The Future of Credibility: “everything is what it is because it got that way”

It's Pandaemonium. I came to praise it, not to bury it.

In his tendentiously-titled Consciousness Explained, Daniel Dennett suggests an outline of a theory of the generation of speech. His outline is expressly directed against the classical notion that when I say something it is because I mean it. More precisely, he is arguing against the claim that there is necessarily within me a Central Meaner, perhaps that being myself or else a kind of homunculus, who holds safe the semantic meanings that I intend, and against which all my attempts to frame utterances are compared for semantic fidelity.

Imagine a great horde of senseless daemons, he says to us, rather than a single central homunculus, each with a phrase or piece of a phrase to suggest. These pieces are mere “found objects” and most are senseless or irrelevant. Irrelevant to what? -- to some kind of goal being held, but, crucially, not necessarily a semantic goal. (Dennett gives an example caricature of a person responding to hostility, starting with the angry flush “Go on the offensive!” and passing through “Cast aspersions on some aspect of his body!” on the way to “Say: 'Your feet are too big!'”. He assures us that we can then go home and curse ourselves for not having thought of a wittier retort.) All these “word-daemons” compete to put their mark on each others' candidates for a verbal utterance, and the stream of language they generate is adjudicated over, “yes or no”-style, by a horde of equally senseless “content-daemons.” This chaos does not end when the daemons together have assembled something that meets a Central Meaner's review. Rather, the putative semantic intention is itself modified by the judgments of the content-daemons upon the candidate utterances of the word-daemons.* The intention is “drawn” through an abstract semantic space toward the candidates, just as the candidates are drawn to the intention. What one has is not a kind of bureaucracy but (in Dennett's terminology) Pandaemonium, and it is “a process that is largely undesigned and opportunistic.”

This, if I may remark, is not a model that would have been taken too seriously as recently as the mid-nineteenth century. It is a thoroughly post-Darwinian conception, this supposition that all this chaotic variation can with only subtle environment constraints and interactions nonetheless construct something complex, something with significance, an utterance with a Meaning. Dennett directs this schematic model against the suggestion that the need for a Central Meaner would give necessity to the opposing theory of consciousness that he calls the Cartesian Theatre, (which is, roughly speaking, the idea that there is a single [physical or abstract] place where consciousness “all comes together”). In doing so he links our naïve sense that there must be a Cartesian Theatre to explain our experiences with our naïve presumption that there must be an agent's design behind anything complicated. There is indeed a commonality between them: both are a kind of turtle-stacking. (Check out the graphic on that page!)

We therefore stand properly advised that in our new century we will brook no contention anywhere that sophistication mandates design. To now pass instantly from the ontologically profound to the perhaps merely interesting, and to keep up to the promise of the title and opening paragraph of this post, we in particular note that in the new society of the Internet there is no compulsion to subscribe to a Central Content Provider.

This is a change from the old model, the model of cable television and publishing houses, where the expense of broadcasting mandated a few powerful players. To be sure, Peter Mansbridge isn't going anywhere, but there's a reason why he's now reading out viewer e-mails on the air each night. It's spelled out in this article, which was first published in a magazine and only later reprinted in the author's blog – so don't hold it against it, that it didn't come from a Central Content Provider, that in fact you (maybe) first heard about it from my pointing it out.

For this is the basis for our Internet Pandaemonium: if you know me and trust me, if I have credibility, then my endorsement of content brings that content to your attention. If you agree it's view-worthy, you refer your own close neighbours to it. If not, it need not branch any more along that path, but remember that my other readers might feel differently. The more it's passed around, the more credence and significance it gains in the greater community: some memes gain tremendous attention, or tremendous notoriety, as any longtime Internet reader can verify. As for where that content came from before -- maybe I found it purely by chance, or from another blogger whom I read, who may himself just be another citizen marking out prose or maybe someone with a mission to find selected or special content, like a museum curator, or from a newspaper or other kind of dedicated, professional content provider. (As exciting as our new century is, we would be too reckless by half to lose our professional content providers: they still deserve our respect, even if their roles are changing.) In this particular case it's all-of-the-above: the article I just linked to isn't the link I followed to that blog. This fellow's blog is in my queue pending final decision on whether I should bookmark – add him to my trusted content providers. The ultimate origin of the article, for our purposes, is the author's blog, (strictly speaking, this origin is in the sense that “eukaryotes come from prokaryotes” rather than “eukaryotes come from primordial soup”, to anticipate parenthetically some ideas from the sequel); but every blogger is an author waiting for his work to be cited and to come to prominence, possibly as part of a greater content-complex (an organelle in a eukaryote). It doesn't happen often: the price, the scarce resource that demands differential survival (speaking so as to continue to anticipate), as always comes from the opportunity cost of a person's limited attention-bandwidth for consuming media. So most bits of content don't achieve total renown, but rather roam and hover between heaven and earth. (Mostly around earth: renown like flight is very expensive to support.)

Now let me make good on those anticipations. I used that word -- memes. If you've never heard it, maybe as always the Wikipedia summary can be helpful. In short, an idea is a meme, or more precisely an idea at the size that can be replicated. It might be as small as an idiom of language or the refrain of a song or as big as the text of a book. This is in analogy to the way that a gene is some span of codons in DNA, not of fixed length but rather defined in terms of its being able to be selected for or against. A gene is supposed to do something meaningful to the phenotype, which is either fit or not according to the environment, and this by backward-translation is a selection pressure against the various genes in the gene pool. The notion of a meme is, I believe, an idea whose time has come. Of the increasing number of contemporary popular works on the subject I can't sanction or sanction any, since I haven't read them. I can, however, (not to use my pulpit to promote too many more books) both on general and specific principles recommend Richard Dawkins' The Selfish Gene, which to my mathematician's eye reads not so much as a book about biology as an extended worked example to support a nearly-axiomatic theory about differential survival of replicators. (This description is not meant to put you off it: that's good praise!) The memes, the so-called New Replicators that “live” not in space but in peoples' minds (hence why they're new – not so long ago there weren't any minds), are first introduced by that name in Dawkins' 1976 book, although I imagine the notion was anticipated by other authors previously.

With the meme as part of our vocabulary, more or less everything we've described so far today takes a single shape. They're all stages of memetic evolution.

And while we're here, what about that Wikipedia? I cited it so casually to give a reference to memes, but isn't that where Wikipedia itself came from – some kind of memetic evolution? Take a listen to this talk by the Wikipedian progenitor himself and decide whether it sounds like the same thing. Not to be tendentious myself but I think it does: each Wikipedia article is itself a meme complex adapted under the pressure of its editors and its editors' minds – that last meaning, perhaps, the memes living in the editors' minds? So a Wikipedia article is a complex adapted to the memetic environment that the editorial community represents. If you think the situation is disqualified because the editors are agents with intentions, remember the parable of the Central Meaner: those intents themselves are memes, or perhaps more precisely certain products of memes (in the sense that the phenotype of an animal is a product of its genotype).

This is encouraging in the sense that it seems this is an example of memetic evolution, but, to get back to the original objection of the last paragraph, if we were inclined to be suspicious of the memes and the prospects for memetic evolution, the fact of a Wikipedia citation about them is not going to be terribly convincing. The mere fact of the citation is mere question-begging! On the contrary, it is good and well, we imagine, to say that animals rose from natural selection; but why is it that we suppose the conditions are present in our world of discourse to make possible a memetic evolution by “natural” selection, even if we believe (“in our new century”) that this is possible in principle?

For if the conditions are not there, a Wikipedia citation must surely be a deeply suspicious thing. Supposing that daemons can make wisdom just by nattering suggestions is no better than saying that monkeys can write Hamlet given typewriters nor than saying a mammal came about by accident. Even Plato knew, despite not having Darwin's idea as a counterexample, that sophistication does not imply an agent's design; but chance alone does not gain sophistication without selection pressures causing differential survival. (The “nonrandom survival of randomly varying replicators,” in Dawkins' one-sentence summary.) So rose the animals, and so rise the meme complexes – supposedly. But is there or can there be really such a thing as the so-called “wisdom of crowds”?

What, for example, to make of such a famous experiment as Kasparov v. The World? Let me recall the circumstances of this event to you. In Summer 1999, Microsoft sponsored a chess game played (at correspondence time control, about one move per day) on their website. Garry Kasparov, recently having regained his status as the invincible champion with spectacular triumphs at Wijk an Zee and Linares in the first quarter of the year, commanded the White pieces, (with the assistance of his usual seconds). Captaincy of the opposing Black forces was given to – everyone who showed up: any person could log on to the MSN Gaming Zone and submit a vote for their team's play. The move with a plurality of votes would be the one made. The match was not so uneven as it sounds: four strong junior players were enlisted to provide brief recommendations to The World team, and still more strong players volunteered, including Alexander Khalifman, who went to Las Vegas in Fall 1999, while the game was still being played, and there won the FIDE World Championship. (At this time, the world championship was still a divided title, and Kasparov refused to play in FIDE's events.) Or perhaps after I, and Dennett and Darwin, praise Pandaemonium you might think the game unfairly balanced in the other direction?

No, it was not so. After several months of tough play, during which at times all three outcomes (win, lose, draw) seemed equally plausible, Kasparov was victorious, (in a queen-and-pawn endgame). Is this the logical triumph of expertise? I would be quick, in my Platonic prejudice, to think it so: understanding trumps grasping because that's what understanding is. Similar one-versus-many contests have been held, albeit with less fanfare and publicity, and they have often been decided for the expert and not the putative wisdom of the mob, but far from always. More recently, Arno Nickel, who holds the ICCF (international correspondence chess federation's) grandmaster title, bested the computer Hydra, by a score 2.5-0.5, in correspondence games; and this unthinking computing beast the same creature of awful power that beat Mickey Adams 5.5-0.5 over the board! But Nickel too has lost a game against The World, in the form of the community on the website ChessGames.com; it concluded in January of this year.

So what to make of all this? In a phrase, it's the difference between a mob and a crowd. Fickle folly can be incendiary, but it knows not how to aim itself nor cares what it burns. If a man leads a mob, it is at most as wise as he, and often less. Irina Krush, one of the World Team coaches Microsoft hired in the match against Kasparov, expressed her displeasure when some of her mob's decisions decided against her and Khalifman. A mob can beat Kasparov if Khalifman can. After it gave such a tremendous fight it is ungenerous, to say a minimum, to put that World Team closer to the monkeys than to Shakespeare, but one idly wonders if simply the presence of better tools, like a ChessBase wiki to replace the primitive forum design of the MSN Gaming Zone, might have made the difference.** Anyway, it's quite possible that to beat Nickel in correspondence games is more impressive than to beat Kasparov, so the proper tools have surely been developed. But the lesson is clear: It is the networking of the crowd that sets the stage for the miracle of memetic evolution, and not the noisy, violent hub-bub of the mob.

And, to get back to our Wikipedia problem after an extended parable, a fringe article is only as good as its first and only editor-progenitor, but a well-eyed treatise is a tremendous thing – as long as Wikipedia itself is properly built to encourage the right selection pressures. Remembering the talk, the last pieces come into place. You have administrators, (Dennett's content-daemons, the judicial counterparts to the word-daemons striving to craft prose), and they, of course, don't need to be experts in the article subject areas, no, not in anything but encyclopaedia mediation, -- and then, more or less, the “undesigned, opportunistic” process of evolution can start. Alexander Khalifman is a tremendous “chess expert,” but that's a word-daemon: he isn't on the right ontological level to exploit a Pandaemonium to challenge Garry Kasparov.

After so much chatter, let us remind ourselves that – it is a fait accompli; Pandaemonium is already here! It came in the form of the blogosphere and YouTube and other networking sites. (Pre-Internet models suffer from limited connectivity and content distribution – somewhat deficient for good examples of a functioning Pandaemonium, and so every film studio executive dreams of that rare word-of-mouth buzz.) Bloggers write and post content and link to each other; YouTube users have schematic personal pages on which they can, in addition to posting their own videos, link to particular other videos or to other users' personal pages. This is the minimum requirement. It doesn't have to be one or two sites in particular, and may not be in future. Indeed the Internet crowd can sometimes be fickle in its endorsement (the various blog-hosting services all rise, compete, and fall among themselves). What's important is just the structure on which the community builds itself, and any sufficiently self-connecting framework could do. From there we little daemons, simultaneously both word- and content-, we take care of things themselves.

And quite well, too. So ends my speech of praise.


(Endnotes.)

* For example, consider the "seductive turn of phrase." Where did the title "The future of credibility" for this entry come from, anyway? It's apparently a play against an earlier entry entitled "The end of credibility," (which, if you're looking for it, was the last of "Three Short Comments from Princeton"). Is this really the best I can do? The two entries don't seem to have a lot to do with each other, other than being about blogging. Maybe they're both about content distribution, from two different sides. Maybe it's just a weak play on words. But it has such a hold that the word "credibility" even gets a mention a little later, in the middle of a very important paragraph. There's supposedly an abstract semantic idea that this paragraph is to communicate, and that idea is independent of its instantiation in text, but in particular the "trust metaphor" to explain the low-level links between blogs is directly connected to this curious title. Can I say which came first in my thinking? For surely the meaning of the text would not be quite the same with a different metaphor.

In other examples, a "seductive turn of phrase" becomes its own justification, divorced from any external semantic concept. An easy way for this to happen is from grammatical ambiguity, say due to excessive editting for style over content.

Incidently, the titular quotation, "Everything is what it is because it got that way," is from D'Arcy Thompson's "On Growth and Form." (Peter Medawar called this "the finest work of literature in all the annals of science that have been recorded in the English tongue." I haven't read it. It's about 1200 pages, which is rather longer than this post.) It is a little more obviously connected to the technical side of the subject matter.


** Everyone who has ever spent a little time slumming on a forum whose theme yields to certain frequently-asked-questions knows that Sisyphus himself hadn't such a futile task. I have examples in mind, but they're such dangerously contagious and stultifying memes I don't dare quote them – they're the memetic equivalent of the flu.


Postscript. Sorry about all the broken links; I've fixed them. They were there because I was composing this entry in a word processor (for the obvious reason) which was automatically turning all the quotation marks into smart-quotes (that angle toward text, like “...”) rather than the (uniformly-oriented, like "...") quotation marks that one needs to put in HTML tags.