Heuristic – Page 2 – God of the Machine

There Oughta Be a Law

 Heuristic, Politics  11 Responses »
Mar 142003
 

It had to happen. Elizabeth Smart’s father, Ed, calls for “Amber Alert” — a program to notify the public of child abductions that is used in 38 states — to go national, at a cost of a mere $25 million. “There is no question that Amber Alert is a necessity,” says Smart, with the usual combination of good intentions and bottomless economic ignorance. “Having it saves children.” Since Amber Alert, by the reckoning of one of its proponents, has been responsible for the apprehension of 47 criminals, whereas America’s Most Wanted has nabbed 746, it might make more sense to call for a national law to broadcast it twice daily, or perhaps a special cable channel all local providers will be required to carry — all America’s Most Wanted, all the time.

Bad luck, it seems, confers instant moral authority. A hitherto obscure person, granted his day before the TV cameras, permitted to say anything he likes, demands — a new law! What could be more American? We need a name for this phenomenon, previously observed in anti-gun crusader Sarah Brady; Richard and Maureen Kanka, parents of Megan and Megan’s Law; and Linda Campion, the motive force behind a pointless New York law allowing relatives of crime victims to testify at sentencing hearings. (There are other instances I’m too lazy to look up, but Kaus says three is a trend.) Any suggestions?

(Update: Paul Dubuc proposes “tragislation.” Not bad at all.)

A Pattern Language

 Culture, Heuristic  No Responses »
Feb 202003
 

My tiny corner of the blogosphere is abuzz with Christopher Alexander. Michael of the Blowhards has an excellent brief essay on him, with a link to Wendy Kohn’s more detailed treatment. AC Douglas also has a few remarks.

Alexander is a (now ex-) professor of architecture at Berkeley whose most famous books, The Timeless Way of Building and A Pattern Language, have inspired a sort of cult. He believes that the key to architecture lies in rules, or recipes, which he calls patterns. The same problems occur over and over, and the patterns, which he claims not to invent but to discover, are well-known ways of solving them. A Pattern Language consists of 253 of these, with photographs and descriptions. Alexander maintains that if you simply follow the patterns, beauty will come. He aims to demystify all of architecture, and to a great extent he succeeds.

His patterns range in breadth from city planning to room decor, and many of them are alarming in their specificity. Cities should contain no more than 9% parking space; political communities should be around 7,000 people (this is reminiscent of the 19th century socialist crackpot Charles Fourier, who recommended 500 as ideal); no urban downtowns should serve more than 300,000 people; most buildings should be no more than four stories high; terraces should be at least six feet deep; every room should have light on at least two sides. Sometimes Alexander buttresses his rather ex cathedra pronouncements with studies and arguments; sometimes not. “Nine Percent Parking” gives a fair taste of his style:

We [he has co-authors] suspect that when the density of cars passes a certain limit, and people experience the feeling that there are too many cars, what is really happening is that subconsciously they feel that the cars are overwhelming the environment, that the environment is no longer “theirs,” that they have no right to be there, and so on… Instead of inviting them out, the environment starts giving them the message that the outdoors is not meant for them, that they should stay indoors, that they should stay in their own buildings, that social communion is no longer permitted or encouraged.

We have not yet tested this suspicion. However, if it turns out to be true, it may be that this pattern, which seems to be based on such slender evidence, is in fact one of the most crucial patterns there is, and that it plays a key role in determining the difference between environments which are socially and psychologically healthy and those which are unhealthy. [Italics his.]

To begin with, nine percent parking is based not on “slender evidence,” but on no evidence. It is a “suspicion,” which becomes a pattern, which becomes a dictum. Here you catch a faint whiff of the crank.

Yet it is a very plausible suspicion, even if the particular number is bogus. Urban landscapes full of cars, like Los Angeles, are depressing. Most of Alexander’s patterns are very plausible, even the ones that never would have occurred to me, like “Zen View”: “If there is a beautiful view, don’t spoil it by building huge windows that gape incessantly at it. Instead, put the windows which look onto the view at places of transition — along paths, in hallways, in entry ways, on stairs, between rooms.” The man who writes this has meditated long and profoundly about why some buildings succeed and others fail.

Alexander generally begins with what people want. You might think that most architects would begin there, but in fact very few of them do. Instead they talk a great deal about form, function, structure, “machines for living,” and the like. Alexander’s solicitude is one of the sources of his unpopularity within his profession and his popularity in the world at large. The photographs in A Pattern Language are of warm, inviting, pleasant places, places that would be fun to live or play or work in. They are not of monuments, large buildings, or what one has been taught to regard as architectural masterpieces.

In Alexander’s cosmology, beauty in architecture consists of satisfying people’s desires, and those desires are immutable. He uses words like “healthy” and “alive” with abandon. As Michael Blowhard notes, it follows that architectural standards are objective, and artistic standards as well. There is a human nature, to which buildings will appeal more or less successfully. It follows further that Alexander is in on the secret. It is this assurance, more than anything, that infuriates his fellow architects, who fancy themselves artists and resent the suggestion that someone has not only encountered their problem, but solved it.

Now I’m all for normative thinking, provided it’s kept far away from the police power. Jane Jacobs, with whom Alexander is frequently grouped, takes pains to show how livable cities grow organically from people’s natural behavior, while top-down planning leads to disaster after disaster. This concerns Alexander not at all: only ends interest him. Some of his grander patterns must be enforced by law, and he does not shrink from doing so. In “The Magic of the City” he writes:

Put the magic of the city within reach of everyone in a metropolitan area. Do this by means of a collective regional policies which restrict the growth of downtown areas so strongly that no one downtown can grow to serve more than 300,000 people. With this population base, the downtowns will be between two and nine miles apart.

He thinks people ought to own their homes. Arranging this is a simple matter: “Do everything possible to make the traditional forms of rental impossible, indeed, illegal.” So you’re not surprised to read this testimonial from one of his former students in Kohn’s article: “Chriss answer to my doubts about The Timeless Way of Building was to say ‘Find out your psychological problem that prevents you from agreeing.'”

Alexander’s biggest fans are not architects but computer programmers. Unless you are a professional, you don’t have a clue how vast his influence is in the field. The most important book written about software in the last thirty years, Design Patterns, takes its form explicitly from A Pattern Language. The authors enumerate thirty “patterns” that make for elegant, robust, even beautiful software. (The mark of a successful new technology today is the appearance of a book on it called Patterns in ...) Software design patterns are very like Alexander’s: solutions for recurring problems in software development. Cities and software applications are both complex systems that must be broken down into components to be understood. Alexander’s ideas lend themselves more readily to software than architecture because a software architect can control every aspect of a project. He need not rule the world to enforce his chosen patterns.

So we’re left with an inhumane humanist, a brilliant crank, an immodest prophet of modesty. Even so, A Pattern Language is one of the most interesting books about architecture, and the world, that you’re ever likely to read.

(Update: AC Douglas comments. I also posted a slightly different version of this to BlogCritics, where there are a couple more comments.)

(Another: Chris Bertram discusses Alexander and kindly mentions me.)

Affirmative Action Number Fun

 Heuristic, Politics  3 Responses »
Jan 292003
 

Goodwin Liu has exposed, in the Washington Post and at greater length in the forthcoming Michigan Law Review, a flaw in the thinking of affirmative action opponents that he calls the “causation fallacy.”

Affirmative action is widely thought to be unfair because it benefits minority applicants at the expense of more deserving whites. Yet this perception tends to inflate the cost beyond its real proportions. While it is true that affirmative action gives minority applicants a significant boost in selective admissions, it is not true that most white applicants would fare better if elite schools eliminated the practice. Understanding why is crucial to separating fact from fiction in the national debate over affirmative action…

…Allan Bakke, a rejected white applicant who won admission in 1978 to the University of California at Davis’s medical school after convincing the high court that the school’s policy of reserving 16 of 100 seats each year for minority students was unconstitutional. For many Americans, the success of Bakke’s lawsuit has long highlighted what is unfair about affirmative action: Giving minority applicants a significant advantage causes deserving white applicants to lose out. But to draw such an inference in Bakke’s case — or in the case of the vast majority of rejected white applicants — is to indulge in what I call “the causation fallacy.”

This is a “fallacy,” according to Liu, because the vast majority of rejected white applicants would still be rejected, even without affirmative action. This fallacy works in mysterious ways. The lower the standards for black applicants, the more rejected whites clear the bar. The more rejected whites with better credentials than accepted blacks, the less certain it is that any particular white would have been admitted if there were no affirmative action. It follows, from Liu’s logic, that the lower the standards for blacks as opposed to whites, the less cause for whites to complain!

Liu makes a big deal of the fact that Gratz and Bakke very likely wouldn’t have been admitted regardless, and in any case couldn’t be sure. He then publishes the following table, of admissions rates at “five highly selective universities” (this is thanks to Ampersand, who takes it from Liu’s full Law Review article, which I haven’t read and isn’t yet online):

SAT score
1500+
1450-1499
1400-1449
1350-1399
1300-1349
1250-1299
1200-1249
1150-1199
1100-1149
1050-1099
1000-1049
< 1000		 Black rate
100%
75%
69.6%
80%
64.6%
73.9%
60%
55.5%
46.2%
40.6%
35.4%
17%		 White rate
63%
51.1%
39.9%
30.7%
25%
22.6%
19.3%
18.7%
13.3%
12.4%
9.6%
3.3%		 Rate w/o AA
62.7%
50.8%
39.8%
30.8%
25.4%
23.8%
20.6%
20.9%
16.2%
15.5%
11.7%
6.7%

One wonders, first, what the raw numbers are. They would be easy to include and would prove instructive. (The nice round numbers in the upper rows in the black column make me suspect that we are dealing with a vanishingly small sample size.) It is fishy that the percentages of whites admitted in the upper percentiles declines without affirmative action. Ampersand comments that “[a] white student with a combined score below 1000 has a 96.7% chance of rejection from a selective school with affirmative action, and a 93.3% chance of rejection if aa didn’t exist. In either case, the odds are overwhelming she’ll be rejected; and the primary reason for the rejection is her poor SATs, not her race.” An opponent of affirmative action might retort that whites with such scores would have twice as good a chance at admission. This is a fine example of how to lie with statistics.

But the overwhelming question about this data is, how does he know? If Bakke and Gratz can’t prove that they would have been admitted in the absence of affirmative action, how can Liu establish the SAT distribution in its absence?

Ampersand also notes how whiny the AA plaintiffs are:

Anti-affirmative action lawsuits are not put forward by whites who would have gotten in to a selective college if only affirmative action didn’t exist. They’re put forward by whites who have such a strong sense of entitlement that they can’t admit they failed to gain admission because, on the merits, they didn’t deserve admission.

Well maybe, but Gratz and Bakke are paragons of virtue compared to Miranda, Escobedo, Gideon, and other plaintiffs in famous Constitutional cases. Spy magazine once ran a little story profiling such plaintiffs called “Dirtball Heroes of the Constitution,” and there isn’t an AA plaintiff who would even come close to qualifying. In any case, aren’t you supposed to take the plaintiff as you find him?

This whole business of percentages disguises the fundamental fact that for every black applicant who is admitted because of affirmative action there is a white applicant who is rejected for the same reason. We may not know which white applicant, but that fact is immaterial. Liu suggests “rethinking the conventional view that a race-conscious admissions policy pits whites against minorities in a zero-sum game,” but a zero-sum game is precisely what it is, and what it has to be.

Assignment Desk: Blog Clusters

 Blogs, Heuristic  1 Response »
Jan 232003
 

Valdis Krebs performed a simple experiment. He looked at the “buddy list” on Amazon of several dozen top-selling political books and graphed the results. (Link from BoingBoing.) The result is two clusters, as one would expect, but with one book in the middle, with “buddies” on both sides: What Went Wrong by Bernard Lewis. (Also, arguably, The Clash of Civilizations by Samuel Huntington.)

The “cocooning” controversy could be resolved the same way. Steven Den Beste theorized last year about blog clusters but without data to back him up. So the assignment, for someone less lazy than I am, is to create a chart, after Krebs, for blogs instead of books, using for data the top 100 blogs and, say, the first ten blogs in their neighborhoods at BlogStreet. This would be imperfect but indicative. How many clusters would there be? Who would be in the middle? Do people often read blogs that they disagree with or are blog readers, like book readers, blinkered by confirmation bias?

 Newer Entries