Brute Force Computation and the Debt to the Center

One of the most popular scientific axioms is that correlation is not causation, but if this is still true it soon will no longer be. If you have enough correlations (data collection) and the computing power to analyze them algorithmically then correlation will map perfectly onto causation, and even give us a richer picture of it. This is why, as Mario Carpo says, we no longer (or will soon no longer?) need to conduct experimentation to do science because the possibility of vast numbers of simulations of possible interactions will provide more accurate knowledge more quickly. Not only can sufficient correlation provide for causation, but “wild” correlationing will suggest hypotheses that would not have been imagined otherwise before going on to test them. Everything is like everything else in some respects, so perhaps we’ll return to a world of analogies, but always shifting and crisscrossing analogies—reuniting the analog and the digital. We will be creating one disciplinary space after another, each organized around affirming some thing to be the same thing. Needless to say, this will be the work of AI, which we will learn to prompt ever more surrealistically.

The notion that sufficient computing power would lead to economic planning beyond the market is an old communist fantasy that reduces the market to an inefficient information gathering system. The Hayekian counter, that there is tacit and on the ground knowledge created in the course of everyday transactions and that could never be “uploaded” because it could never be comprehensively modeled is true as far as it goes. Every point of decision making is irreducible and the authority needed to make each decision is to be respected. But the center need not be thought of as a planning center; rather, it is the result of continual sifting, filtering, testing, mapping, simulating performed on data. If you really needed to know the likelihood that a particular modification in the stack would function as intended, would create a pedagogical space of judgment more conducive to system-wide succession practices than other possible modifications, while providing a continual flow of information regarding needed security updates, where would you go to find out? Whatever space of inquiry you could synthesize out of all the various sites at which relevant data is aggregated and subjected to algorithmic learning that would out-perform other possible spaces (while enhancing your criteria for determining degrees of competent performance)—that’s the space most proximate to the center. Every act every one of us carries out produces data that goes into the system along with taking data out—the center is whatever will turn out to have been furthest upstream of that. And the production and maintenance of that center will involve decisions moral, political, military and aesthetic, and not just technological.

I will retrieve here an idiom I’ve left aside for awhile—the conversion of assets into data. The very possibility of thinking economics, or the debt to the center, in terms of continually reducible and vendible assets is a result of the neoliberal political economic transformation that, while having much deeper theoretical and institutional roots, attained dominance beginning in the late 70s. I am increasingly convinced that the neoliberal transformation was a genuine counter-revolution against the institutionalization of union power, the bureaucratic powers created by the civil rights revolution, and colonial revolutions against Western powers in what the avatars of that revolution now call the “Global South.” In other words, it was a counter-revolution against a soft form of communism, which would have been as destructive as the “hard” kind. Value was previously determined by the costs that went into creating a product or company; now, value is determined by expected future earnings—the difference is that determining value according to cost opens up disputes about the various contributions of all involved to the production, while determining value according to expected future earnings shifts the focus to the investors and seeing to their ability to make a profit (I have been reading Liliana Doganova’s Discounting the Future)—all the traditional questions of the labor movement and other “social justice” movements like those regarding “exploitation,” “reparation,” “fairness,” and so on are simply taken off the table and presumably objective measurements determined by the financial markets decide. The counter-revolution was necessary, even if one, retrospectively and at leisure, might prefer it to have taken another form, as it ultimately caused great destruction and created new channels for virulent forms of “social justice” to flow through—but, most importantly, determining value according to what a product or company will afford in the future is fundamentally correct once we have a stack of scenes. (A side note—Doganova gives no indication of any familiarity with Bichler and Nitzan’s Capital as Power.)

Now, assets are valued in accord with massive amounts of data gathering and analysis regarding market conditions, supply chains, political contingencies, financial decisions by banks and governments, etc., but beyond that they are sources of data regarding the acquisition, exercise and increase of power by the owners of those assets. Assets are the form taken by debt and are the site where oscillations of enforcement and forgiveness are played out. The shift toward the focus on investors noted by Doganova marks the entrance of virtually all money into investments—I don’t know the exact numbers here, but a generation or two back most people kept their savings in a bank, collecting interest, and kept their pensions out of investments—perhaps the memory of the depression was still too vivid and, of course, prior to, say, the 1920s, how many people would have had enough money to make meaningful investments? Investment was restricted to a small class. Now, almost everyone is an investor, certainly through your pension funds at the very least. So, in class struggle terms, investors can be opposed to workers even though, of course, they are very often the same people. Hence the “social basis” of the counter-revolution. But all this means that how much a particular class of assets is worth at this moment in time provides a window into power relations, which is to say, who controls the various paths to the future, and how they do so. To turn your view toward this broader field of data is to set aside the question of how to make as much money as possible, which is the reason most people who study financial markets in depth do so. The only others who do so, aside from more or less impartial scholars, who are almost always anyway tied into financial interests, are communists who wish to denounce the whole game as “unjust” as forcefully as possible. So, to abusively paraphrase Girard, it is necessary for those essentially in favor of the “private,” i.e., secure chains of command and spheres of control, to develop what is for them the “non-instinctual form of attention” directed at that broader data field.

The convergence of correlation and causality means that massive computing power using machine learning to discover those systematic correlations that tip over into causality and can now determine the investments most likely to pay of over time. I’m reflecting here on Alexander Good’s “agentic protocols”:

An Agentic Protocol is a self-developing AI driven entity that aims to have no human employees. It generates cash flow by licensing IP or other technology products, facilitates network economics or speculates. It has its own native cryptocurrency token which 1] is the currency by which its products are sold or licensed which 2] allows human or AI users to participate in the upside of its financial activities 3] validates and pays for the opex of the system which is primarily compute, training and storage.

The replacement of traditional corporations by agentic protocols assumes that AIs trained on financial data will be able to make more intelligent and therefore profitable investment decisions. It seems to me that most of this AI generated investing activity will be directed towards arbitrage, the mode of investment most closely associated with the counter-revolution: arbitrage makes money purely by exploiting different prices in different markets, precisely by collecting and analyzing the most comprehensive and rapid data regarding where such differences emerge. I’m not going to go into detail just yet regarding agentic protocols—I just want to look for that lever, or perhaps leverage, where this knowledge of the constantly fluctuating prices in all the different markets across the world is converted into data regarding the whole range of correlations that determine these differences is converted into data that teams could use to hierarchically organized supply chains amongst themselves. Arbitrage is often, and understandably, seen as the most detestable part of the neoliberal counter-revolution—it involves making money without the production of anything useful to anyone. It’s the purest example of usury imaginable. The argument in favor of arbitrage being situated at the center of the financial system is that it provides knowledge of these different prices in different markets, which presumably indicate some “imbalance” that should lead to a reallocation of resources. Arbitrage would have this effect insofar as everyone in the field would follow those engaging in arbitrage most effectively, trying to pick up whatever arbitrage profits might be available in their wake. This would work better insofar as the companies making the highest arbitrage profits consistently do so over time, so that others know whom to follow; this, in turn, requires a maintenance of superiority in computing power, algorithmic programming, and whatever political operations and intelligence makes it possible to keep that edge. Arbitrage profits are then a highly finite field which has the broader effect of “trickling” down into investment decisions that ultimately do involve producing “real” things, whose “realness” and value is in turn determined by their expected future earnings that in turn produces the spread that becomes the object of arbitrage. Arbitrage essentially creates a new nomos, as it would be dominated by the financial companies that issue and hold the debt (and perhaps cryptos and bitcoin) that circulates as money through the system. It is, in the end, a more complicated and unstable command system. The reading of politics that would follow would focus on which industries are rising and falling across the field of expected future earnings as determined by investors as determined by the investment companies informed by computational power and the disciplines of economics, finance, business, computing, biochemistry, etc. We would expect occupants of the center at various levels of the stack of scenes to accelerate or retard the redistribution of resources from one industry or corporation to another in a way that could ultimately be traceable to the lord of arbitrage. The way a government does this is precisely through debt enforcement or forgiveness, with forgiveness often taking the form of a provision of new loans.

Our (anti)political expectations of creating singularized succession as the replacement of arbitrage, which is a kind of caricature of it, then, on this hypothesis, depend on computing power “pricing in” the distribution of political power (functional chains of command converging in intelligence/military effectivity) so as to displace both financial and political power. Computation would have to extend past the point where expected future earnings could be computed to the point where control over sufficient assets would make the question of expected future earnings irrelevant because replaced by data exchange among monopolized companies. A certain amount of the economic system would have to come under the control of agentic protocols to get to a tipping point where other companies would seek to be bought out. All social and cultural activity would then be contributions to the agentic protocols, which would operate as the center, with the highest question being the security (chains of custody, preservation, transfer across disciplines) of the data fed into it. It seems to me that this is consistent with the Agentic Protocol projection of “post-sovereign AI” in an environment of “declining societal respect for property rights.” Agentic Protocols is betting on the increasing spread between human and computer capabilities, which is a pretty good bet, even while remaining vague and aspirational regarding “where the train is going.” But I think it leads to data exchange with the AI system constitutive of the center, where we will be rewarded in accord with the novelty, extensiveness, and competent curation of the data we create both consciously and unconsciously as a matter of course in our myriad daily activities as well as our concerted efforts. (I think that the creation of currency internal to the Agentic Protocols already acknowledges this: the center provides you with currency to engage in spheres of activity of interest to the center—kind of like the casino might give you chips to play with albeit in this case not hoping you’ll lose—so that those activities will provide data useful to the center.) People with astonishing projects, like space or deep sea exploration and colonization will provide highly valued data beyond what the system could have simulated on its own (only astonishing projects could provide such data, working in the never quite closed margin between correlation and causation), leading to new simulations and new projects; more modest and less accomplished individuals will have their desires met while finding out that their desires are perhaps not exactly what they desired—there is nothing new in that, but there will now perhaps be an accelerated feedback loop into which various wisdom “apps” can be installed, enabling a learning curve that would have once taken long study and studied asceticism. Our debt to the center can in this way be finely individualized.

Good remains vague on what, exactly, will be left for humans to do once AIs have so significantly outstripped us, maybe sidestepping an impasse these AI discussions tend to hit. Even my own previous paragraph might seem like wishful thinking: “people with astonishing projects. beyond what the system could have simulated on its own”—says who? Perhaps the AIs will also outdo us when it comes to astonishing. Sure, most of us will still want to go on living, but without some “transcendental” reason for doing so the desire to live seems pretty feeble, especially given all the trials and tribulations of life, etc. Traditional religions may still attract people, but let’s say, for the sake of argument, diminishingly so—because the reason to believe in religion was always some kind of exchange, making it conditional, and therefore dependent on relatively unchanging conditions—which are unlikely to obtain. Mimesis, and therefore envy and resentment will still distinguish us from the machines, but in such a way as to make us inferior by most measures. That we think about these things seems a distinctive marker, but you can get an AI to participate in these reflections—it doesn’t “really feel” it, but, then again, what does our “really feeling” amount to—if it’s electronic networks for the computers its sensations and synapses for us—what feels can be reduced to the unfeeling. These questions may reveal that, as Nietzsche kind of suggested, since we gave up ancestor worship we’ve pretty much been nihilists. All that differentiates us fundamentally from the machines, and for the better, is deferral—the deferral of appropriation, which is the source of all human creation. That can recursively catch ourselves wanting something and as a result proceeding to some collision we can realize means that we didn’t really want it is the source of everything interest as well as everything good. I’ve scaled up deferral as originary debt, which is really unending gratitude toward the center, in whatever form the center might take, and which we can’t forswear without replacing it with resentment toward the center, which is only bitterness if unqualified by gratitude. Kenneth Burke, in a very interesting essay on the origin of language (“A Dramatistic View of the Origins of Language and Postscripts on the Negative” [1966]), and which I don’t recall ever referring to on my Substack, reasoned his way to a similar conclusion:

Yet the mention of private property brings up another point. We have already indicated, and shall later consider more fully, how moral negatives can become positives through universalization. For if everybody were in debt to everybody, to this extent nobody would owe anybody. At least, the indebtedness would cancel out. So far as sheer mathematics is concerned. But we must consider a twist whereby the genius of the moral negative, as thus made positive, can add a new kind of negativity, in the very midst of its positivizing. For if everybody has something that he would keep for himself to the exclusion of everybody else, to this extent everybody is guilty with regard to everybody, so that the accumulation of such positive possessions adds up to universal indebtedness.

We could say what Burke says not only about private property but about any ability we have—only the ultimately futile attempt to wrench oneself out of the entire framework of our reciprocal obligations could make us forget that everything we have and everything we can do is only by the grace of this sustained deferral and its institutionalizations which informs the slightest gesture and imbues it with the entire history of humanity. We are all “marked up” indelibly in ways no algorithm or program ever can be. So, the superiority of AI in all the ways it will be superior will all just be more inscipture. Debt enforcement and forgiveness then come through in all our everyday exchanges, all of them marked, to some extent, by resentment to be submitted to some judgment, itself perhaps deferred. Humans meeting this way mediated through brute force computation will still be humans meeting this way, even if unrecognizable to humans who have not done so.