Thursday, October 30, 2003

Longhorn

As expected, Microsoft intends to extend its desktop monopoly in the networked world via Longhorn, it's much delayed major new operating system upgrade.

Good luck.

The Internet has been the application platform of choice for about six years now, and this is not going to revert back to the desktop just because Billg would like it to. Funnily enough, the open source folks are only just starting to twig this change themselves -- there was some recent post I am too lazy to find wondering if a service that runs on open source software carries any open source obligations a locally executing application would (answer: no) how how licensing might be altered to change this.

Microsoft's discomfort with this brave new world becomes clear when you consider they think they are in competition with Google because it means you are consuming processing cycles remotely, not locally. When you think about it, this is kind of true, but their response (compete with google using MSN, not make NT the real server-side Unix killer it was meant to be and, oh yeah, make it free) shows you how poorly they fit in the networked world.

Longhorn attempts to redress that by integrating the Windows client with the (not yet, but soon to be) Windows server, all running on WinFX APIs, the networked equivalent on the infamous Win32 APIs. On top of this, data becomes integrated by residing not in files, but in a database that itself is part of the OS. This is a good move in general, but I'm guessing this will mean bad things for file compatibility across systems, something that is pretty easy in this age of plaintext, pdfs, and html documents. Finally, Microsoft is developing the post-browser browser, which will work best when talking to MSFT servers and will be distinct from IE (which will continue to languish as it does now). You can see bits of this coming together already in the new Office launch, which has built in collaboration features that Lotus Notes has had forever but no one could find (or, more tellingly, cared).

It all adds up to a big chicken and egg problem that I don't think even Microsoft can structure a solution to. I don't think Windows is going away any time soon, but all the action is no longer running on Win32 APIs, it's all on the internet operating system.

Monday, October 27, 2003

Government without the Romance

Chicago Economist (and Nobel prize winner, natch) James Buchanan has this rather excellent post on public choice theory. At its heart, public choice assumes that government workers put themselves, their family, and their friends before other people (as do all decent human beings) and see what the consequences are for public administration.

I did not know the historic context around public choice theory or Ken Arrow's (Chicago econ, Nobel prize too) work demonstrating how elections results are artifacts of the voting regimen, so it was interesting to learn that this work came about against a backdrop of socialism. It was also interesting to note that Buchanan settled on a constitution to protect democratic systems from majority tyrannies -- formalizing what James Madison & Co had figured out several years earlier.
A somewhat loose way of putting this is to say that in a constitutional democracy, persons owe loyalty to the constitution rather than to the government. I have long argued that on precisely this point, American public attitudes are quite different from those in Europe.
I'm not sure what to make of this, given that the US produced the West Wing while England opted for Yes, Minister.

It's also worth reading this nice piece by Arnold Kling, on how ideological bias creates closed systems that lose, in the long run, to open systems. I know in this internet era open is all the rage, but it's also worth remembering that if you find yourself someplace where everyone agrees with you, you are someplace very weird.

Oh yeah, and while I'm pointing out good things by Arnold to read, check out this nice piece on why FCC Chairman Michael Powell is the most misunderstood man in government. In that article, Arnold also mentions David Isenberg (who often strikes me as foolish, but make up your own mind) and the deeply unimpressive Stiglitz (Nobel prize, econ, not Chicago) who I heard speak about 6 months ago. Stiglitz is speaking in Boston tomorrow, I won't be there.

Upgraded to 10.3

OK, I shelled out $120 for Pan-thire and the installation seems to be flawless. Mail is much better. PDFs are faster. The whole system feels more responsive (OS X.2 was stately, not snappy). So far, so good.

Sunday, October 26, 2003

Over the air, in the clear

You might say that the current spectrum/media morass the US is in today began with a congressional mandate, designed to *weaken* the then brand new TV companies, that any television signal broadcast over the air, had to be broadcast "in the clear" -- that is, without any encryption. I guess the idea was that everyone should have access to TV at zero cost. The effect of this was to slow down the deployment of cable (when that came around) and then to slow down the deployment of satellite (when that came around) and to transform media from something consumers value and pay for to a way to aggregate eyeballs and sell to advertisers, with the predictable consequences for program quality.

Within the media biz, everyone pretty much believes that free, over-the-air TV broadcasts are dead, and that folks will get their content from cable, satellite, or something else like that. The idiotic digital TV boondoggle seems to mandate a digital flag for over the air broadcasts, essentially saying that "in the clear" is finally going away after all. Good riddance.

Tuesday, October 21, 2003

More on prediction markets

Winterspeak reader TM send in this link that describes prediction markets in some detail. It's worth reading. It also includes links to the fabled HP forecasting markets that people who write about this stuff love to mention.

I joined Ideasphere recently and will write up the experience after I have had time to go in and set up some bids.

Monday, October 20, 2003

And now for something completely different

This is from the back of a bottle of Newman's Own Light Italian Dressing.
The Great Salad Dressing Baloon Race Across The Boot of Italty An armada of baloons loaded with Light Italian. The starters gun - Bazoombah! They all rise majestically into the air. Newman's Own Balloon, with fewer calories, more taste, and secretly propelled by charity, flies faster that Kraft and further than Wishbone. First across. First on the ground. El Piloto quaffs much quaffs of Newman's Own Light Italian in victory. A medium light Italian starlet, daughter of Butch Cassidini, named Bitch Cassidini, leaps into the balloon basket, kisses Piloto, her lips smeared with Newman's Own Light, she murmers, "You taste of Sicily, of Vesuvius, of Naples, baby," and patting his fanny she whispers, "and no fat".
Remarkable.

Thursday, October 16, 2003

The end is nigh

No -- not a Cubs v Sox World Series (the Apocalypse really would be upon us then). Apple iTunes store for Windows. Oh yeah -- AOL will offer it too.

Nice

My buddy from school, TC, sent me "claim" on whether or not Krugman would win the Nobel Price for Economics by 2040. Bid/Ask is currently 52/55, so his odds are about 53%.

My take? Well, since the prize cannot be awarded posthumously, and Krugman's about 55ish now (I guess), I'd give him a 95% chance of being alive to recieve the prize before the claim expires. I also know he is a John Bates Clark medal winner, many of whom have gone on to win the Nobel prize as well. I don't know how many though, and I don't know how soon after, so let's assume that 50% of Clark medal winners go on to win a Nobel prize too.

So, I would put Krugman's chance at 95% (might die) * 50% (predictive power of Clark medal) = 48%, which implies the price (at 53%) is too high, but not by much. Someone more diligent could easily put together better numbers.

I also notice that there aren't any claims on any Clark medal winners who do not have an op-ed column in the NY Times, suggesting that whomever put this claim up is either a huge fan of Krugman's (for his political position) or hates him (for his political position). Given that NY Times readers and Democrats (who tend to like him) far outnumber economists (who seem to think very little of him) I would guess that the Krugman claim was submitted by a fanboy. (Note: By bias would also lead me in this direction, so I probably feel more sure about this than I should). This may explain the clearly outrageous 80%-90% peaks the price history shows, but it says something about the power of markets that this optimism has been tempered to the far more sober 53% it is now. Or maybe a bunch of economists just discovered decision markets.

Tuesday, October 14, 2003

Reading prediction markets

I wrote about prediction markets recently and it might be worthwhile to go over how to read their numbers. This claim, for example, is about Bush winning in '04, and it gives him about a 54% chance. The bid is 53, ask 56, which I interpret as being (on average) about a 54% chance, with the spread between bid and ask being the cut the exchange takes to make its money. So, if you think Bush is going to lose for sure, you need to think he has below a 53% and you could take that side of the bet -- which would also lower his predicted chances. If you think he's going to win, you need to think he has above a 56% chance, which would raise his predicted chances. I'm going to sign up as soon as I have the chance, and give a more detailed account of what it's like. (Here's another competitive exchange)

Soft Skull Press

I actually met Sander Hicks back when I was in NYC and I was on his mailing list for a while. He was very left wing, so I'm surprised he decided to screw his workers by not paying them. So much so.

Monday, October 13, 2003

Apple farmers in upstate New York

Clay Shirky has a new essay outlining how the RIAA's law suits are driving users to paying services or ever-more-resistant-to-lawsuits filesharing networks. Previously the domain of corporations, encrypted, private networked storage is now being used by consumers to swap mp3s.

An interesting side effect of this is that obscure music gets dropped first, while popular tunes remain as available as ever -- a loss for labels, listeners, and artists alike.

Richard Thaler, behavioral economist at Chicago, has a note in this book (The Winner's Curse) about apple farmers in upstate New York (Dick used to teach at Cornell). According to Dick, New York apple farmers would leave baskets of apples by the roadside with a cash box secured to table the fruit was on. Passers-by were expected to buy apples and, on the honor system, pay for them by putting money in the cash box. But honor has its limits and the cash box was screwed tight to the table, so people could not walk away with it.

This system used to work pretty well, although economists would say it would fail because people would simply walk away with the apples without paying. To me it represents a very efficient sort of practicality -- I'm sure some people do walk away without paying for the apples, but not so many that it's worth having someone standing guard. Or trying to apply DRM technology to fruit. Or suing 13 year olds. The best digital music service will take the same common sense approach as New York apple farmers -- there will be some theft, some precautions, but people will overall have a good experience and pay for it.

Thursday, October 09, 2003

More Dumb Slate

Between Krugman and Sureweiki (sp), Slate used to have a great econ and business section, but they both left, one went to seed and the other to the New Yorker. Since then Slates financial columns have been pitiful, with this latest one from being particularly poor. Without going into the central economic insight which explains why falling prices in one sector frees up money and cash in all others (and therefore creates jobs there), it's good to have one killer econ fact that undermines the usual dumb arguments you get over this stuff. The article asserts "There's no question... that Chinese imports are pummeling American manufacturers." Oh really?

The truth is that American manufacturing has increased output by 50% over the past ten years -- which is pretty amazing when you think about it. It is also doing it with far fewer people as the economy as a whole continues to move from manufacturing to service, just as it moved from agriculture to manufacturing. But the image of a flood of Chinese imports crowding out domestically made products is just wrong -- and hard to maintain when you point out that America is manufacturing more today than it ever has in the past.

Wednesday, October 08, 2003

Ahnuld (Kling) throes down

Arnold echoes the disappointment I, and many of my econ friends, feel towards once-good economist, now ridiculous hack, Paul Krugman. Worth reading.

Ahnuld

I overheard some people today remarking that they could not believe Arnold won the California election. I presume they did not check out a prediction market which put Arnie's chances at around 85% yesterday (IIRC). Current odds on Arnie becoming President -- 10%.

One thing I like about the exchange is that the abstracted act of putting a wager on a pundit-esque outcome, while considering whether the current bid/ask spread puts your hard earned after-tax dollars at risk at favorable odds, alienates you to precisely the most useful degree when considering a political event (or prediction) that may make your blood boil. Sometimes your thinking does become clearer when it is just numbers on a screen instead of flesh and blood people.

The whole thing makes me think of U Chicago judge extraordinaire Richard Posner's book on the poor quality of public intellectuals, his argument being that demand for quality in pundits was low so the market did not supply any. A public list of bets, along with a win/loss history strikes me as being just the sort of thing you would want to judge the quality of your pundit of choice.

Consider the advantages of such a system: 1) you have a permanent record of positions and statements without requiring personal obsessiveness or the pundit's honesty, 2) less chance of claiming to be misunderstood, 3) a focus on pragmatism (if your insight does not translate into a bet, does it matter?), and 4) an acceptance that life is probabilistic, not deterministic, and it is the cumulative tally of wins and losses that matters, not perfect alacrity every time. By calculating actual wins, you would see if the pundit actually held contrarian (undervalued) views that signal greater insight and understanding -- which is presumably why they think people should pay attention to them in the first place.

I visited Neal Stephenson today who was signing copies of his new book Quicksilver, which kinda seems like a sciencey version of Pynchon's excellent Mason & Dixon. Stephenson is incredibly dull in person, incredibly. If the guy droning on in front of me did not match the picture on the back of his books, I would not believe it was the same fellow. A surprising disappointment given how rich and thoughtful his books seem to me. But (for some reason) it did get me thinking about how, over the past few years, the Internet has gone from being this wonderful groupfest to this kinda scary place which is 75% crazy and wants to destroy everything you hold dear. I wonder if people felt similarly dislocated in the early days of advertising when they were suddenly told to "buy soap" by some guy on a screen, so they did. Just as everyone ignores adverts now, I wonder if, in a few years, online screeds on how the current evil baby eating administration is destroying everything will seem quaint, like those retro ads for instant cake mix, and be tolerantly ignored. Of course, a public record of gains (or losses) on prediction market would be a healthy innoculant too.

Monday, October 06, 2003

Jane + SCO v. Linux

Jane does not get enough people yelling at her at home, so she's published a piece on how SCO's lawsuits might limit Linux adoption in the enterprise. She got her wish of more yelling, which she responds to here.

What do I think? I think Linux will do very well in the enterprise, no matter what happens with the SCO suit, which I think might have legal merit but only because copyright laws (and software patents) are utterly asinine. It's true that there may be some infringing code in the kernel, but that can be rewritten. Moreover, businesses are filled with software that they don't really have the licenses for, and it's true some large, exposed companies care about that deeply, but there are also plenty of smaller companies who just don't pay much attention to arcane licensing rules. So the SCO suit may deter some, but not all, and I think the all are very big.

Jane also asks "who is going to fund development?" The answer, on the server anyway, is anyone who makes complementary assets, including hardware, middleware, and applications. The usual suspects include IBM, HP, maybe Sun, Oracle, etc. etc. Certainly for hardware makers who support many operating systems, being able to unify and optimize around just one, even one as not-quite-standard as Linux, represented tremendous efficiencies and savings.

On the client it's a different story, one closer to IBM's mainframe business once the minicomputer appeared. People predicted that IBM's mainframes would go away and you know what -- they didn't. They are still around, doing payroll or whatever, and IBM still maintains and services them as before. They don't sell a bazillion of them any more, and they are a small percentage of the market now, occupying a rather uninteresting niche, but they are still around.

The Windows client is similar -- it's not going to go anywhere, but it will become a less interesting, less important part of the market. The important (most used) applications have shifted from the local (and complex) Word, Excel, PowerPoint to the hosted Yahoo!, Google, and Amazon. And guess what -- they run on Linux. I don't see why replacing local instances of Windows with Linux is all that important since Linux is already running all the important apps -- remotely via the browser.

Friday, October 03, 2003

Powell interview

CNet has a great interview with FTC Chairman Michael Powell. Read it all. Two things that were new to me:

1) Powell believes in the "end-to-end" principle. "What are we going to do to reach the world we want to see--one in which applications are very separate and can be run over IP?"

2) The furor over the FTC's decision to slightly relax media regulation was idiotic. I guess I'm glad some people are writing about Smart Mobs, but maybe Dumb mobs need some attention too.
All I would say is that if we're going to have a dialogue with our citizens, let's tell them that the difference between 35 percent and 45 percent (the portion of Americans any individual TV station would be permitted to reach) is four or five stations nationwide out of 1,400. Do you believe that democracy lives somewhere between 35 and 45? If it's 38, it's going to kill democracy? And if it's 36, then that's just fine? I'm going to be candid about it: It's preposterous.
(Thanks to Lawrence for the pointer).

Thursday, October 02, 2003

Moneyball

I just finished reading "Moneyball" by Michael Lewis, which tracks the rise of the Oakland A's under Bill Beane who applies statistics and science to baseball. It is a great book, and I recommend it to all.

My old Chicago professor Dick Thaler (along with Cass Sunstein--who lived on the South Side of the Midway) have a nice article talking about Moneyball. Their main question is: how come it took people so long before they figured out what Bill James, the fellow who figured all this out and now works for my new local team, the Red Sox, has been writing about for years? The basic explanation for efficient markets is the same as the explanation for why you don't find dollar bills lying around -- someone will have picked them up already. But baseball has been leaving literally millions of dollars on the sidewalk for decades.

If the baseball labor market, with so much available, recorded data, remained horribly inefficient for so long, how much trust can we have in the regular labor market? This is an appropriate question given the recent hubbub around executive pay and wages in general. Thaler makes the case that interviews, for example, are a terrible way of predicting performance and some sort of regression (equation) looking at things like test scores etc. do much better. This means that taking time to interview candidates--unless you are going to be working with them personally--is a waste of time and you're better of relying on an equation to judge the value of someone as a potential employer. When people hear this they tend to be revolted, and I sympathize with that, but I also think there is a lot to it. Regressions are better at picking winners in state fair animal contests, and job candidates.

I had lunch recently with a mathematician who ran statistics for the Air Force, and his job was coming up with the equations they use to select which applicant gets to be a pilot. I've also met a clinical psychologist who studied atheletes and criminals, but now worked on designing tests to pick future traders. It's nice to see that rationality in hiring is not restricted to baseball.

Wednesday, October 01, 2003

Clarification

A couple of folks have written in asking about my recent post on Krugman's Unequal Exchange essay. I apologise for being obtuse. I'll try and do better this time.

The gold miner/fisher parable is actually a very standard labor economic model for talking about how greater risk demands greater return for people to take it on. Or at least, it's what we used in class at Chicago. So imagine my surprise when it's trotted out to argue that the gold mining town should have *greater* redistribution, quite opposite to it's usual moral.

The sad truth is that this is one of those times when people's intuition about what's fair is not a reliable or helpful way to think about how to organize a country. I am very sympathetic to how counterintuitive, or even repugnant, basic economics is, but Man is not at the center of the universe, and supply curves slope up, demand curves down.

The insight I would hope an economist (esp. a humane economist) would bring to a distribution discussion was that if a society, for reasons for technological change, switches from a fishing to a gold prospecting economy, the costs of redistribution are going to soar precisely when the urge to redistribute rises as the gap between rich and poor becomes wider.

That's what an economist would say (or at least take as a starting point). Krugman is an economist. He did not say that. Disingenuous? It seems so to me, which is why I don't read him any more, but you're free to differ.