THE VIEW FROM CHAOS MANOR
View 340 December 13 - 19, 2004
Highlights this week:
For boiler plate, search engine, and notes on what in the world this place is, see below.
If you are not paying for this place, click here...
For Previous Weeks of the View, SEE VIEW HOME PAGE
If you intend to send MAIL to me, see the INSTRUCTIONS.
December 13, 2004
There is a very great deal of mail, which I will get up as soon as I can. I do have to go out to Fry's today.
Thanks to those who have recently renewed subscriptions, and the new subscribers.
The new BYTE column with my Christmas gift suggestions and a call for nominations on the annual Orchids and Onions Parade is up. Send nominations as separate mail and include the word orchids or the word onions in the subject so my rules will sort it properly, and don't include anything you want a quick answer to. If you want your name published (assuming I publish the letter; not all will be) include your name as a signature, and if you want your email address included, put that in as a signature as well. In other words, I do not in orchids and onions mail make any effort to identify the sender or include name or address unless that is in the body of the letter; I won't paste out of the header.
I'll words on several subjects later. Now I have to get to the mail, then a trip to Fry's.
|This week:||Tuesday, December
I remain unaccountably tired, but I did write a decent essay in mail yesterday. It will have to do.
Friday the 13th fell on a Monday this month.
I hope to have a couple of essays done shortly. Meanwhile, there's interesting mail. Or I thought it interesting.
I have recently been reminded that the New York Times comes by its fussbudget superior demean rather naturally:
"That Professor Goddard with his 'chair' in
Clark College and the
Everyone in the rocket business is familiar with that Times quote, and
whenever the science press corps would hold a fete for Walter Sullivan
you may be sure that someone would quote it.
December 16, 2004
Over in another conference we were holding a discussion on evolution. One participant asked if anyone had ever created life in a test tube. The obvious answer is no, nor does anyone know how to do it, but there are plausible explanations for how it might have come about, and mathematical models for demonstrating how, given what we believe about early conditions, there might then have been evolution to more complex life forms; from which, eventually, all that we see about is would be generated. As we get closer to the present era, the mechanisms for producing the results we see are better understood, and in some cases can be said to be well understood.
That would have been the obvious answer from someone trying to engage in an intellectual discussion. Instead, the response was to attack the motives of the questioner, accusing him of being a closet "Intelligent Design" advocate. In the course of that response it was said, flatly, that one cannot understand evolution -- which is the fundamental thing to understand in biology -- without understanding higher mathematics.
To which I said:
You know, I can't do the stress dynamics for large structures, and I have long ago lost the ability to do some of the math for aerodynamic analyses, but I can see whether a bridge or a building falls down without understanding the math of those who designed it.
A long time ago I was on a preliminary design team for Boeing for the WS 110A, a very important contracts for Boeing. We bid flush rivets and some of our best aerodynamic technology. North American, a fighter company, won the contract for a SAC bomber because they had done their homework with wind tunnels -- it turns out that in the critical flight regime there is a boundary layer 30 or 40 mm thick. North American had external antennae rather than flush rivets...
Yet I dare say their aerodynamicists weren't better than ours, and certainly their mathematicians and programmers (FORTRAN for the UNIVAC in those days) were not better than ours.
I include the incident of the WS-110A as an illustration of what I would have thought a fairly well known phenomenon: most complex questions can be explained to an intelligent readership without the use of jargon or resorting to the internal technical workings of the theory of the explanation.
The rocket equation is actually pretty simple math so I am unsure why "rocket science" has become a synonym for the intellectually difficult, but one can explain how rockets work, why they work better in vacuum than at sea level, ISP as a measure of efficiency, and why Single Stage to Orbit is at the edge of the possible on Earth but easily accomplished on Mars without using a single equation or resorting to jargon. Feynman does a pretty good job of showing the paradoxes in quantum theory without using mathematics: a thorough study of the most simple experiments and their result doesn't take any math at all to get your mind boggled. And so forth. Evoking this reply:
look here, jerry, i doubt richard feynman would agree
> [Quoting a previous remark by another participant] We're supposed to think, almost required to think, that critters evolve by high-energy sunburn...they have weird offspring (which a lot of us manage without cosmic rays). These tads are occasionally an improvement on their parents, or in human experience at least think they are, and so they have lots of kids....To believe this, you have to believe that you can improve a car engine by firing a rifle at it.
[The writer of that] is obviously unfamiliar with experimental evolution experiments like those of Lenski et al., where they have actually taken populations of E. coli, evolved them for 10000 generations, and sequenced them to find out that - yes - mutations DID occur and the mutations DID boost fitness in the lines of E. coli that survived. Because E. coli cells can be frozen down you can do something that you can't normally do, namely a direct comparison of the genomes of the 10000 and 0th generations.
He's also unfamiliar with the fact that random point mutations are *not* the only way in which genome content evolves - there are insertions, deletions, duplications, genomic island integration events, recombinations, etc.
BUT - most importantly - he doesn't understand the *fundamental mathematics* behind the reason that evolution works! And he doesn't want to know!
The reason is that randomized search *with selection* is actually a *provably good* optimization strategy. Evolutionary algorithms are actually *used in nonbiological contexts* in computer science. Amazingly, that implausible ol' natural selection thing actually works even when genomes are circuit plans and phenotypes are transfer functions! Brief summary here:
Mathematically speaking, natural selection is a stochastic hill climbing algorithm in fitness space. If you don't know what that means - or find it AMAZING and IMPROBABLE that a stochastic algorithm could be provably optimal simply because it's "random" - what can I say?
Interesting. Perhaps. What is asserted here is that there are mathematical models that can show certain kinds of progress toward PREVIOUSLY DEFINED GOALS: that is, there is a selection algorithm built into it. These are known as cellular automata and many of you here will be familiar with them. I didn't say that, though. Instead I answered, far too flippantly:
Well, I suppose I will remain stupid and
ignorant, not smart like you.
Which got the answer it deserved. So I composed this, which is in fact the reason for putting you through all this stuff. I said:
I will be among the first to say that it would do me no harm to have explained to me some of the concepts of evolutionary selection that are thrown at us like weapons; and since all of us here have had at least some mathematics, even if we have not used matrix algebra and factor analysis for a while, we could probably understand the instruction. In my case I have been to many of the cellular automata conferences, and I have some notion of how such things work; in fact I have written a couple of automata generators. But I could use some instruction in these techniques of describing evolutionary selection.
That of course is precisely what we have not been given.
As a novelist all this is useful and amusing, but I can find plenty of examples of this kind of irrational behavior in many places.
If the world has come to the point where, in order to function as citizens and political advisors and novelists and essayists and such like, we must learn new forms of higher mathematics whose nature can only be hinted at and given in references in textbooks, and whose effects can only be explained in esoteric terms understandable only by those with the proper gnosis and initiation into the arcana, so be it; but perhaps a few of us will be permitted to say bullshit.
There are few ideas so complex that their effects cannot be explained without resorting to higher math. I made use of incomplete gamma functions in order to make models of equipment failure with and without maintenance, and lives depended on some of those models; but I never had to explain what an incomplete gamma function was to the generals who had to decide on maintenance strategies. I had to explain what the function predicted, and demonstrate that its 'predictions' pretty well fit the data and evidence we had of the past, but I would never have attempted to put that kind of function up on a briefing chart.
The notion that there are things beyond the ken of ordinary intelligent people has been around a very long time, and so far, at least, the notion has turned out to be bullshit.
I don't have to know a lot about climate models to see if, given the conditions of 1950, it comes close to predicting the actual conditions in 2000. As a policy maker I wouldn't have to know much about the internal workings of the model if I found that while it supposedly predicted the future perfectly, it could not account for the past. Knowing only that I could make a rational decisi0n involving billions in public money based on the notion that we have a valid prediction of the future. This time for sure!
I think most of us here understand how statistical predictions work. We [Paul Horst and his students] were perfectly able to demonstrate the workings of the University of Washington Grade Prediction Program to people who did not understand matrix inversion, error matrices, and factor analysis. For that matter it's not all that hard to explain to people what Spearman's g is without going into the actual mathematics. Sure, it's EASIER to explain g to statistically sophisticated people, but one doesn't have to make statisticians out of people to explain it.
I have found, in a rather long life, that those who continue to refer to matters supposedly beyond my comprehension and who couldn't rationally and without calling people names tell me what those arcane factors did or predicted, were in fact confused themselves. People who really understand their subjects don't throw up their hands and act appalled at the ignorance of those who ask questions they have trouble answering, and don't use a fog and smokescreen of their supposedly better intellectual equipment to evade questions.
There followed a long letter from an economist, who disagreed with the first writer on many points, and asserted in the beginning, regarding the New Right's economics:
"But what use is an NR ideology that places the utmost importance on the enforcement of private property rights in material goods going to be in an economy where value inheres in information, more often now created by an open source community and distributed gratis in a virtual commons? Where the most valuable material posessions are land rights, ownership of which is based on hereditary luck? Where the bulk of the service economy is moving into education and medication, two superior goods where the public state has an advantage over private firms in both productive efficiency and distributive equity? "
These are serious questions, which deserve some serious thought.
First: value inheres in information, much of that created by an open source community and distributed gratis in a virtual commons.
There is a good bit to think on here. Political economists have speculated about such societies for some time. The usual proposed remedy is distributism: rights in property except periodically some of those are skimmed off and distributed to all; the difference between that and a welfare state is that the recipients are given property, not reduced to serfdom by getting a salary for standing in lines and tugging the forelock to a clerk. Belloc, the author of The Servile State, had a lot to say for distributism, but not for welfare scheme with administrators.
The clerk in a welfare state, on the other hand, has what amounts to a property right to a job, with both civil service rules and a public employee union to make it difficult to take that property away regardless of the quality or even existence of job performance. Are these cases covered in the models?
The tendency for regulatory clerical staff to multiply their functions -- to demand that people get permits for nearly everything -- is pretty natural too. Parkinson may have been whimsical but his data about the growth of such organizations without regard to their actual work -- Whitehall expanding its HQ staff at a time when the Fleet and Army were being reduced, the Colonial Office HQ staff growing despite the ending of the empire -- his data were real. I haven't seen very many economists study the tendency of the state to turn into a Permit Raj and the effect that has on an economy. I gather that in Britain it is nearly impossible for a landowner to build a chicken coop on his own property without getting a multitude of permits and employing a licensed contractor.
India seems to be a very interesting example of the New Economy. The last time the Wall Street Journal looked into it, there were still people awaiting trial after years of incarceration: their crime? Stealing from the state, to wit, riding the train without a ticket. The Journal found one person who had been there a lot longer than any possible sentence for the crime: but no trial was scheduled. We like to think that Modern Progress means the gradual elimination of such things; but I wonder if there are not powerful mechanisms promoting there growth and evolution? India has certainly developed some splendid universities and does an admirable -- I use the term advisedly -- job of educating smart people to take their places in the information economy. Whether it has done a comparable job of dealing with the bottom part of the social structure is another story.
Polities evolve, and predictably. When I was Deputy Mayor in Los Angeles the Civil Service was smaller, and we had 80 Civil Service Exempt positions. Today the Civil Service is MUCH larger, there are 600 Civil Service Exempt positions, and yet the police force has very few more sworn officers, and there are streets that have not been paved for 30 years. Elementary civic services aren't provided, but there are plenty of "mental health" programs (to which so far as I can see, there is no end: unlike mistresses and corruption which does have some limits). The city spends a lot more money than it did when I was part of its governance, but I am hard pressed to see, by observation or by statistics, a lot more returns on that government investment. Are such matters modeled in the new economic models? I have not seen any, but I am willing to believe they exist.
The feedback mechanism between the Prison Guard Union and its donations to politics, and the tendency to keep sentences for drug offenses long, and to build more jails rather than look at possible alternative sentencing, seems well established; I haven't soon such feedback loops in most economic simulations.
I will agree, confess, admit, that most of my economic thinking is done without hard mathematics; but when I ask for the hard mathematical models on certain questions, what I get is arm waving. I have yet to see a model of the global economy that takes into account the cost of unemployment and underemployment, and welfare costs, and the costs of downward class mobility brought on by export of jobs. I am more than willing to believe that a 10% tariff is not an optimum solution, but what I don't see is politically feasible alternatives to the problems caused by taking people who have long thought of themselves as middle class and casting them into the general labor pool. IN small numbers those may not be important; if the numbers get large, the importance grows.
In a word: politics generally trumps economics (whether models or realities); and the ability to explain the economic theories to those who have to live with their consequences seems to me something rather important.
And here I return to my original point: while many things can't be done without applying mathematics, and it may be true that one doesn't really understand something until it can be reduced to equations, most (I believe all) fundamental questions can be explained to intelligent people without resorting to arcana. I need not know the mechanism by which cyanide kills to know that it is deadly and to know how to detect it. And I need not know the internal details of a complex model to see whether its output has any relationship to reality.
December 16, 2004
Niven came over and we took a long hike.
We need a copy of Joseph Epstein's 1988 article "Who Killed Poetry?" and while I have found plenty of on-line references to the article, I haven't found a copy. I have long ago discarded my back issues of Commentary, where it appeared in the August 1988 issue, but I bought and paid for it (by subscribing) at the time.
I suppose this raises interesting questions about copyright. I want to show the article to Niven and to read it again myself; but I can't find it.
If anyone can find me an on-line copy, I'd be grateful as it will save a trip to the library and the onerous task of using my little hand scanner to copy it, then edit it (or make a Xerox copy). Anyway I'll keep looking. Not long after that article appeared I met Dana Gioia, now Chairman of the National Endowment for the Arts, at a Robinson Jeffers Society meeting in Carmel. Gioia wrote an article in Atlantic about Epstein's piece. In any event, Niven and I need to read it, me again, he for the first time, for a work in progress.
I have found a way to get the article by paying for it. This is minorly irritating in that I owned it and gave the copy to my local library, but mostly because they don't have a convenient way to just buy that darned thing. They want all kinds of information. And I subscribe to their magazine. The Atlantic has Gioia's commentary on Epstein available simply and easily. I wish places like Commentary had a Paypal link I could use to just give them the money and let me have my article.
I'll probably end up getting my wallet 0ut, getting out the credit card, filling out their form, and the rest, using more than $4.95 of my time in doing it. Amazon makes this a lot easier. I wish others would.
And in fact I did, and they wanted more information than I like to give, but at least their insistence on a phone number was satisfied by 818-555-5555 so they don't have that. A book store charging me 5 bucks for a book doesn't need all that information about me, but what the heck.
I now have the essay in pdf format, and I have printed a copy. For a moment I forgot that to print a pdf file you have to do it from inside pdf, not from the Windows print command.
And Aleta sends this, on Christmas and cats...
December 18, 2004
Holidays. Long strenuous walk with Sable.
December 19, 2004
This is a day book. It's not all that well edited. I try to keep this up daily, but sometimes I can't. I'll keep trying. See also the monthly COMPUTING AT CHAOS MANOR column, 8,000 - 12,000 words, depending. (Older columns here.) For more on what this page is about, please go to the VIEW PAGE. If you have never read the explanatory material on that page, please do so. If you got here through a link that didn't take you to the front page of this site, click here for a better explanation of what we're trying to do here. This site is run on the "public radio" model; see below.
If you have no idea what you are doing here, see the What is this place?, which tries to make order of chaos.
If you subscribed:
If you didn't and haven't, why not?
For the BYTE story, click here.
Search: type in string and press return.
The freefind search remains:
Entire Site Copyright 1998, 1999, 2000, 2001, 2002, 2003, 2004 by Jerry E. Pournelle. All rights reserved.