Monday, July 14, 2008

Chomsky! Pinker! Get In Here!

[OK, OK, it is a Lisp blog but if I can refute Chomsky's entire body of work on language acquisition in one post, why not?]

These academic PhD geniuses have a problem. They are geniuses. They think they can figure everything out by thinking, like Feynman fixing the neighbor's radio at the age of three or something close to that by thinking. Great story, buy the book. Forget which.

Chomsky and Pinker merely repeat the error of the AI crowd who sniffed at NI (I just made that up, it means "natural intelligence") and said, "NI?! NI?! We don't need no stinkin NI!". They said that because they were geniuses they knew that what NI did with pretty much organic chemistry they could stomp with (drum roll, please) reason.  Ooooohhhh, excuuuuse me!

I guess I can leave Pinker out of this, he did not add anything to Chomsky, he got drugged in only because a correspondent mentioned his "work" on blank slates. I love that "work": Oh, let me lean back and draw on my pipe and I have a PhD so if I think babies learn something too fast they learn them too fast, dammit! As if. And believe it or not I am coming to my point. But first a science trick, a prelude to my witnessing of the refutation of Chomsky. It has no bearing on anything else but it involves a cute girl and something I could not believe when I saw.

There I am in a sports bar/shore bar celebrating a great week almost delivering my world-changing Algebra software when I see a lovely lass drop her drink. Three or four feet, to the floor. It was a plastic cup, which helped a lot come to think of it, because a glass glass being more rigid would not have absorbed the energy and there would have been a wicked rebound preventing what happened from happening: the drink landed upside down and just stood there, trapping half the drink inside. The bad news being there was no elegant way of finishing said drink, but that is not my point. Did I have a point?

Well, if I had to have a point, it would be that no one else was all that impressed. Clearly I need to spend more time in bars late at night watching drinks being dropped and I do not think it is possible to spend more time in bars late at night than do I so I guess I need to find bars with clumsier people, good luck on that.

Oh, and the other useless and irrelevant to the demolition of Chomsky/Pinker thing I learned is that it takes just one unauthorized speech on world affairs (well, he was mumbling pretty badly, I am not sure it was on world affairs) over the unattended band microphone to get escorted unceremoniously out the front door at the sports bar in question. Right, Chomsky.

So I look up and I see a skateboard competition, lucky enough to be watching the winner's winning run, and lucky enough for my brain to clear long enough to understand what I am seeing, a Johnsonian kicking of the rock I refute you thus Dr Berkeley of Chomsky/Pinker by some sixteen year-old if that. What did he do, and why did my brain have to clear? Two good questions.

What he did was give himself a push across a flat floor in a reasonably small room (no gigantic ramps and concomitant acceleration/stored momentum) and upon reaching a ramp at the same time push off in such a way that the skateboard (unattached to his feet for those who do not know) elevated with him to a height just above a rail (er, just an elevated horizontal pipe mebbe a foot off the floor) such that he could land on the skateboard in midair just as the tip of the skateboard landed on the rail and using his/their forward momentum proceed to slide along the rail for a couple of yards before again levitating off to land somewhere else to commence another trick. 

And before the anti-blank-slate morons can scream "Oh that is easy, I have the algorithm right here" allow me to break their hearts and add that as he did the initial elevation springing the board into the air with him (a yawner in these competitons) he took the trouble to pop his deck into a barrel roll (or two, who knows at that speed?) before regaining it to land the rail toe-slide. If the MIT geniuses think that is easy lemme just say that I have spent years in Central Park watching the same people practice on skateboards for years and they still cannot jump over a small backback and land on the skateboard on the other side.

So... these people have skateboard organs? Or do you think learning to barrel-roll a jump into a rail toe-slide is more important to them than is mastering language to an infant so their motivation is stronger? This would be a reasonable belief if one has never been in the presence of an infant who will join in any conversation it hears when it neither understands nor can generate one intelligible syllable.

Oh, and why did my brain have to clear? Because what I saw was a commonplace: inconceivable performance easily explained by sufficient practice. If a recent blog had not been on my mind I would not have given the skateboard performance a second thought, nor the motorcycle-flipping organ manifested in the next story on ESPN.

Come to think of it, this is about why Lisp wins: we simply are not as smart as we think we are. I like to say that I cannot write good code but I do know when I have written crappy code. Recognizing is easier than generating. Hell, crappy is just one bit, right? But how do we produce a huge stream of uncrappy bits? Good luck.

What we want then is an agile language that lets us create awful code really fast and fix it as fast as we identify the problems. Pretty soon our software organs cut in and we are creating not-so-awful code up front. Even better.

Now I just have to figure out an elegant way to finish a drink upside down on a barroom floor.

7 comments:

MaysonicWrites said...

Put some saran wrap down next to drink, slide drink over (losing slight amount).

Gabe said...

1.) Poke hole in bottom of cup
2.) Insert straw
3.) Enjoy!

Kenny Tilton said...

haha, I love you guys. I should have said "dignified"! And I should have known an engineering audience would focus on, well, the engineering. :)

Dave Roberts said...

Come to think of it, this is about why Lisp wins: we simply are not as smart as we think we are. I like to say that I cannot write good code but I do know when I have written crappy code. Recognizing is easier than generating. Hell, crappy is just one bit, right? But how do we produce a huge stream of uncrappy bits? Good luck.

Amen, brother!

Anonymous said...
This comment has been removed by a blog administrator.
TruePath said...

Your missing the point of the Chomsky argument. It's not merely that that learning language requires an algorithm that none of us can think of. The argument is that it requires an algorithm which we have theoretical reasons for believing can't exist.

More or less the point is that there are so many logically possible ways that language could be constructed that no possible algorithm could ever determine which language was actually being spoken. Maybe it's right or maybe it's wrong but it has nothing to do with us being clever.

It's like using a one time pad to encrypt your communication. If your enemy still manages to decode the information (and it was truly random) you know he didn't develop some fancy algorithm to crack the encryption, no such algorithm exists.

Kenny Tilton said...

@Truepath: "It's like using a one time pad to encrypt your communication. If your enemy still manages to decode the information..."

That is the analog for language acquisition from which Chomskyans wish to argue? Good grief.