Sunday, December 28, 2008

Tuesday, December 23, 2008

MoJo on auto credit

Via cshalizi's delicious, Mother Jones comments on auto industry lending practices. It turns out that sleazy salesmanship w.r.t. the cars themselves isn't the half of how auto dealers screw consumers.

On the bright side, the article says that 900 car dealerships will fail over the next year.

Monday, December 22, 2008

Pandora for disconnected operation (an open letter)

Dear Pandora executives:

I hear that your business model's in jeopardy because of increased licensing fees for streaming Internet audio. I suggest that you diversify your income stream by licensing your database for media players, enabling them to dynamically construct playlists using the player's local collection.

Most modern music devices have the capability to construct dynamic playlists based on ID3 tags (for the less technical in the crowd, ID3 tags are how your iPod knows the artist, album, etc. of a music file that you download). The problem is that although ID3 tags include features such as release year and "genre", these don't suffice to construct a cohesive sounding playlist. Pandora's dynamic playlist construction based on musical qualities handily beats construction via arbitrary categories like genre.

Conversely, although people increasingly own rich media Internet devices with unlimited data plans (like the iPhone and the G1), WWAN Internet connectivity is still not pervasive enough to provide a seamless streaming radio experience.

Meanwhile, the cost of storage (both hard drives and flash memory) is falling so dramatically that people can save both encyclopedic music collections and a fairly substantial database of metadata about those collections on local storage. (Any worries about "giving away the store" by letting people cache slices of your database locally should be alleviated by the realization that people will require periodic updates, as the music industry — despite their insistence that digital piracy is strangling them to death — continues to produce an avalanche of new music every week.)

There is a window of opportunity here. By providing a mechanism for good dynamic playlist construction, you could get a licensing fee on every player sold. And since you'd be providing metadata about songs, instead of the songs themselves, you would pay no streaming fees.

Alternatively, you could at least write software that opportunistically plays a song from the local device when available, instead of streaming, thereby reducing streaming fees and improving audio quality on such devices. (When someone listens to a station long enough, chances are that the user will already own a nontrivial fraction of the songs that come up.) Notice that if you write this app, then it is a tiny, incremental step to simply disable the streaming engine and play songs from the local device only. This would be less dramatic than caching the Pandora database itself locally, but it would reduce network bandwidth to an RPC to the Pandora server on song switches (and again, you'd avoid the streaming fee).

Once you have Pandora cached locally, you should take a page from Wikipedia and Delicious and allow people to collaboratively tag music with your attributes. (There's a potential spam problem here, but there are defenses which I think would be fairly effective.)

It seems likely that you've thought of these ideas. Maybe you've even found good reasons not to implement them. But just in case you haven't, I thought I'd suggest them.

Cheers, Cog.

Saturday, December 13, 2008

3 observations prompted by overnight air travel

(Attention conservation notice: Hastily written to pass the time in Charlotte NC airport between a red-eye and a connecting flight. Adjust your quality expectations accordingly.)

1. One of the most important pieces of information when walking through an airport is the current time. Yet in a surprising number of airports one can walk for many minutes without seeing a clock in one's line of sight. (Long digression on central planning & incentives nipped in the bud here.)

2. I always thought of the PSP (PlayStation Portable) as a niche device principally used to play games (at least in the US).  However, the young woman sitting to my left and the guy sitting 2 seats to my right each had one, and were using them exclusively to watch movies and listen to music respectively. Evidently Sony has been doing better selling these than I thought, and people have succeeded in using them for more purposes than I would have expected. I know that these devices can play video & music, but the experience of acquiring content, managing your collection, and loading it onto the device is hardly seamless; but maybe this deters people less than I thought. (Demographic note: the guy was also reading a novel of the obvious pulp fantasy nature called "Vorpal Blade".)

3. See the attached image for your amusement. Offered with the sole comment: What would you think of this ad if you didn't know that colds were caused by a type of virus called a "rhinovirus"?

Tuesday, December 09, 2008

Vocabulary PSA: loin and groin

In the past month or so, I have had two conversations wherein people did not understand the literal meanings of the following words, as applied to human anatomy:

loin
The meaty part of your lower back and sides, below your rib cage and above your pelvis.
groin
The crease where your torso meets your legs, running from the side of your pelvic bone (at your hip) to between your legs.

Due to the excessive use of these two words as euphemisms for genitalia, many people seem to be under the mistaken impression that both of these words refer literally to genitalia or the genital region. In the case of loin, it's pretty egregious, because your loins are nowhere near your genitalia. In the case of groin, it's simply annoying, as there's no other word that refers to that specific part of the anatomy.

This has been an Abstract Factory public service announcement.

Sunday, December 07, 2008

Two possible bailouts

A comment on my previous post suggested that I'm not considering the impact of throwing hundreds of thousands (maybe millions) of people out of work during a recession. Actually, I'm sympathetic to some auto industry employees, namely the ones who have no influence over retail operations. Due to accidents of geography and economics, a lot of Midwesterners got sucked into a career in an industry that has no future. But it's not necessary to save these pernicious companies (and the clowns who manage them) in order to help their employees. Consider the following two possible bailouts.

Bailout One: Give the automakers $25 $34 billion in low-interest loans and pat them on the back. This is the mainstream bailout plan that's emerging from Washington. Yesterday morning, they shook loose $15 billion from taxpayers' pockets by diverting money temporarily from a fund for environmentally friendly cars, but this won't be the end of it.

If we do this, what will we get for our money?

Well, GM has a net annual income of negative $10 billion (extrapolating optimistically based on 2008q3 performance; in fact, extrapolating from previous quarters yields a much worse figure). Giving GM, say, $20 billion means kicking this corpse of a company roughly 2 years down the road, at which point you will have to kick it again.

This is not an accident. Here is a fundamental truth about the global market for automobiles: there is currently more automobile-building capacity in the world than there is demand for automobiles. As a consequence, no combination of innovation and smart business practices can sustain all auto companies at their current size. Either (1) some automobile manufacturers somewhere in the world must go out of business or radically reduce their capacity, or (2) governments all over the world must sink money into automobile subsidies indefinitely. Notice that (2) amounts to a welfare program for auto employees, and an extremely inefficient one: some of the money that would go directly to paying for peoples' food, shelter, etc. instead goes towards building a pile of cars that nobody wants. (Meanwhile, those cars will be unloaded on the market at artificially low prices, which will lead to a greater than economically optimal level of automobile use, which leads to more traffic jams, smog, CO2 production, etc.)

So we're back to (1). If some auto companies somewhere in the world have to radically downsize or go out of business, the optimal outcome is that it be the most dysfunctional, wasteful, and stupid ones. Can anyone guess which companies those are?

Bailout Two: As of their last SEC filing, GM has 252,000 employees. For $25.2 billion, I propose cutting every employee of GM a $100,000 tax-free check and telling them to go make a new life for themselves. Meanwhile, force GM into Chapter 7 bankruptcy (i.e. liquidation, or the sale of GM's productive assets to companies that can make more productive use of those assets).

Ford employs 224,000 people, and we could offer similar terms ($100,000 to each employee) for $22.4 billion total, while sending the company into Chapter 7. Chrysler, having recently been bought from Daimler by a private investor, does not have current SEC filings, but Wikipedia claims that it employs 58,000 people, making it a comparative bargain at $5.8 billion overall.

Note that with a well-managed Chapter 7 selloff, the plants, equipment, technology, etc. will turn a profit for the buyers and continue to contribute to the economy of the region. (With an ill-managed selloff, the value disappears down a black hole; but if you assume bad management, then lending the companies money will be a disaster too.)

The total cost for bailout two, for all 3 automakers, is $53.4 billion, or $178 for every U.S. citizen. For this price, every employee of the Big Three receives a one-time tax-free bonus of roughly twice the annual median household income in the U.S. I contend that if you can't make a new life for yourself with that transition fund, then you're beyond help.

I'm not an economist, but I don't see how bailout one is obviously superior to bailout two.

Of course, bailout two is just a crazy idea of mine, and nobody who matters is considering it; nor would they if it were suggested to them. For reasons that are not entirely apparent to me, opinions on this matter have coalesced rapidly around a narrow range of alternatives, all of which involve lending massive sums of money to companies with a track record of destroying economic value instead of creating it.

I'm not saying my bailout is actually a good idea. I'm saying that it's not demonstrably true that Washington's bailout is any better, and yet it's gradually becoming more and more obvious that Washington's bailout is inevitable.


Incidentally, it's been interesting watching all the liberal blogs come around gradually, albeit begrudgingly, to a pro-bailout stance over the past few weeks, despite the absence of evidence that Congressional Democrats have the backbone to demand adequate oversight or other conditions. It's like the prelude to the Iraq War: all the Thoughtful Liberals agree that Something Must Be Done and so everyone piles onto the bus, even though the driver doesn't show any indications of being able to get it right. Maybe people think that if you're on the bus, at least you'll be heard. But that's a totally false assumption. Thomas Friedman's War was not George Bush's War; guess which one we got? Likewise, Paul Krugman's Bailout will not be Barney Frank's Bailout; guess which one we'll get.

I can't help but think that political calculus has played a role in these liberal blogs' turnaround. If you destroy the Big Three, then you cripple the UAW and anger people in Midwestern swing states at the dawn of a period of Democratic governance. It's understandable for politicians to think this way. People whose jobs do not depend on being elected, however, should advocate for what is good, rather than for what is politically expedient; because by advocating for what is good, you make it more likely that someday the good will be politically expedient.


UPDATE 5:36 PST: OK, for the record, here are two reasons why Cog's Grand $100k Giveaway wouldn't be sufficient even if you could get everyone to agree to it. (a) It does nothing for employees of the Big Three's upstream suppliers. (b) It leaves the automakers' huge pension obligations unfunded, which leaves the pensioners in the lurch. The proposal was less in the vein of "Let's do this now!" and more in the vein of "Look, this is a lot of freaking money and we haven't even begun to consider whether this is really the best way to spend it."

Saturday, December 06, 2008

Why I hate auto manufacturers

Consider this: Almost every car company forces you to deal with that semi-sociopathic middleman known as the car salesman.

The function of a car salesman is to use information asymmetry to introduce inefficiencies into the market. By using deception, social engineering, and emotional manipulation, the salesman attempts to get you to irrationally pay more than the true market value (which used to be a total mystery, but which you can now find on the Internet; of course car companies hate this). Instead of selling cars based on their actual worth, the car salesman enables the company to price discriminate based on how uninformed, impatient, and manipulable you are, which of course correlates strongly with your social position, access to information, etc.

Here is how a rational car market would function. First, instead of car dealerships, there would be third party "test drive centers" which kept models of every recently released automobile for you to drive. When you were serious about wanting to buy a car, you would go to a test drive center, pay a hundred bucks, and test drive as many cars as you want. At the end of the day, you would sit down at an Internet terminal and order the car that you liked best. There would be staff at the test drive center, paid on a livable salary, not on commission, to help you sort through the information if you wanted. Every manufacturer would simply have a sticker price, which would be true prices instead of today's farcical "sticker prices" because auto manufacturers would have to compete with every other manufacturer on a transparent market.

Needless to say, auto execs would not reorganize the market this way if their mothers' lives were hanging in the balance. They prefer to force all their customers to sit through a several-hours-long soul-corroding encounter with a man who you know, you know is lying to your face. Encounters like these are social pollution: they sully the population's common stock of faith in humanity. And car companies feel that this social pollution is absolutely vital to their bottom lines, and so they cultivate it.

It maddens me to think that my tax dollars will be going to these people. Liquidate the executives, liquidate the plant workers, especially liquidate the dealerships, but in any case liquidate these companies. Kill the companies dead, carve up their assets, scatter the pieces to the winds. And let Midwesterners find a way to build things of value without deceiving people and grabbing handouts from the public till.