this is fine, but when you get a multivariate normal distribution (that is, a vector random variable that is jointly normally distributed, as opposed to a vector of independently normal random variables), after three dimensions it becomes impractical to do this integration. It is illegal, in economics or statistics literature, to mention this problem without also mentioning that it is called "the curse of dimensionality."

The solution, in practical terms, is to generate random variables that are distributed according to the distribution you are interested in, and then see how many of those fall in the rectangle you are interested in. As you perform this iterative process, you will converge on the probability you are interested in. The GHK algorithm is an improvement on the "crude" version I just described that uses some clever weighting to get a better estimate with fewer random draws.

So, since the Matlab code was taking 31 hours to run on the dataset of interest, and despite the fact that my optimization efforts cut that down by two thirds in Matlab, I thought I'd try to see if I could improve on that with a lower-level language.

I searched long and hard for some Java or C# matrix math libraries. I found some, of course, but nothing I felt comfortable with investing time in, because they looked too half-baked or bit-rotten, or whatever. So, I finally came to terms with the fact that I was going to have to use C++, because I knew there were many math libraries freely available for it.

The first thing I realized is that while there are some good libraries for this stuff, they're not really for the uses I was thinking of. Many of them are for creating gigantic sparse matrices, and that was just too heavy-weight. I'd never see a speed improvement for a much smaller matrix like I was working with. Eventually I invested some time in a highly lauded library called Blitz++. It promised good speed and some very complete matrix classes.

However, after I'd familiarized myself with the library and its very flexible matrix implementations, I found that it had no matrix multiply routine. This struck me as a rather strange omission, and I blew over an hour poking around because I just couldn't believe it wouldn't have it. Of course, it really didn't. They offered some very clever hack involving their tensor notation and the associated operators which in practice executes in O(n^4) time, no thank you. So, I wrote my own matrix multiply routine, no big deal, I've done it a million times.

However, what quickly became a problem was all the C++ template hackery that the Blitz++ fellows think is so clever. It isn't. It's painful. Compiling my little 150 line program takes more than half a minute, due to all that template crap. If there's a simple bug, it could very well generate 40 pages of meaningless error messages because every phase of the compiler is totally confused. Using Blitz++ drove me slowly insane, and now I don't really know what to use. There are some other math libraries I found, but none of them have the specific operations that I was hoping to get pre-written, as far as I can tell. The more complete ones seem to promise more template nonsense as well.

It makes you want to use Java or C#, is really what it does. I'm sure the performance would still have been a major improvement over the entirely interpreted (and notoriously slow) Matlab, and I might well have been done by now because I could have just written the damn code, instead of screwing around with libraries all day.

I'm somewhat at a loss for how to describe this movie. Most of the things I normally key in on in movies just don't apply this movie (like cinematography, directing, acting, and writing). Still, much like its (spiritual) predecessor, South Park: Bigger, Longer & Uncut, it's a brilliant film.

Some people, of course, won't be able to decide if they want to see this movie until they know whose side it's on. Given that it's really hard to pin the movie down on anything, that might, along with its R rating, explain why it opened third in the box office. I've seen some complain that the left is given a harder time by this movie, but I disagree. It's true that Bush and his advisors are entirely absent from the movie, but I can easily see it being the case that whoever greenlighted this movie didn't want to risk the box office consequences of cutting themselves off entirely from Bush supporters (after already cutting themselves off from people who don't think they want to watch marionettes, and people under 17). The criticism of the war on terror couldn't have been clearer or more piercing to me. The members of Team America talk like police on COPS might, showing up in helicopters with sirens on them, and blowing up much of major foreign cities in their bumbling pursuit of terrorists. Then after finally killing the terrorists (and razing much of, say, Paris), they stand around, waiting to be fawned over by the locals, who are about to do no such thing. You can't fail to pick up the idea that fighting terrorists through bombing cities largely effects third parties who have nothing to do with the two parties in actual conflict.

In any case, it's ridiculous to argue about who has it easier in this movie. No one who agrees with either extreme is going to declare the movie's verdict fair. Although I very much respect Parker and Stone's intelligence, and I respect the freedom comedy has to tell the truth, the much more important feature of this movie is that it's hilarious. It's the funniest movie I've seen in ages, and everyone in the audience obviously thought so too. The main character's final speech about the war on terror must be seen.

Update: I hate to admit I read cnn.com, but this article is related to some of the things I talked about above.

I've always wondered how the east coast can have such a strong concentration of the first-rate schools. On the east coast, you'll find tons of top schools that people just have to go to: Harvard, MIT, Yale, Columbia, Cornell, Princeton, UPenn, Chicago, Michigan, Brown, NYU, Dartmouth, Northwestern, Duke, UVA, and plenty of others. The west coast has Stanford, CalTech, Berkeley, UCLA, and maybe UCSD. That's what you've got if you're, say, west of North Dakota (And I'm counting universities here, there's plenty of well-respected liberal arts colleges on the west coast, though I suspect those are also outweighed by the east coast). Only five of the top 30 schools are west of Texas, according to US News.

This isn't really surprising, given that the east coast has been settled for much longer, but it is a bit strange that most of the top students on the west half of the country get shipped to the other side. You wouldn't expect it to last over time. Berkeley, UCLA, and UCSD are big schools, but that can't possibly make up for it. The workhorses of the west, the "midrange" mentioned above, are the other 7 UC schools. No other state school system is well respected on this half of the country, and the California State system isn't very good.

Given this lopsided distribution, it's only natural that over time, more schools in the west would slowly evolve and move up the ranks, right? Having gone to one of those workhorse schools, I'm glad to see it. In fact, you might have expected this to happen at USCB: How many other schools can give professors offices with an ocean view? It probably has a better faculty than you might otherwise think it could afford.

I don't know if UCSB can ever get past its reputation, but it's pretty clear that it is improving dramatically (in reputation, at least; clearly if it's being considered a better school, it's been a better school for some time). Indeed, for the small difference in the quality of education, there's a large difference in tuition between the UCs and the Ivy Leagues. That cost-effectiveness is in itself something to be proud of, I think, but it clearly doesn't translate into respect or earnings. Just try telling it to an Ivy Leaguer; they'll nod and smile but wouldn't be caught dead there. Perhaps some day the "other" UCs will be as well-regarded as the "other" east coast schools, and provide a stronger representation for the western half of the country on the lists of the top schools.

What a wonderful thing that would be. Another coast filled with excellent schools, but without the aristocratic names and self-importance. Instead of old neo-classical buildings and snow, simple, laid-back architecture in paradise, and with respect.

This is a powerful movie, and it might be one of my favorite sports movies. However, that's probably not for the reason that most people like sports movies. Typically, sports movies are about struggle, overcoming adversity, persistence, and eventually victory, or something like that. A sports movie where the main characters don't win is almost unheard of.

If you aren't paying attention, you might think that Friday Night Lights is just that sort of movie. However, what it's really about is having an unhealthy attitude towards sports. The movie is quite clear about where it stands on this, as announcers complain that not enough money has been spent on the high school's football stadium, and superstar kids can barely read their college recruitment letters. The kids in the movie are told that they don't have any opportunities aside from football, and they believe it. Because they believe it, it's true, and it's heartbreaking.

Aside from that, it's a pretty typical sports movie. Billy Bob Thornton is great as the football coach (but then, saying that is like saying that Courtney Love is an amazing actress when she plays a crackwhore), and all the other Texans put in very natural performances. The movie depends on that authenticity, because it's very much not in the traditional Hollywood style of, say, Jerry Maguire. The camera is frequently handheld, and it cut very quickly, like a music video. Some parts of it just seem like a highlight reel, fast-forwarding through time. Of course, all sports movies do that, but this one makes doesn't try as hard to hide it. That's not a criticism, either, it works well.

Here's just a list of several interesting articles and blog posts I've read recently around the web.

Thank You, William H. Meckling
I found this article about the economist(s) who killed the draft on Marginal Revolution. It's a great read. Here's the heroic Milton Friedman shaming a general:

Of course, Meckling wasn't the only hero. Milton Friedman was very persuasive. One of Meckling's favorite stories, which his widow, Becky, recalled in a recent interview, was of an exchange between Mr. Friedman and General William Westmoreland, then commander of all U.S. troops in Vietnam. In his testimony before the commission, Mr. Westmoreland said he did not want to command an army of mercenaries. Mr. Friedman interrupted, "General, would you rather command an army of slaves?" Mr. Westmoreland replied, "I don't like to hear our patriotic draftees referred to as slaves." Mr. Friedman then retorted, "I don't like to hear our patriotic volunteers referred to as mercenaries. If they are mercenaries, then I, sir, am a mercenary professor, and you, sir, are a mercenary general; we are served by mercenary physicians, we use a mercenary lawyer, and we get our meat from a mercenary butcher."

Linux Kernel 2.6: It's Worth More!
In this paper, David Wheeler uses some of Barry Boehm's constructive cost models (COCOMO) to estimate the cost of privately redeveloping the Linux kernel from scratch. He arrives at an answer in the neighborhood of $612 million dollars, though he says (and I completely agree) that this is probably only a lower bound. It's almost certainly much higher than that. Previously, David Wheeler wrote a similar paper called More Than a Gigabuck, which attempts to estimate the same statistic for an entire RedHat distribution (as the title suggests, it comes out to more than $1 billion dollars, though I would again be surprised if it was actually that low). The much more interesting question is how much Linux is worth, but that's much harder to estimate (Slashdot, of course, can't tell the difference).

How to Watch Tonight's Debate
I largely agree with Timothy Noah's sentiment, but I think he's wrong in practice. When talking about the added expenses of Bush's social security privatization plan, he's being simplistic. It's true that privatizing social security will require a lot of borrowing (I believe those figures). But the longer you wait to do it, the more expensive it gets! Since John Kerry doesn't want to privatize it, he gets to count the transition costs as zero, but that masks what is really going to happen. In fact, the Concord Coalition document that Noah links to says about as much:

However, the key issue in evaluating the fiscal implications of a Social Security reform plan is not its immediate 10-year cost but whether it achieves long-term sustainability. Both the costs and benefits of reform should be assessed over a time frame that goes well beyond the next decade. Incurring a modest near-term budget cost as part of the transition to a genuinely funded and sustainable system may not be fiscally irresponsible.

The Concord Coalition then goes on to say that they don't think Bush's plan will achieve long-term sustainiability. However, I still take issue with the way people account for the costs of inaction. Arnold Kling explains much better here and here. Note that Kling assumes there is no cost to increased government borrowing, and that this simply shifts who pays when; this ignores the effects of crowding out, which can have contractionary effects. Still, it's better than doing nothing, but nothing is probably what we'll get:

The more dire the long term problem grows, the greater the cost, and the political resistance, of doing something to fix it now. So, for instance, debate over private account reforms for Social Security stalls on the size of "transition costs," which are much less than the cost of not reforming, but must be borne in the present. The question now is whether we've passed the point of no return: Whether the near-term pain required to fix things will continue to multiply with the long-term cost of inaction at a rate that keeps us locked on course. Perhaps, sometime soon, citizens will be struck with the spirit of civic and intergenerational responsibility, bite the bullet and clean up the mess their predecessors made. Or perhaps, if we're lucky, the scam will last long enough for us to pass the bill on to our kids.

Justice Dept. wants new antipiracy powers
Last friday, just after the debate, we watched Real Time with Bill Maher. Fox News's Tony Snow and the asshat Lieutenant Governor of The Worst State in the Union lost their shit when Bill Maher pointed to the recent corporate tax cut and said that nothing has changed since September 11th. I don't fully agree with Bill, but I think he definitely has a point. If our priorities have changed so much, then do we really need to increase funding for the FBI so they can hire more agents to crack down on illegal song downloading?

"The department is prepared to build the strongest, most aggressive legal assault against intellectual-property crime in our nation's history," Attorney General John Ashcroft, who created the task force in March, said at a press conference in Los Angeles on Tuesday afternoon.
In an example of the Justice Department's hunger for new copyright-related police powers, the report asks Congress to introduce legislation that would permit wiretaps to be used in investigating serious intellectual-property offenses and that would create a new crime of the "importation" of pirated products. It also suggests stationing FBI agents and prosecutors in Hong Kong and Budapest, Hungary, to aid local officials and "develop training programs on intellectual-property enforcement."

It basically speaks for itself. If September 11th had really changed our priorities, we might stop this nonsense (and plenty of other nonsense) because we have other things we urgently need to spend money on that will give us a higher return to our safety and happiness.

Mickey Kaus made a post today that gave me an idea. Gerrymandering is an old problem, and it continues to be a huge problem. Redistricting is necessary from time to time for ensuring that district boundaries are drawn fairly with respect to the changing concentration of the population, but the problem is that the people who do the redistricting are the politicians affected by it. It's been said that gerrymandering lets politicians choose their voters, instead of letting voters choose their politicians.

What if we could come up with an algorithm that creates district boundaries, and all states were required by law to use that algorithm for creating districts? That is, what if we could create an algorithm that takes as input the geographic shape of the state and information on current voter densities and outputs a set of district boundaries? That would take the control entirely out of the hands of any partisan force and place it instead in the hands of mathematics and the voters (by virtue of where they choose to live). Notice that it doesn't take in any information on what party the voters are registered with. To borrow from John Rawls, it would be a set of district boundaries that everyone could agree to if they didn't know what party they were in, only it works by not knowing the party registration of the voters. In essence, a tangible, real-life veil of ignorance.

Of course, there are difficulties. When I took data structures and algorithms, one of the projects involved implementing several different algorithms for taking a group of points distributed on the x,y plane, and clustering them into n groups. Constructing the optimal clustering is not computationally feasible (with all the footnotes that go with that), so these algorithms were mostly based on heuristics and didn't always generate good results (and they could still take a long time to run).

I imagine with some research, it should be possible to construct an algorithm that has desirable properties. For example, it should create districts of roughly equal size in terms of population, connected, non-overlapping, not too concave, and it should try to make district boundaries go through sparsely populated areas. It is essential that it not require any "tuning parameters" or human input (except in the case that it needs firmly set parameters to be tuned once and then used everywhere). That's too many goals, and they can be contrary to each other, I know. Still, the clustering doesn't have to be optimal, it just has to be unbiased and better than what currently happens. Ideally, the key properties and performance boundaries of the algorithm should be proven mathematically, so everyone can be comfortable with it. With some work, I think something promising could be done.

However, I think even if it is possible, it won't be done for the obvious reasons: it requires the people who are affected by it to agree to it, which is the original problem. It's too bad, it would be an interesting combination of law and technology.

Update: Macneil writes in to point out that this is actually an old idea. For example, see here. I did some google searches last night, but nothing came up. Now I'm finding all sorts of stuff, so I must have been using the wrong search terms.

Today marks the beginning of the endgame for the 2004 presidential election. This morning, at 8:30am, the BLS released the last employment situation report before the election. Bush supporters had placed high hopes on this report finally showing strong job growth, especially because the BLS was due to issue a revision to the data for the year leading up to March 2004. The Bush administration had been expecting an upward revision of about 300,000 to 400,000 jobs. This jobs report didn't get close to fulfilling their hopes, as the upward revision is smaller than forecasted (and smaller than the average upward revision in past years), this month's job growth was weak (96,000), and last month's was revised downward. A good news strikeout.

The candidates are going to have to react quickly, because their second debate is tonight. I don't know how people weigh the different debates, but I imagine the earlier ones are probably more important to people. By the third, they've probably seen what they were watching to see, so unless someone makes a major mistake, it's probably the least important. We'll see if Bush can reverse last thursday's performance, and if Kerry can turn in a similarly strong one. By tonight, we should have a better picture of where the election is going.

By the way, this story is hard to believe, exactly the sort of nonsense that Salon mass produces. The photo is hard to argue with, but who knows? They cite Bush's "let me finish!" statement as evidence, but I don't see why he couldn't have been responding to Jim Lehrer looking like he was about to turn it over to Kerry, which is what I assumed at the time. I don't see the mainstream media picking up on this story anytime soon. Still, I'll be watching the space between Bush's shoulder blades with interest tonight. (By the way, I got that link from electoral-vote.com, so please don't think I actually read Salon)

Now that it has been a couple of days since the debate, some better polls have been conducted about who won the Vice Presidential debate. The results firmly suggest that Cheney won (link from electoral-vote.com). However, those states and cities also agree in similar numbers that Kerry won the first presidential debate.

Damnit, shoulda stuck to my guns!

The other day, J. Bradford Delong, an economist a Berkeley articulated a really good reason to privatize social security. Actually, he wrote about several good reasons, but since this is actually an iPod post, I'm only interested in one:

At present, your Social Security benefits are yours only by grace of Congress: Congress could cut them if it wished. But if your privatized Social Security account were *yours*, then it would be yours not by grace of Congress but by right of property: courts would stand ready to defend it against any casual attempt to cut or confiscate it.

(Arnold Kling discusses the issue in more detail here, if you're interested).

The other day, I was thinking what a strange thing it is that I still have yet to buy any music from Apple's iTunes Music Store. I'm a big fan of Apple's, I am in favor of intellectual property rights for digital music files, and I think self-protection mechanisms are probably the best solution. I'm all about their pricing. So if I'm such an ideal customer for them, why can't I bring myself to do it?

One big reason is very similar to what Brad Delong brings up. When you download a song from iTunes Music Store, you don't really have a very firm guarantee about what you're buying. They do spell out what the software allows you to do, but they can change that retroactively. In fact, they have done that. If you purchased a song with the first version of the store, you were able to burn a given file to CD ten times. Subsequent versions have brought that number down to seven. That's hardly a dealbreaker, since I don't even plan to do it once.

Still, it makes me uneasy. If I pay for an album, what exactly am I buying? Purchasing a song from iTunes Music Store (and let me make clear that I realize that this is not entirely up to Apple; most of these criticisms should be read as being directed against the RIAA where appropriate) is actually the purchase of a usage license. You don't have any rights except what they give you, and you certainly can't consider your file a piece of your own property. In the license agreement they reserve the right to change your rights retroactively at any time, so I don't really feel comfortable with the purchase.

Does commercial song licensing happen this way? If you're a movie producer, and you want to license the latest hit song for your blockbuster, do they give you the rights, take your money, and then warn you that if they want, they can instantaneously change the terms of your usage agreement? You would feel comfortable with that? I doubt it. Yet that's exactly what they expect you to do when you download music from them, only one dollar at a time.

Then on top of that, they tell you that you are not in fact buying a license to use the music, you're buying a license to use that one particular file they send you. If your hard drive crashes, well, you're just gonna have to buy a new license so you can download it again! In other words, they're not about giving you the increased convenience that digital files give everyone else, they want you to inconvenience yourself by manually backing up your gigantic music collection to dozens of CDs, or pay to download everything again (having people buy things over and over again is a big theme with the record industry). I can understand that they have bandwidth costs, but losing a hard disk is not the uncommon event that a house fire that destroys your CD collection is. It happens all the time, and I'm not sure most people really understand how to back up their files.

I know the record industry is barely comfortable with the fact that everyone in the world can run around willy-nilly without a 1984 big-brother screen in their living room, automatically charging people when they hear a song. It's a tough life. But I think that if they would make some changes to their policies, they could get a lot more people onboard.

The way iTunes organizes your music is fantastic, and for me, CDs have basically become a distribution system for my mp3 files. When a CD comes out, I'm faced with a choice between paying $9.99 for the it at iTunes Music Store, and going to the record store a few blocks away and getting it for around $12.99 or $13.99. I'm still choosing the record store, and the reason why is directly related to their licensing terms.

With the CD, I do pay more. But I also get a lot more for my money. If I drop my laptop and the disk breaks, I can just rerip the CD without spending any more money (than the replacement laptop is already going to cost me). I can play the CD on as many computers or devices as I like, including my car, without the hassle of burning CDs or other contortions. And, I can lend the CD to my friends, if I think they might like it; who wants to be unable to lend their friends a CD? (It's stating the obvious to say that the 30-second song clips on iTunes are practically useless compared to getting a couple of good listens to the full CD). In short, the CD is worth more because it conveys usage rights that are a strict superset of the rights given to me by the license agreement. And they cannot revoke those rights after the purchase of the CD. Is it any wonder that people pay the few extra dollars to buy a CD and not have to worry about the inconvenience and uncertainty?

Of course, there is a difference in price, and it is significant. But apparently it's not a big enough difference to make up for all the problems, because I'm still getting the CDs. I suspect there's also an effect where people pool their purchases, buying CDs and letting their friends rip mp3s of them, so that in effect they are experiencing a lower price for the CDs that they don't absolutely need to have. This makes the price difference with the digital file seem less appealing if you have a group of friends with similar music tastes.

Still, I do want digital distribution, especially given how unused my CDs become once I've ripped them, and I think it would be more attractive if they made some changes. First of all, they should respect the terms of the license in effect when you buy the song, and waive their right to change the terms of the purchase after the fact. This will prevent the bait-and-switch problem that arises from the fact that the files are not your property.

Then they should let users listen to the entire CD at a very low quality level, instead of giving them 30-second song clips. The Flaming Lips have a very crappy version of Yoshimi Battles the Pink Robots on their website. It lets you listen to the entire album streamed in Flash, and it sounds like garbage. Still, I listened to the entire album several times all the way through, and then had to get the CD to stop the stutters and hissing and the fact that I had to be on the web and playing my music through Flash. I'm pretty sure I would have bought the album anyways, but I don't think someone unfamiliar with their music would have heard what is good about the album from a group of 30 second clips.

Amazon does something very similar to this, letting users search inside the entire book and get specific pages out. However, they put tight limits on how much of the book you can see, so that you don't just read the book online. This might lose them a few sales due to people who were just looking for a specific piece of data, but those weren't exactly likely buyers anyways. They probably get additional buyers who satisfy themselves that the book contains a lot of what they're interested in (which they can't ascertain from just the first few pages), and feel comfortable buying it.

The mechanism could work similarly on iTunes Music Store. Interested, registered users logged into their accounts could stream the entire album (counted as 50% or more) in very low quality, say, 5 to 10 times in a 2-year period. After that, sorry, you've gotten a good idea of what's in there. Streaming the album in this way isn't a good substitute for having bought the album because it sounds crappy, it limits the number of times you can hear it, and it doesn't let you send the album to your iPod or through AirTunes to your stereo.

Loosening up on the usage restrictions like this also makes it possible for users to share CDs they think their friends might like. Instead of lending the CD, a user can simply say "Go to iTunes and stream this CD, I think you'll like it." Since people probably have a better idea of what their friends like, I would be surprised if this didn't result in a net positive for sales.

Finally, iTunes Music Store should give users some recourse if their hard disk crashes, other than buying the whole shebang over again. It doesn't necessarily have to be free, but they do have your purchase history, and they should offer you a steep discount to download the files you've bought again in case you need to. It could be a sliding scale, like 50% off for a collection of less than 100 songs, and 70% off for a collection of 100 to 500, and 90% off of bigger collections. That should deter people from casually downloading songs over and over again ("Hmmm, why do you want to do that?"), without leaving them totally screwed if they lose their music collection or have to switch to a new computer.

The best proof that the vice presidential debate was an embarassing failure for the Democrats is the universal consensus amongst Kerry/Edwards supporters that the debate was a draw. This is an nonsensical as the same response to the first presidential debate from Republicans.

Not that Edwards was without his good moments. Almost immediately after the debate shifted to domestic policy, he started to not suck. Some of his answers were actually good, especially his obviously-rehearsed defense of the fact that he's a trial lawyer, and very convincing, substantive discussion of malpractice suits (even if I disagree with his solution to the problem). His closing statement had the potential to be powerful, but given the preceding 90 minutes, it felt too fake and rehearsed. I can understand why he reminds people of Clinton, when he's giving a speech. But Clinton can talk like that when he's speaking unrehearsed, and Edwards can't.

Cheney is a fantastic debater, and his performance tonight was easily the best of any presidential or vice presidential debater in the past 4 years. If you didn't understand anything they were talking about, you'd conclude that Cheney had a better grasp of policy. His answers were coherent and substantive, convincing. Edwards debated like Bush, falling back on pre-packaged attacks to the point of repetition. All through this, as Cheney wiped the floor with Edwards, he did it with a calm, pleasant manner that said Edwards was getting what he deserved.

Cheney took it a step further: he has an instinct for when he can cross the line and go in for the kill. Several times Cheney declined to respond to a question instead of filling up his alloted time, exactly those times when Edwards was sounding weak and repetitive. When he cited Edwards's attendance in the senate, it made Edwards look pathetic, and Edwards had no good response. He just got Bush-like and cried about "distortions" before changing the subject. I think that even people who don't understand the policies being discussed very well can sniff out that sort of weakness.

I think Edwards erred when the moderator asked him about flip-flopping and spoon-fed him some Bush flip-flops. He claimed they'd been consistent the entire time, a claim that strains credibility. I don't know why it's such political death to say "We extended you our good faith when you claimed that Saddam Hussein was on the verge of having these weapons and immediate action was necessary, and now we've seen that you were making those claims irresponsibly," or something to that effect. Instead, by doubling down on the "total consistency" angle, they've cut themselves off from that direction for good. Kerry was closer in the first debate when he pointed out the kindergarten insight that you can be certain and wrong. Sometimes it can help your case to admit your faults, it takes an attack a way from your opponent (8 Mile had a great example of this, but you probably don't want to deploy that attack unmodified in this context). Edwards did something like this late in the debate, when he admitted that he didn't have the experience record that Cheney did, but that didn't mean he was wrong (or something to that effect). Unfortunately, I think people had mostly figured out what they needed to from the first half of the debate, and turned their TVs off by the time things were going well for Edwards. Fortunately, I think most people aren't going to be too swayed by the VP debate anyways.

Watching this debate, I couldn't help but wonder whether maybe Bush hasn't pioneered a long-term innovation in the role of the vice president. Obviously, the vice president hasn't had an important role in previous administrations. Lyndon Johnson said the most miserable time of his life was being Kennedy's VP. He was used to having incredible power as the Senate Majority Leader, and took the VP slot thinking that "power is where power goes." He drafted an order for Kennedy to sign that would have made him something like "co-president," but obviously Kennedy declined.

However, Bush has gotten a lot of mileage out of Cheney. Cheney is obviously a heavy-weight policy wonk, and does a lot of heavy lifting. The amount of hatred that is evident for Cheney probably does draw some of the heat away from Bush himself. And Cheney's position probably helps him get things done. Watching this debate, you couldn't help but wonder if things might have gone differently if Edwards had been the presidential candidate, and someone like Madeleine Albright, William Cohen, Colin Powell, or some other well-spoken, highly intelligent, long-time policy expert who is too old and obviously disinclined from politics to be any sort of threat to the president.

Update: I woke up this morning to find that the consensus appears to be that Edwards won the debate. Apparently the second half of the debate resonated with watchers more than I thought it would. When Cheney started destroying Edwards early in the debate, and Edwards sat there fidgeting with his pen and burying his face in the coffee mug, before failing to counter any charges and instead repeating his vague allegations, I thought it was all over. Cheney went on to make unbelievable statements that he should have known better than to try to get away with, but my fears were that Edwards was so damaged at this point that it was all moot.

This morning, there's several developments on what Cheney said. Obviously, his claim that he never said there was a connection between Saddam Hussein and Al-Qaida is so ridiculous I doubt anyone has bothered to point it out. But Cheney incorrectly gave the URL for Annenberg's Political FactCheck (factcheck.org) as "factcheck.com." George Soros snapped up that domain and put a message up on it. Also, the Kerry campaign has produced a picture of Edwards and Cheney on the stage next to each other at some event (which of course doesn't prove that they actually met, but if it gets enough circulation, it's as good as proof).

Andrew Sullivan, William Saletan, and Chris Suellentrop and others whose opinion is probably better than mine all disagree with me. More importantly, the first instant polls are coming back in favor of Edwards, especially CBS's poll of undecided voters, which shows that they favored Edwards. Given that more people are probably reading about the debate than actually watched it, if it continues at this rate it might add up to a functional win for Kerry/Edwards, even though I'm still not sure what they saw last night.

Another update: Apparently Fred Kaplan and RealClearPolitics got the same read as I did. Kaplan says:

Cheney came into this debate with a weak hand. The war in Iraq is going poorly. Casualties are mounting. Meaningful victory is increasingly elusive. The rhetorical foundations of the war—weapons of mass destruction, Saddam's connection to al-Qaida—are thinner than ever and acknowledged as such by more insiders every day.

Edwards referred to these trends and to the administration's misjudgments and misrepresentations. But he only referred to them. He didn't pound them home time and again.

The Bush-Cheney campaign has tried to deal with the administration's problems by switching the perspective—warning that Kerry will only make things worse. And Cheney did push this button over and over: "You're not credible. … You have a record that isn't so distinguished. … Your rhetoric would be a lot more credible if there was a record to back it up. … I don't believe [Kerry] has the qualities we need as a commander in chief"—all uttered with a tone of parental disappointment.

Kerry defused this tactic very effectively last Thursday. Edwards didn't tonight. It may be that reading a transcript of tonight's debate on foreign and defense policy—or, better still, a point-by-point outline of the transcript—might make Edwards come out the winner. Cheney, after all, spouted a fair bit of nonsense. And Edwards pointed out that some of it was nonsense. But my guess is a stronger impression was made by Cheney's withering assault on his challengers' qualifications for office. Edwards shot down some of the specific attacks, but he didn't mow down the barrage.

Another update: Andrew Sullivan elaborates here. He's convincing me.

Here are my thoughts on the debate.

I watched the entire debate, and half an hour of the post-debate coverage. When Kerry started off in the very beginning, I winced at his Floridapander, but Bush managed to screw up by not having a much better, more involved counter-Floridapander prepared, and lamely echoed Kerry's remarks.

However, that was the last time Kerry made me wince. After ten minutes, I told Claire I thought Kerry was winning. For the rest of the debate, Kerry was - dare I say - excellent. Every single question, he started with a very strong short answer, and then elaborated. Every statement ended right as time ran out with a closer that left an impression that the answer was firm, consistent, and complete. Kerry spoke calmly and fluidly, with a very strong prosecutorial manner. It was strong, polite, and commanding; he left nothing frivolous for anyone to complain about like in the 2000 debates.

Bush was the opposite, almost immediately he started coming apart. He would start his responses with a pause of several seconds while he fumbled for words. His replies felt like rehearsed pleasantries devoid of specifics, and he seemed surprised by Kerry's performance, holding to a debate strategy that was outdated from the very first question. Kerry wasn't very specific about anything, but he certainly gave you the impression he was being concrete and substantive to the extent that a 2 minute response allows. Bush shifted around, seeming to hold onto the podium like he was on some scary rollercoaster ride at times. Bush kept trying to stick irrelevant barbs in there, to the point that it felt like a non-sequiter; "...oh yeah, and flip-flopper! But anyways..." (Not a real quote). At one point, he started off a reply by highlighting a vague term that Kerry used, and then found himself with nowhere to go with that line of criticism. It looked weak.

And when Bush actually was talking substantively, I got the impression that someone who wasn't keeping up to date with current events wouldn't know what he was talking about. Kerry actually managed the unbelievable feat of talking about policy, and then explaining what it meant quickly, simply, and without any condescension at all. It was shocking.

I thought one key moment that was both good and a missed opportunity for Kerry was when they were talking about building a coalition. Kerry said the obvious, which is that our coalition was a joke. In his rebuttal, Bush talked up our allies, and made a big deal about Kerry's having forgotten Poland in his list (that military powerhouse). Then in his counter-rebuttal, Kerry detailed the countries who are in the coalition and then detailed their troop levels, making obvious what a joke it is. The missed opportunity was his not giving the comparison to the coalition that Bush's father assembled for the first Gulf War.

Kerry made a few minor errors in judgement, I thought. On a few minor points, he leaned on a touchy issue, like Kyoto, which hadn't been mentioned by anyone. He also told an anecdote about JFK trying to convince France about the Cuban missile crisis, which I fear was ill-advised, even if I agree with his point. It came around and ended well, with him pointing out that no one in the world would say to us today "We don't need to see the pictures, your word is good enough." Still, I wonder if perhaps someone could come away thinking he was trying to talk up France the same way he was trying to mention every single swing state as many times as he could.

This was compounded by his "passing the global test" line. I didn't have a problem with the line in itself; the question was about how the United States should deal with pre-emption, and Kerry was highlighting how important that issue was to get right, given how dangerous pre-emptive war can be. Bush wasn't able to do anything with it; as I said above he just drew attention to the line and hit a brick wall. I'm not sure some uninformed voter would get what Bush was trying to point out, but who knows what the magic of campaign ad editing can do.

The losses due to all this were hopefully swamped by what Kerry did for himself on the flip-flopper charge. His demeanor was so firm that he countered it with just his body language and speech pattern throughout the debate. Bush's canned lines bounced off him as if he didn't hear them, and it worked. Then Kerry explained the simple fact that you can be certain and wrong and did it masterfully. A clear, memorable moment in the flip-flopper issue.

However, I don't think any of that is the real issue. Just as the talk of the 2000 debate was Al Gore's sighing, the most important thing to happen in this debate was everything except what was being said. Kerry looked strong and confident, Bush looked defensive and nervous. When Bush was attacking Kerry at one point, Kerry nodded calmly and wordlessly said "I agree," neutralizing the attack while Bush was still in it. In contrast, Bush would often stand there leaning, doing his pursed lips thing, and looking pissed off. ABC's post-debate coverage said that their instant poll had 45% saying Kerry won, and 36% saying Bush won. CBS got similar results, and CNN got an even wider margin for Kerry. I try to be sensitive to those demeanor issues, since they seem to be so incredibly important to people. Still, they're not very important to me, so I don't really trust my own opinion here.

I'm not sure at this point that it's going to add up to a Kerry victory. I have managed to follow this election so far with almost no hope, and I hesitate to get my hopes up now. Everyone knows Kerry won, and the conservative pundits are pushing to make the limited case that "it was a draw." Nonsense, even that limited claim looks ridiculous. For the first time, I felt like I wanted Kerry to be president, and not just that I wanted someone other than Bush. Given the median voter theorem, that's not good enough. Still, I follow politics ridiculously closely and had managed to not see Kerry speak since January. It isn't hard for me to imagine that many swing voters tuned in and saw Kerry for the first time tonight. And it isn't hard for me to imagine them liking him.

Mickey Kaus has some more thoughts here.