TOWM quote of the day comes from Ed Sperling writing for Forbes:
The timing of all these pieces couldn’t have been planned better, which is interesting considering none of it was planned at all.
He’s talking about how server farms managed to drastically cut their power bills recently. It depends on several things, all of which happened to crop up at about the same time. They are:
(1) Moore’s Law hit a bump: you can’t keep cutting the size of chips in half without leaking a lot of heat, so hardware manufacturers started switching focus to multi-core processors.
(2) BUT – most software wasn’t ready for multi-core, so server farms just kept running stuff on older chips, but this required a huge power bill for cooling.
The solutions were:
(1) Virtualization got better, so it got easier to run old software on multicore systems without having to rewrite it, which is what people did.
(2) Server farms started looking for ways to cut their power bills.
What’s interesting for me about this article is that Ed Sperling seems to think it’s a minor miracle that all of this stuff got accomplished “without planning.” This is a common misconception about the free market, actually – that there’s “no planning.”
In actuality, Mr. Sperling shouldn’t be the least bit surprised, and that’s because MORE planning goes on in free market economies than in planned economies. Free market economies are the economic corollary of Linus’ Law(Eric S. Raymond vesion): “given enough eyeballs, all bugs are shallow.” Substitute “bug” for “market inefficiency” and you get the point. When literally EVERYONE is allowed to participate in the economy at the planning level, problems get solved a lot faster than if you leave it up to corrupt panels of politically-appointed experts. This is why market economies will always outperform planned economies in the long run. And this is also why it shouldn’t be surprising that what Sperling considers to be a minor miracle actually occurred. There’s nothing mysterious about how server farms figured this out. Give them a power bill that they actually have to pay and customer demand that they actually have to meet and they have a real incentive to solve their problems. By contrast, if internet space was something apportioned by the government on the basis of “social need” or what have you, then what we would get instead would be a bunch of people saying “well, the internet is a social good, so they should have power to run their machines,” and some pol would snap his fingers and it would be so, and all kinds of opportunities for innovation would be missed. And they would be missed not just because the financial incentive had been taken away, but also because the number of entities running the servers would have been much smaller. Rather than hundreds of independent server farms, you would have a handful of government-approved server farms. And so there would be fewer chances to innovate. Ditto the hardware manufacturers, of course. Absent marketplace competition, it’s hard to see what would motivate anyone to design multicore processors in the first place, much less adopt them. Moore’s Law could safely reach a trough, and the one or two quasi-government entities could just sit on their thumbs and say “well, that’s as good as it gets, so here ya go!”
The truth, you see, is that there is TONS of planning – ACTUAL planning – going on in the real economy by real market actors with real budgets and real profit reports and real bosses breathing down their necks to get real work done.
I think there are two problems here.
The first problem is linguistic. The word “plan” in English calls up an image of someone with a spreadsheet listing all his resources, which he then allocates. The point is that it isn’t a word that encompasses much innovation or creativity, or even more than a single-stage process. But the word is inadequate, because in real life planning is often (even usually) done over probabilities about uncertain resources, and it can span several stages in development. When I “plan” a move in Chess, for example, it’s true enough that I focus more on the board as it is now, since that’s something I know with certainty – but a good Chess player will, of course, plan several moves ahead, in response not just to the most likely, but to an array of possible responses from his opponent. And for this analogy it’s probably telling that Warrent Buffet and Bill Gates aren’t Chess players but Bridge players – Bridge being a game where planning is over uncertainty, because you don’t know what the other players have in their hands and therefore have to take odd contingencies into account.
The second problem is an undue focus on planning. It turns out, you see, that economic activity isn’t all about planning, but also about agility. People love to use boardgame analogies for economics (I just did, after all), but there’s a realtime aspect to running acompany too. In these cases, martial arts make a better proxy – because one of the things you learn in martial arts is never to commit yourself before you have to, and in fact to use it against your opponent when he does. The point is that since a fight happens in real time, you can’t always see what will be coming your way, and so you need to be prepared to react quickly to anything. The businesses that succeed are, unsurprisingly, often the flexible ones that use agility rather than strength (which here is “throwing money at the problem”) to respond to whatever comes their way. And so some aspects of economic success we can actually chalk up to anti-planning, to not committing oneself to any particular course of action before all the relevant information is in.
It seems to me that it is these twin errors – (i) an undue focus on planning as being something that is done in a single stage over known quantities when in fact it is typcially done in several stages over largely unknown quantities and (ii) an undue focus on planning rather than efficient reaction as the wellspring of success – that leads people like Mr. Sperling astray. The answer to his astonishment is that his perception is both too big and too small. It is too big to see that a lot more planning than he gives credit for DOES go on – each and every day and by more people than he imagines. And it is too small to see that planning is not the only thing that accounts for economic success. A lot of it has to do with flexibility and reaction time as well.
Since I do not make these two mistakes (though I fully admit I had to train myself out of it – it is a natural human tendency to err in this way, I think), I am not at all surprised to see that the computer industry coped with a potential profit shortfall by both reacting to the situation and implementing contingency plans. I wouldn’t have expected any less, and indeed I expect that every industry goes through just this kind of “crisis” just as successfully at least once a decade.