By Kevin D. Williamson
Sunday, May 10, 2015
‘Finally, neural networks that actually work.” So reads
the headline in Wired, and, really, haven’t we all been waiting? (Yes, we have,
even if we do not know it.) The article concerns artificial-intelligence
innovator Jeff Dean, who as an undergraduate at the University of Minnesota 25
years ago created a rudimentary “neural network” — a computer system
sophisticated enough to learn — but was hobbled by the available computing
power of the time. Now working at Google, he’s helping to create vastly
powerful and subtle networks that recognize faces and spoken language.
A few pages over, there’s a wonderful if unintended
counterpoint: a profile of Megan Smith, an ex-Googler whose official title
these days is “chief technology officer” . . . of the United States government.
She is the third person to hold that post, which was created under the Obama
administration. “Why can’t the federal government have websites and digital
services that are awesome?” she asks.
Why, indeed?
She might start by asking what it takes to get fired for
incompetence: CGI, the web-design firm in which one of Michelle Obama’s college
cronies is a senior executive, was lambasted for the debacle that was the
Obamacare rollout, and eventually fired — or as “fired” as the politically
connected get in Washington; some months after having its Obamacare contract terminated,
CGI was awarded a multimillion-dollar IRS contract to manage . . . Obamacare
issues. Mrs. Obama’s friend in the firm, Toni Townes-Whitley, was promoted from
senior vice president to chief operating officer.
It’s enough to fry your old-fashioned meat-based neural
network.
These are interesting times, for those with eyes to see.
The news is full of economic stagnation, but the suburban Starbucks where I go
for my afternoon caffeine jolt apparently generates enough cash to justify a
periodic Brinks pickup: That’s a lotta latte. Behind the Brinks truck is a new
BMW i3, one of those nifty little electric cars that are slowly displacing
their lumbering internal-combustion ancestors. A strange phenomenon: When the
first hybrids came out, they were popular among my enviro-lefty friends, who
gave up their old FJ Land Cruisers and embraced the Prius as an act of consumer
self-mortification. The much more radical full-electric cars, such as the
Tesla, are showing up not in the garages of my eco-nut friends, but in the
garages of my car-nut friends — indeed, practically all of the Tesla owners I
know are full-blown right-wingers. (Okay, selection bias in the sample,
admittedly.) Tesla owners are an enthusiastic bunch: Last summer, a man I had
known for about 42 seconds handed me the keys to his new $100,000 sedan with a
mandate: “You have to try it.” We were in Montana, so there wasn’t much to hit
other than the odd stray bison, and it was indeed the sort of thing that makes
me feel like I’m living in a William Gibson novel, which is one of my favorite
sensations.
The electric car is a great example of Arthur C. Clarke’s
third law: “Any sufficiently advanced technology is indistinguishable from
magic.” The day before yesterday, the question about electric cars was: “Will
they ever be good enough to really compete with gasoline-powered automobiles?”
Today, the question is whether makers of traditional internal-combustion
engines should even bother trying to catch up or just throw in the sparkplug.
One of the rarely appreciated aspects of the capitalist
model of innovation is that the wealthy subsidize the development of products
for everybody else: The mobile phone is a case study in that process, as is the
electric car, as indeed were ordinary cars. The firm that developed the first
automotive air-conditioning and power windows was a high-end marque that
despite its landmark innovations is no longer with us: Packard. The Bonfire of
the Vanities–era financiers who carried the first mobile phones paid for much
of the research and development that made them ordinary products for
non-gazillionaires. My own financial means at the moment do not, alas, afford
the purchase of the new plug-in hybrid from Porsche — which is a million-dollar
supercar — but the technologies developed for the 918 Spyder will make their
way through the marketplace the same way that the automatic transmission
(Oldsmobile, 1940), the supercharger (Mercedes, 1921), and the independent
suspension (Mercedes, 1933) went from being expensive options on cars for the
rich to being standard equipment on your Hyundai. We get our futuristic
21st-century cars the same way Johnny Cash got his Cadillac in 1976: One piece
at a time.
F. A. Hayek wisely observed that this sort of
experimentation is socially beneficial in no small part because the rich can
afford to make mistakes and to follow technological dead ends: The technologies
developed for the $100,000 Tesla or the $800,000 Porsche may not end up being
the winners in the long run, just as many of the features of the 1983 Motorola
DynaTAC were surpassed by those of competitors; but chances are that your high-tech
supercar is not your daily driver, much less your only means of conveyance to
work or the vehicle you use to get your kids to school, and if you own one, you
can probably afford to make a sub-optimal choice. Most of the people facing the
Betamax-VHS dilemma in the late 1970s were well off enough that going the wrong
route caused no economic hardship.
Nobody missed a meal because his Apple eMate was a bad
investment.
The early-adopter process is a way of channeling capital
that is beneficial and relatively painless — and even fun, for a certain kind
of person. (The founder of this magazine so loved WordStar, an early word
processor, that he appeared in an advertisement for it and endlessly sang its
praises: “I’m told there are better programs,” he told Time, “but I’m also told
there are better alphabets.”) That process is also ruthlessly efficient: Packard,
Polaroid, Zenith — the corporate graveyard is full of ghost brands, legendary
names that have outlived their products.
Technological and economic evolution is like biological
evolution in one important aspect: Death is the instrument of life, and the
process of innovation is inseparably linked to the process of extinction.
And that, if U.S. Chief Technology Officer Megan Smith is
actually wondering, is why the federal government can’t have websites and
digital services that are awesome. Non-performing
federal agencies do not go bankrupt, federal bureaucracies do not see their
shares tank when they do poorly, and government entities do not have their
assets acquired by more effective competitors. Political bureaucracies are creatures
doing violence to the evolutionary equilibrium — dinosaurs running amok in
modern technological civilization, and Jurassic Park taught us how that turns
out.
Chief technology officer? Sure, couldn’t hurt to try.
But if anybody knows a really good dinosaur wrangler . .
.
No comments:
Post a Comment