The Computational Universe

Searle’s Chinese room example makes good sense to me. If a machine (or a person, as in the example) merely manipulates formal symbols according to some rulebook it itself did not create, then the fact that it generates intelligible or useful answers—or, indeed, even manages to pass the Turing test, which would be a staggering feat—is not evidence of thought. But because we are incapable of determining whether those around us are actually thinking or just juggling symbols and responding in ways we expect thinking people to do, we cannot know, but for our common-sense confidence in their humanity, whether the internal processes of those around us are any different from those of a computer.

This truth led us to a larger question, one concerning “consciousness”—something the class was very unwilling to attribute to a computer (and rightly so, if Searle’s objection holds true, as I think it does). It seemed implicit in the discussion that consciousness is the foremost feature that distinguishes man from machine or lower-order animal. I would contend that this sentiment is only partially true. For it might very well be the case that the mind works after the fashion of the computer, the only difference being the sophistication of the algorithms it employs. The human would thus become, in principle, fundamentally knowable. If that reductionist view of the self is true, then a particularly able computer would be just as much human or conscious as any of us.

Only if our minds work in a way entirely different from the computer’s algorithms is consciousness a valid concept. Someone making an argument from religious grounds might suggest that it is the soul that animates the mind, which therefore works in ways not so crudely formulaic as the computer does. Indeed, if the mechanical view is untrue, then the mind must be, to some extent, unknowable. For if its processes were known, then they could be compared against those of a computer and, in principle, superseded in quality by a more capable machine. An inimitable human consciousness can thus be considered in a distinctly mystical light. Though I think myself unfit to comment upon the matter from such a perspective, I do know that we must in some way account for our shared (I think) insistence that something separates us from a hypothetical, perfectly humanlike machine.

Sir Tim Berners-Lee seems to be one of those brilliant men cursed with a feckless tongue. I’m sure he had good things to say last week, but for the most part, he just didn’t say them. That is not to say, however, that I did not find several lines of argument in his presentation worthy of consideration and remark. I was particularly interested by his discussion of Wikipedia, the online encyclopedia anyone can edit at any time, and other such decentralized ventures in social terms. The parallels between these and various social and economic constructs—democracy, the market—are obvious. The “fractal tangle” upon which Wikipedia is built mirrors the structure of a modern society, which is created out of countless social organizations and relationships, just as the market is sustained by even more numerous combinations of buyers and sellers. (And like the market, Wikipedia will be stable and positive over an infinite time horizon, but not necessarily so at any given moment.) All the complex relationships upon which Wikipedia and the market are both founded—an entirely uncoordinated mixing of decisions on a grand scale—dissolve into one large community, political, social, or economic (in a way not altogether dissimilar from that of the cellular automaton, we might note).

Despite the power of such relationships to fulfill the desires of parties on both ends of a deal or contract, there are pitfalls, as Berners-Lee noted. He noted a tension between the ideal of a decentralized network controlled by none but enjoyed and improved by all and possible commercial abuses (fraudulent emails from Nigerian princes temporarily bankrupt or somehow unable to their own open bank accounts to store away their uncountable millions, for instance). There is a similar tension between promise and pitfall in the realm of intellectual property, where the line between dissemination of information for the good and fraud and other abuses is sometimes murky or too easily crossed.

Turing machines are tapes theoretically unlimited (so that any program, regardless of its length, might be executed), fully erasable, and divided into squares of a given size, which each contain one and only one symbol and are read singly. More efficient programs than this exist, to be sure, but this one was conceived in order to illustrate the scope of possibilities that can arise from simple rules. For, according to the Church-Turing Thesis, a Turing-Post program are able to reproduce any program that can possibly be constructed—as, indeed, can a program written in C++, Java, or any other commonly used programming language.
Though a universal Turing-Post program can be created to act in the same manner as any possible Turing-Post program, indeterminacy still inheres in the whole system in the form of the halting problem—we have no ability to determine whether a given program will continue ad infinitum or at some time end. Consider Conway’s Game of Life or Langton’s Ant, which, proceeding from simple rules, quickly achieve a complexity beyond our comprehension (they are therefore cases of “emergent complexity”). Both, of course, can be represented by Turing-Post programs.
The combination of Turing-Post programs’ impressive capacities and the halting problem, then, in some ways encapsulates our ambivalence concerning computers themselves, which might be considered to be nothing more than exceedingly intricate Turing machines. For while they are capable, in principle, of solving most questions short of the metaphysical, in practice their raw calculations, much like the Theory of Everything about which we read, sometimes lack the explanatory power necessary for true understanding.

I’ll admit that I didn’t take the time to read everyone’s post, but there seems to be general agreement that the Scribbler robot isn’t a very precise machine. It doesn’t always travel in a straight line, it turns inconsistently, it blunders around obstacles, sometimes smashing right into them when its sensor doesn’t function as we might hope it would. Even when traveling back and forth over exactly the same time interval, the Scribbler travels different distances, tracing a noticeably uneven line upon the drawing paper. It seems difficult, then, to relate this simple device to the complex machines that assemble cars, stamp metal, stitch fabric, and do a host of other tasks using intricate motions repeated with the utmost precision.

Then again, the Scribbler is capable of executing a series of reasonably complex maneuvers, and the thing isn’t entirely incorrigible, for the action of its motors can be fine-tuned to correct for errors. Not only that, with the appropriate skill (which I know I do not yet possess), one might program the Scribbler to perform some sufficiently elaborate maneuvers to draw tolerably good pictures and play decent music. And, equipped with its stall sensor and obstacle sensor, the robot is, at least in principle, very similar to the Roomba vacuum cleaner, which, in addition to other ventures, has enabled Brooks’ iRobot corporation to turn a modest profit. Likewise, the Scribbler’s component parts aren’t much different from those in robotic dolls, which enjoy several advantages in practicality over the Scribbler. The imprecision of their motor’s actions matters little, so long as that imprecision remains within tolerably low levels (we wouldn’t want a baby’s face to sneer when the relevant program called for it to smile, for instance); the necessary parts are fewer and need not relate to each other as exactly as the Scribbler’s motors must in order to generate the requested shapes and lines; and the toys have some value beyond the classroom, where the Scribbler schools the ranks of the computer illiterate in basic robotics and programming. In short, simplicity is a virtue, and though the Scribbler is perhaps more complex than robotic dolls currently being sold, it suits our purposes well enough.

I’m Daniel, a Mathey freshman from Cleveland, Ohio (well, from the exurbs, really, but let’s just say it’s Cleveland) who will probably major in economics with certificates in Latin, Finance, and maybe Greek. I also intend to study abroad junior year, probably in England. I enjoy reading history, golf, working out, running, playing with my adorable black Labrador, and traveling.
I decided to take this class, as it seems many of us did, in large part to correct, at least partially, my appalling lack of understanding when it comes to computers. That I’m not incompetent at running programs doesn’t excuse the fact that I don’t really know how they work—something of which my more computer-savvy friends remind me on a regular basis. If I enjoy this class, I imagine I’ll take a programming class as well, but this class has the advantage of the ST label (it’s always nice to kill two birds with one stone).
I’m also looking forward to learning about cryptography and the mechanics of the hardware behind the electronics we use on a daily basis.
Finally, let me just say that those chairs in class—the ones that lean back so far you feel you might fall right over—are awesome….Now I’ve done it. 200 words.

Update 2:00 P.M. on 2/11/06: I use a Dell Latitude notebook.