Totalizing projects (Jan 11, 2006)
In one sense, by "totalizing" I simply mean a intellectual programme that aims for global scope in some specific area of inquiry. For example, PlanetMath has the unofficial goal of becoming a comprehensive math reference work (I say "unofficial" because this goal hasn't been recorded on the PlanetMath Goals page, but I'm not sure what it would mean for PlanetMath to have an "official" goal). Wikipedia advertises itself as being a free encyclopedia that anyone can edit (thus, it aims to be global in person-space; and it is generally considered to have good, and certainly improving, coverage of "encyclopedia content space"). Cyc (and projects like it) have to do with another kind of "encyclopedia" all together, namely, a "thinking" one; the idea is to get global, actionable, coverage in the computer of something called "common sense"; as I understand it,
(approximately speaking).
This last example seems to have a chance of sharing the next sense I understand for "totalizing". A computer with very robust "common sense" abilities might (in theory, or say, by definition) be able to read the other resources mentioned here, and, in a sense, subsume them. This sense is both technically and legally tricky; a sketch of the legal issues is given on this page, but our understanding of these issues is far from complete. The technical matters (philosophico-mathematical and also socio-hack-a-nomical) are probably even harder to deal with (but luckily we have had a number of top minds, from Leibniz to Stallman, and various places in between, thinking about these things).
I think of the situationist's "total revolution" as belonging to (and mutually extending) this second category of "totalizing projects", and, in particular, I think of "free culture" as being one of the key current-day instantiations of the "totalist" vision. (Note, these views are probably somewhat atypical, but they are at least roughly compatible with the standard views.) The key step to a third and final understanding of "totalization" is the step that takes us from purely intellectual projects to projects that take the same perspective on the physical world. Given the interconnections between "information" and "reality", this step is probably not unintuitive. I certainly acknowledge that to make this interpretation jive with the standard simulationist-hyperrealist thinking, we'll need to do some work, but the point is, roughly speaking, that by simulating reality in an increasingly "total" way, we come to have increasing "control" over reality. Of course, since reality is practically speaking infinite, and simulation necessarily finite, increasing control ought not give us the illusion of complete control. A totalizing project is total in scope and extent, but seems unlikely to obtain truely total control. (I won't say that this couldn't happen, however, I am unable to conceive of what it would actually mean if it did happen; hence, I say it is unlikely, so that I can focus on things I might have a better chance of understanding.)
Depending on which of the three "levels" of totalization we're talking about, the justifications may change. Individual users might like the feeling of helping to build something big, important, interesting, and beautiful (the wikipedia ~ pyramids analogy, although note that the latter was, as I understand it, more of a slave-labor sort of operation than a happy/smiley volunteer/freetime activity). They may get significant "content-level" benefits from participating in wide-scope intellectual projects, and they may also have a chance to learn some interesting "meta-level" skills (social networking, coordination of large effort, working with other people, etc.). In the case of totalizing AI projects, the benefits are similar, but are of course related mainly to the content areas of AI ("how do we think", "what does it mean to think in general", etc., and in the case of free AI, "how do knowledge communities and cultures work"). Of course, the AI crowd is all the closer to the third level of totalization (should they choose to go that route), and there may be certain psychical (and later, tangible) benefits of participating in the build-up of a "total revolution". The actual concrete benefits to be obtained from total revolution itself are (as above) increasingly difficult to conceptualize the more "total" things become. But certain likely effects can be sketched, or at least imagined: e.g., reducing transaction and other logistical costs, controlling pollution, various improvements on the technological and medical fronts, greater empowerment on the social front, as well as all that fun we'd have with all the knowledge that would be available. Well, no doubt there will be those who can more easily imagine a total revolution going awry, and us ending up in some dystopian end-state. I can't argue this matter point by point, but I think a fair stock response is that we could end up in a dystopian end-state without total revolution of the kind I'm speaking of, too, so what do we really have to lose? Of course, this discussion is probably too vague for many people to appreciate; I guess that's OK, because we can come back and fill in more details later. This is, incidentally, is one of the usual tricks of totalization.
Sure. Let's take the Hyperreal Dictionary of Mathematics project as an example. By itself, it is sitting roughly in between the first two levels of totalization, that is, between global content coverage in some specific domain, and strong, general, "common sense" AI. Of course, the HDM wouldn't be expected to have "common sense" about shoelaces, or oranges, or birds, or human emotion; rather, as indicated by its name, it would tend to have common sense – or just broad, increasingly "total", knowledge about mathematics (and, yes, the "real world" as it applies to mathematics, and, presumably, mutatis mutandis, applications of mathematics to various things in the real world). If you're familiar with the ideas of the HDM project, you'll notice that it may have an edge on, or, from a slightly different point of view, something very valuable to contribute to, a project like PlanetMath, in terms of the "global scope" goal we mentioned above. This is because managing the huge amounts of information relevant to building a comprensive resource in mathematics becomes an increasingly computational problem. The completion of the PlanetMath project may require, and so, imply, the HDM (in some form). This is one of the features of totalizing projects: that they do, in fact, tend to become more and more total. Similarly, the HDM, with its scholium system and its (putative or hoped-for) broad and deep knowledge of mathematics, seems poised to make the move to other knowledge domains, graduating, as it were, from something at level 1.5 to something around level 2 or level 2.5. And from there on, we presumably get swept up in the totalizing drift, and move along to level 3 (and whatever rarefied things follow).
We don't have a great deal of hard facts to go on yet, but from the theorizing above, it seems to me that we should come away with a slightly heightened appreciation of what we have to do with the HDM project. We, or whomever pulls off the totalizing project in mathematics (for, perhaps there are others working on HDM-like things; certainly, if we admit the distributed operation of the various MKM projects, there are plenty of people working on HDM-related things even if they aren't aware of the HDM and aren't thinking about HDM-like goals), will have subsumed the rest of mathematics and mathematical AI. We were already essentially aware of this at the inception of the project. What was less clear then was the fact that at that point, we will be poised to fulfil a broader totalizing destiny.
This essay is written after some reflections on Turing's Universal Machine, one of the prototypes of the totalizing programme. If this essay ever becomes worth dedicating (i.e., instead of confounding), I'd like to post it to his memory. I think that more work and a better treatment of the history and future of totalizing projects would be in order before we get to that point.
To the faithful reader who has accompanied me on this chaotic journey through time, space, and identity: prepare yourself to embark along any and all of several new dimensions that lie before us.
For the totalizing project must be inclusive; holistic. The needs and wishes of the multitude must be considered, and perhaps (as part of the project), re-engineered.
Consider by way of example, the market. Typically, anyone can buy and sell in the market – reserving the right to refuse business to anyone would be a stupid financial mistake – but, just as typically, everyone must conduct their business with a registered agent of the market, and that is the key point. I can not go and buy stock on Wall Street and then turn around and sell it to my neighbor. No. I must sell it back to an agent of the market. (Now, agents may of course wish to make sure you are solvent before they will sell to you.)
Stocks are, in this sense, very different from, say, hammers.
In particular, if we have Market A and Market B, by default there is no "totalizing condition" that says that persons trading in Market A can automatically trade in Market B at trivial cost. It is not simply a matter of making a representation of Market B within Market A; at any rate, the market prices in B aren't the meaningful prices in A, for even registered agents of A must pay a fee to trade with agents of B.
Only if anyone is free to act as an agent in a given market (or, which is the same thing, agency is associated with no specific cost) do we have a totalizing condition.
In the "real world", this condition obtains: I am free to sell a hammer to my neighbor for $5 if he and I agree to the deal.
Costs may apply of trading between any two commonses that don't match an appropriate totalizing condition. For example, two license colleges need not be compatible at all; and individuals "on the ground" attempting to move content between the two colleges will have to try their best to jump through series's of hoops (any of which may or may not be aflame and/or perched above the abyss).
Notice that the "free market" scenario is nevertheless rather different from the "free content" scenario; the mechanism whereby the control valve is constructed and maintained differs. In the market case, the list of agents is controlled; these agents, however, can both buy and sell stocks with anyone. In the license college, anyone is free to enter (contribute content), but no one can leave (take licensed material and transfer it to another license college). Easily in, not easily out, says the Lobster in the lobster pot.
Confronted with challenging situations like these, totalizing projects have two options. Either to destroy the more difficult traps (markets, license colleges), or render them irrelevant. We have, historically, considered the latter method.
How does Captain Kirk get out of trouble? Almost invariably (skipping by a few details), by being teleported back aboard the Enterprise. I can't remember whether the consensus is that he is formed anew out of basic particles upon arrival or not, but it doesn't really matter for my argument.
Approximately the same method is used by PlanetMath and the HDM project. When we find some content we're interested in, we "beam it" from its original source to its new, freer home, by rewriting it. The relevant semantics are, of course, unchanged.
This precise approach is not always possible. Stocks traded on the free market correspond, in a suitably general sense, to things out in the "real world". There is no way to semantically replace these -- but one can build near-equivalents. For example, if we wanted to depreciate Coca Cola in favor of, say, Crystal Pepsi, we would have to manufacture a situation in which the demand for Coca Cola was rerouted entirely in the general direction of Crystal Pepsi. The specifics of implementing the switch could of course be somewhat challenging; the point is, they involve a switch at the level of semantics. Which is to say, the level of belief, desire, and in some cases, even more material facts.
As I mentioned in the first section, total control is likely to remain impossible (indeed, it is only likely to be meaningful as an abstraction). And, of course, the totalizing project will have to balance the spheres of control of individuals operating within the aegis of the project. One model is to give everyone equal access to control over everything within the project (thus each individual's scope is total); but in reality, we would assume that the economics of the situation drive some degree of specialization among the project's agents, just as we're expecting to see differentiation between the elements of the project, although not necessarily to exactly the same degree.
However, it seems reasonable to expect that the individual agents who are working on the project will come to depend on it to an increasingly high degree, indeed, to exactly the degree at which the project succeeds in being "total".
What precisely this means about the livelihood, psychology, and indeed, the lives, of the agents who make up the project is not completely clear. Since the scope of the project becomes increasingly "total", perhaps it will develop mechanisms to substitute for individual will (if indeed such a thing ever existed; I actually suspect this is a meaningless question). What is clear is that insofar as the project is "total", agents will be tied inextricably to it; and its aegis will expand in every dimension. It may very well "take over the lives" of the participating agents; which is perhaps to say no more than that it will become a way of life. (It may be that "agency" as we know it will disappear in a field of ubiquitous subjectivity and fully exchangable viewpoints; these seem like very totalizing things to do.)
For example, insofar as its agents have reproductive needs, these needs will be met within the project; but even more importantly (I hope that this development has been obvious), the agents will be serving the reproductive needs of the project. The project will in any justifiable sense of the word, become alive. And insofar as it really is "total", it will subsume all life that it comes in contact with (along with everything else).
The project will be the ultimate foreign invader; perhaps working its magic invisibly, but (as is of course its nature) fully and completely. Insofar as it is total there will be no escaping its grasp.
Parents, its 10 'o' clock. Do you know where your culture is?
Well, as I mentioned, there are limits; light cones, for example, that we presumably won't be able to see beyond (by definition).
I think that insofar as we do have "will", it will guide us. If the will is to build totalizing projects, that is what we should do; after all, we would only build them (if we are rational) if we think they are to our benefit.
Of course, we might not be rational. We might be quite wrong in our thinking. But at this point in history, one can't build a totalizing project in an afternoon; there still remains plenty of time for checks and balances and evolution along the way.