Arnold Kling  

Hayekian Linux

Biofuels Sense and Nonsense... More Hawkish than Mankiw...

A reader sent me a link to something Linus Torvalds wrote in 2001.

You know what the most complex piece of engineering known to man in the whole solar system is?

Guess what - it's not Linux, it's not Solaris, and it's not your car.

It's you. And me.

And think about how you and me actually came about - not through any
complex design.

Right. "sheer luck".

Well, sheer luck, AND:
- free availability and _crosspollination_ through sharing of "source
code", although biologists call it DNA.
- a rather unforgiving user environment, that happily replaces bad
versions of us with better working versions and thus culls the herd
(biologists often call this "survival of the fittest")
- massive undirected parallel development ("trial and error")

I'm deadly serious: we humans have _never_ been able to replicate
something more complicated than what we ourselves are, yet natural
selection did it without even thinking.

Don't underestimate the power of survival of the fittest.

And don't EVER make the mistake that you can design something better than what you get from ruthless massively parallel trial-and-error with a feedback cycle. That's giving your intelligence _much_ too much credit.

Comments and Sharing

CATEGORIES: Austrian Economics

COMMENTS (13 to date)
caveat bettor writes:

The more complex a design is, the less it was designed?

This sounds a bit religious to me, because the data does not fit the hypothesis.

John Thacker writes:

Although surely evolution had more time to work. Ruthlessly massively parallel trial-and-error with a feedback cycle is a very good development technique. But certainly in any real development cycle run by humans, there's a room for a designer. The Honeycrisp apple was certainly developed faster with a directed effort than completely undirected effort ever would. There's a balance between design and feedback. (And in actual program design, there's collective human intelligence in the feedback, just like in economics.)

Of course, all this is inherent to analogies to evolution that try to do too much. Constantly analogizing processes that are "parallel intelligent design with openness and feedback from users" with evolution is likely to confuse people about evolution itself.

Matt writes:

That's stupid. I can't believe smart people actually believe that's possible. What a leap a faith! Since you're talking Linux, let's use that as an example. It wouldn't matter how many times you tried throwing together a random set of ones and zeros (billions of tries, trillions of tries, whatever you like) it is *impossible*--did you catch that?, IMPOSSIBLE, to create the Linux operating system to do all the things it does just by chance. And you think that the whole universe just *happened*? You people really have greater faith than anyone I've ever met.

LOL. Friggtards!

Brad writes:

I'd have to see context, but I'll (I'm sure wrongly) assume he's making some kind of point about open source software development. And I'll give that crowd lots of credit for reimplementing the obvious and making the economics of that kind of activity cheaper, but for the 1,257th time, someone name something new and revolutionary that's come out of open source. Th incentive that drives creativity isn't there. Real artists see no virtue in being starving artists.

fundamentalist writes:
That's stupid. I can't believe smart people actually believe that's possible. What a leap a faith!

Right on! Some things are so incredibly stupid only an intellectual could believe them, to paraphrase Orwell.

John Pertz writes:

I think the dissenters in this comment section are incorrectly conflating Torvald's view of unplanned order with the rational planning and design that occurs amongst agents who comprise the aggregate order. You can design something for sure. However, without the process of survival of the fittest or trial and error that design is largely worthless. It is this ruthless "unplanned" system that constantly beats out the bad information and elevates the good to its highest use value that is responsible for the incredible complexity of human beings or for that matter the market economy.

jb writes:

Caveat - no, the more complex an entity the less it was designed.

Matt - not worth responding to.

Brad - innovative/revolutionary stuff that has come out of open source... I assume you mean "paradigm shifting" - The Rails web platform is one - it truly shifted the way we software developers do web development. Depending on how nitpicky you want to be, the original Wiki was an open source project. (The source code was made freely available). Gnutella is a software system to distribute data in massively parallel ways w/out a central authority. Condor was in the 90s the very first distributed "virtual machine." It was public domain. Sendmail was revolutionary in how it handled delivering email in structured ways. It is open source. Numerous revolutions in "private browsing", TOR, for example, are open source, because of the nature of what they do. The program "Make" revolutionized building large software projects. Jabber is a revolutionary way to interact with instant messaging. IRC was, for a long time, the only way to do large-scale community chat. The Internet "News" program/server family was the first Internet-scale forum and blazed a huge trail there.

Need something more "user friendly"? How about domain names? The original invention was created and shared openly in a social environment. The incentive that drove creativity in this case was the frustration w/the cumbersome way to find websites before then.

That's just off the top of my head. I'm sure there are quite a few more revolutionary open source software systems out there.

But you've inspired me to write a blog post where I will list all the different types of software that were revolutionized by open source. Perhaps that will end this ridiculous claim that there's no incentive to innovate in Open Source.

Jared writes:

Matt & Fundamentalist: Go out and learn something, anything, about Evolutionary Computation. Thousands of very complex systems have been designed through such random processes, and they not only work, but many work far better than the best human-designed solutions. Read about Holland or Koza and then tell me "only an intellectual would believe it." You might as well be arguing against heavier-than-air flight because you have no legitimate arguments beyond "ha ha dumb scientists!"

Brad, as to motivation, a lot of useful systems get designed because one or two motivated hackers want a better solution to their own problems, not because they expect to get rich. I suspect that's what's motivated most inventors in the past as well. But the difference between today's hacker and yesterday's garage shop tinkerer is that the better mousetrap he built for himself couldn't be shared with thousands of others because there were only as many as he could make.

Also, I think the motivational aspects of wanting to solve a problem for its own sake is often overlooked. Most people getting paid to design the next corporate widget aren't going to be staying up late an thinking about it, or wrestling with the problem in the shower, forgetting to eat dinner because they're lost in thought. But a lot of people working on problems they've chosen to attack, problems they find interesting, will do those things just for the satisfaction of finding the answer. And those same people overwhelmingly support free software.

Brad Hutchings writes:

jb, You forgot to list GIMP, as it was the first graphics editor ever, right? Also, "make" was introduced in PWB/UNIX and rolled into System III and V, both commercial releases. Can you point out a single application or computing paradigm that my Mom might know about that became popular through open source development and release? i.e. that did not require a company to refine and popularize for the non-geeky masses? The closest open source has come is making components, and the real market innovation in those cases is the price being (near) free.

Troy Camplin writes:

He forgot to include self organization, emergence, information, and dissipative structures. These also all contributed.

We are so impressed with ourselves for what we have done with physics and linear dynamics, but the fact of the matter is that we have done almost nothing with a tiny corner of reality, using an almost insignificant, marginal form of dynamics. We've even done a few cute things with chemistry, and even fewer cute things with biology, but the fact of the matter is that we have not come even close to doing much of anything with either one of them, let alone tapped into the nonlinear possibilities of physics. Linear, entropic structures can be designed quite well by humans. But when we have tried to do anything with anything expressing nonlinear dynamics (which is something like 99.999999% of the universe), we show how woefully ignorant and incapable we truly are.

Dr. T writes:

Matt completely misunderstands evolution and then makes an incorrect analogy to software development. Evolution is not about starting with a broth containing adenine, guanine, etc. and creating DNA from scratch. Its about starting with existing simple organisms that undergo changes to their genetic code to make them better, more complex, and more survivable.

The appropriate analogy for coding is to start with simple libraries, objects, and subprograms that can be altered, aggregated, reordered, etc. Torvalds believes that with massively parallel processing and millions of iterations (where bad code dies out and good code reproduces) programs could evolve that are more complex and functional than programs designed by a software team. This already has been shown to work on a smaller scale.

Matt writes:

I'm perfectlhy ok with the idea that many simple systems come together to form more complex systems, which can be put together with other systems forming even more complex systems. It is a stretch (read, farce), however, to think that these simple systems just melded together randomly and somehow their interfaces just worked. Hmmm. Cool. Yer so smart.

It's funny how upity "scientists" get when someone says "ha ha, stupid scientist", but you should know that what really impresses me is mathmatics. A scientist is a mathmaticians piss boy. If you can prove your stuipd theories with mathematical equations, then let's talk. Otherwise, "ha ha, stuipd scientists".

Oh, and Jared, the question is have *you* read Koza? He makes a stronger case for needing complex thought processes (since that "design" word has been so horribly bastardized) before *creating* simple systems that can interact with each other. In his work, he's shows (probably unintentionally) an utter necessity to have a creator who is... wait for it.... intelligent.


Dr. T writes:


There is no arguing with someone who believes that science is nothing unless it can be encoded or converted or described by mathematics. The religion of Math cannot be falsified.

Comments for this entry have been closed
Return to top