The Tar Pit

The name of this article is indeed inspired by geostationary satellites, whose design principles you will need to understand in order to make sense of what I'm talking about. In order to (at the very least superficially) understand the design principles behind geostationary satellites, you will need to understand closed-loop feedback systems; and in order to understand closed-loop feedback systems, you need to understand the basics of systems, i.e. control theory. Did I not say this before? learn your control systems, already, kids!

Having said that, let us unwind the call stack bit by bit -- a control system is anything whatsoever that may be modelled by this black box:

  Input   +----------+  Output
--------->|  System  |--------->
          +----------+

Judging by this general picture, the keen reader will no doubt intuit that there are quite a few things that can be modelled as, more precisely: some place where one puts something, such that one may take something else out. For example, your day job is that place where you put your time and (physical, intellectual) resources, getting in turn money. Another example would be any engine which turns, say, electricity -- because y'all like electric cars and whatnot -- into mechanical work; while the opposite is also possible: for example a hydro plant takes the mechanical work of a water's flow and converts it into electricity. Another example would be the state, which takes your money and gives you promises, but in any case, I suppose that's enough exemplifyin' for now.

A particular class of control systems -- "control" because you supposedly control the inputs, obtaining the outputs, sorta like when you eat or exercise; you do exercise, don't you? -- are called "closed-loop feedback" systems, because they look sorta like this:

  Input       +----------+      Output
---------x--->|  System  |---x---------->
         ^    +----------+   |
         |                   |
         +----- Feedback ----+

This diagram has a few items added over its predecessor: two "x"es, one on the output arrow and one on the input arrow; and the two are interconnected via another arrow labeled "feedback". In other words, the output is now "fed back" into the input somehow. This "somehow" is not to be simply waved away, since there are in principle precisely two ways of feeding back outputs back into the inputs: one is additive, that is, positive feedback, while the other subtractive, that is, negative feedback.

For example, you get the first type of feedback, the positive one, whenever you place a microphone in front of a speaker which takes its audio input from the very same microphone. And if the audio feedback sounds bad, then the feedback loops of the so-called democracy, or otherwise the network effects where everyone and their dog move to a town near you, driving electric scooters and shitting where they eat, those are devastating to civilization itself -- but I digress.

The more interesting (for reasons which will become obvious in a moment) type of feedback, the negative one, is the one where your teacher slaps your fingers with a ruler so as to help you fix your sloppy writing. Otherwise, examples of negative feedback include your car's ABS, or indeed, the geostationary satellite referenced above, or any sort of linear adaptive system. So how does a geostationary system work, then?

Say you place an object in orbit, meaning that it's at a distance where it lies in imponderability, that is, without being pulled by Earth's gravity. But the Earth rotates, and thus said object may at some point hover over Romania, while at another point it may lie over China and so on and so forth. However, the whole point of stable satellites is for them to be accessible from a fixed point on the surface, so while in orbit, any geostationary satellite must contain some mechanism for automated error correction.

Applying this to the model above, the input will be the angular velocity of the satellite, while the output will be the object's motion. Thus, the feedback will make use of the output in order to adaptively compensate the angular velocity, in order for the resulting velocity to be zero. This will keep the satellite at a fixed position with respect to the surface of the Earth, and more importantly, at the position desired by whomever launched the thing in orbit.

Now that all this is clear, we can proceed to examine how The Geostationary Truth Machine works, by analogy -- the problem with Dijkstra's naïvete is that he doesn't specify precisely "by analogy" to what, and the what is as important as the analogy. Anyway, the system under scrutiny is an extension of The Profiling Machine. The input of this new system consists of particular states of DB, in particular of P, as per the spec. The desire is to drive O, and in particular the query results of agents, to certain (cor)relations, the outputs being... whatever, but more importantly, an atomization, which in practice means either as many clusters as there are individuals, or otherwise a single cluster.

As for the so-called "truthness" generated by the Machine, it really doesn't make any difference whether it has anything to do with reality or otherwise. Regardless, I am quite convinced that this works if not precisely as written here, then quite closely. I did say, didn't I? Maybe now we understand each other that I'm not doing rhetorics for the sake of bullshit.

Filed under: computing.
RSS 2.0 feed. Comment. Send trackback.