09 September 2015

A Thing Of Beauty

If you didn't watch the Serena-Venus match last night, you missed a thing of beauty. You also missed, in a subtle way perhaps, an object lesson in experience. I kind of hoped the match to be three 7-6 sets, maxing out on The Girls. A few points into the first game told me how wrong I was.

Back when I got out of school, the USofA was in the worst recession in decades, so I ended up in a job that was not, shall we say, all consuming. Which fact lent me ample time to fill. I was new to Boston, and didn't cotton to the bar scene. Taking stock, I concluded that I should fill some of this ample time with activities previously foreign. I chose karate and tai chi, in that order. One might consider these to constitute something of a schizophrenic decision. I recognized this, and attributed it to a quest for balance. Having been a bookish type up until then, well...

I studied karate with George Gonis in East Boston; he was still in the phone book last I looked. Among other things, his teacher lived in the Himalayas, which he visited every couple of years for instruction and testing. I studied long enough to have seen the result of one such trip (job in DC called after about a year in). Unusual. What was also unusual was the nature of the instruction. Of particular note was that only black belts were allowed in, or even watch, black belt classes. The only time the rest of us got to see black belts perform was at the school's tournaments, about every six months.

Even then, the 1970s, karate mostly meant tae kwon do to most folk. In other words, lots of flailing of arms and legs. As is common, tournament was by belt class and each had both forms and sparring. The lower belt forms were mostly of fixed form lasting maybe a minute, with sanchin being the main exception which had a unrelated sequence and was about twice the number of steps. The black belt forms were all of unrelated sequences, the first black being the shortest and simplest. If I recall correctly, the highest level black belts who participated were six and their form went on forever; each black belt level had only one form to learn but it was a doozey. Or so it seemed. Not like the movies.

When it came to the black belts' sparring, now that was jolting. In the movies (then and now) and those tournaments one sees today on ESPN and such, sparring is all flailing. Not in this school. (Tournament was held in a community center across the street from the dojo, i.e. a basketball court.) The opponents took defensive stance, George called 'kumite', then ... nothing. They just looked focused ahead. Another note: this was a traditional style, so gi and barefeet. Well, it wasn't nothing, if you paid attention. One would inch, usually something less, forward. Eventually one sensed a weakness and struck. It was over in much less than a blink of an eye.

George taught the new students himself for a month or two before handing us off to the instructors. One of the bits of wisdom he imparted, largely I expect because it was something newbies all want to know, "what's it like to be a black belt?" George: "A black belt is one who can kill or maim with a single blow." So, the black belt sparring demonstrated that.

So it was with Serena and Venus last night. Serena is top of the game. Venus, when she's healthy as it appeared during the Open, is a smidgen less. But just a smidgen. They were both serving at 120+. Think about that. The notion that two healthy Williams would need three 7-6 sets is foolish. They stayed even in the set until one left a tiny margin. Bam, a single blow. So, it went 6-2, 1-6, 6-3 Serena. Makes perfect sense.

Among many so-called creative professions it is often said by the old hands that as one gains experience, one simplifies. As a young practitioner, you feel the need to show off all your, more less perfected, skills. As you get more experience, you realize that there are a few base realities in your field, and getting them right is a life's work. So, you set off to do that work.

Alas, in the quant and database professions, this very important lesson has yet to be learned. I attribute this largely to academics: in order to attain a graduate degree in either field, a paper, expounding new research, must be produced. For the quants, finding new frequentist methodology got to be a pain in the ass. Some, I know not who, clever grad student decided to disinter Bayes and open up a fresh field of "new research". Bah. Equally for databases, although not so much from academics, but rather from the dropouts who can't grok algebra: NoSql and xml. Double bah.

For quants the reality is that it all comes down to squared differences.

For databases (boo!), it's orthogonality.

You thought I was going to say third normal form, didn't you? Admit it. No. The driving principle of the RM, and the SQL database when done intelligently, is to organize the datastore into orthogonal bunches. If you do that at all times, you'll end up with some normal form that'll be almost impossible to screw up as you extend and develop the application. Orthogonal data hasn't side effects in either code or data.

Independence is a wonderful thing.

No comments: