Organic Technology, Kludges, and the Inexhaustible Oppression of the Real
A bonus post on medieval aphorisms, tape dispensers, virtual reality, and silly 1980s AI research
If you're not reading Thaddeus and crew at Machinae Ex Deo, like, what are you doing? If you find our stuff interesting, you'll almost certainly find their stuff interesting.
Thaddeus recently had a post titled "Workshop as Ecosystem" that synthesized several illustrative passages together from a range of badass books and ultimately concluded, according to the title, that the workshop is an ecosystem. Three points stand out that I wish to comment on, two as supporting stories and one as a contrast.
Organic Technology
Let me add one more reference to the idea of technology as organic or the workshop as an ecosystem. Consider these passages from St. Thomas Aquinas’s De Regno, Chapter 3, paragraph 19. Seemingly to remind the reader of an oft-quoted and obvious principle, Aquinas mentions that "whatever is in accord with nature is best, for in all things nature does what is best." In the same tone, he says "if artificial things are an imitation of natural things, a work of art is better-according as it attains a closer likeness to what is in nature." This follows pretty straightforwardly from a Christian view of creation, does it not? God creates, we create; God must create perfectly,1 so we will do well to imitate him.
Now, to be clear, this is not simply biomimicry - I think the examples Thaddeus gives tell that well. I cannot articulate in its entirely what nature is in the sense Thomas gives it, but I can offer one illustration: fractality.2 Fractals, shapes where parts are similar to the whole in some way, seem to be a staple of natural beauty, present in a pretty literal form in coastlines, mountaintops, trees, and river runs. What Thaddeus's argument would say would be to expect to find the same principle at work within beautiful artificial things. That certainly is the case - consider Gothic cathedrals or even Aztec stone carvings. They certainly have a fractality, a distribution of beauty across many scales. I think the reason Mandelbrot and friends "discovered" fractals in the 1900s was because we had forgotten that aspect of beauty with the advent of modernist design and were surprised to have found a mathematical formulation of it.
Time and time again, I’m reminded: the medievals probably thought of it first.
Kludges
After giving the passages that illustrate the theme he's been seeing, he describes two contrasts to good design. The first is a kludge.
We should avoid kludges: tacked-on series of technologies that don't have harmony in form, aesthetics, construction, or theory. Kludges have lots of 'loose ends': byproducts that must be dealt with or simply ejected, nonfunctional bolts and ports.
My tweet-response would have been "isn't an old European city simply a set of kludges built on top of each other?"3 If a kludge is truly without any harmony in form, aesthetics, construction, or theory, then yes, kludges are to be avoided - but to meet all those criteria, you have a very small category.
There's a kind of beauty to that would commonly be described as a kludge. If beauty is about proportion and relation, then an important question in judging the beauty in any work is asking what should be in proportion to what. A beautiful kludge, an 'elegant hack', is where the fix is in proper relation to its rather-constraining circumstances. Isn't that the awe of MacGyver - to take unassuming materials in an unassuming circumstance and create something new?4
This is a rather unimpressive example of MacGyvering, but this is the example of a kludge I had ready at hand. My cheap, disposable transparent tape dispenser broke. I could toss it and buy a higher-quality one (maybe worth doing in the future), stop using this kind of tape (good long term goal), or continue to use the dispenser with its sharp edges (unwise). Or, I could… simply tape the dispenser back together. This taped dispenser is ugly when considered by itself - or rather, considered as an object simply to be seen. But it is fitting - beautiful, even - when considered within the greater order of my desk and my life. It is a small thing, so fixing it with more visual appeal would take work and focus that would not be properly proportioned to the problem of its simple visual ugliness. And so the tape[d] dispenser, as kludgy as it looks, has a fitting relationship with the ecosystem it is part of.
This definition of a kludge is also why under different circumstances, like more time or better skill, the kludge is ugly as sin.5 Furthermore, this characterization of a kludge doesn't preclude a certain beauty when the goal is to be intentionally imperfect, i.e., where the fix is not properly related to the circumstances of the fix itself, but instead related to the idea of fixing or making in general. A kludge also gives permission to cast off the modern delineation between maker and user and begins to open a gradation of skills, a gradation obvious to anyone who has begun to make.
Kludges are a sign of poor design- but reconsider whether that thing in front of you is actually a kludge.
The Inexhaustible Oppression of the Real
6Later on, Thaddeus describes an implication of this vision of technology as organic:
It seems to be something like this: do not map a pre-conceived functional model onto things. Yes, man does work, the cow makes milk, and the grass photosynthesizes. It is also true that the cow eats the grass and the man drinks the milk. But these are a mere minutia of all the actions that each does. The grass grows roots, it communicates with microbes, it respirates - to enumerate only a few activities. Cows and men are even more complex yet - but their complexity is of an unknown nature, constantly revealed through work and engagement.
There are two stories I wish to share to illustrate this insight, one that is historical and one that is personal.
First, the personal one. In my day job, I was charged to build a virtual environment for designers to discuss and perform design tasks in so we could study how designers use virtual reality. I learned that creating a world is… not easy. In theory (and in sales), virtual objects are as easy as"drag and drop" - a 3D couch made in one setting can be copied, with minimal marginal cost, to another setting. In practice, objects are never what they seem to be. Even the objects I spent significant time on to get working right - the eraser and whiteboard marker - were shadows of their real selves. You couldn't use the virtual marker to tap, or fidget, or make marker swords, or toss in the air. I was burned time and time again by the temptation to reduce some object (simple things!) to just a few uses, and implementing them. I think this is the practical trouble with the current dream of the metaverse: the real world is just too complicated to repeat, and we take so much of its complexity for granted. This makes our pre-conceived, inflexible mental models the source of many problems.
This personal experience helped me make sense of a historical arc that the artificial intelligence research community had gone through from the 1960s to the 1980s. I recall seeing quotes at a certain place in our computer science building, spoken in the 1980s from a Turing Award winner, that the great challenge for AI was how to express common sense knowledge in logic.7 The desire was for a formal system to capture and express all the common-sense words we encounter every day. I kid you not, the example he goes with is canning - defining a can being sterile in terms of being sealed and not containing live bacteria, and the not-containing-live-bacteria is defined in terms of three smaller atoms, “is it a bacterium?”, “is it alive?”, and “is it in the container?” I get vertigo just thinking about such a system - we have to define life before it can even start canning! To be fair, he uses this example to illustrate that simple propositional logic can’t be used to discover a rationale or a mechanism for canning, which is certainly true. But the future direction he proposes is “to allow the full use of quantifiers [like any, all, some, none] and sets and have strong enough [programmatic] control methods to use them without combinatorial explosion.” The focus he has is not on the non-atomicity of the atoms, but on finding the right structure for the atoms to be arranged.
To put common sense knowledge entirely into logic is the science-scale version of the researcher-scale virtual world. The world is given to us, and so it is not something we can do anything with completely, especially with our simple logical categories. It appears to me that this logic-focused approach to AI is very modernist and Enlightenment-flavored.8 The fact that there are other stories where a lack of an ecological vision of technology (supplanted by an Enlightenment one) has lead to problems indicates that Thaddeus is certainly on to something.
In both the personal and historical cases, the immense depth of the real (as opposed to the symbol-based, or what can be understood and articulated) seems oppressive - it weighs upon the researcher, present at every turn, and stymies every plan laid out. If, as a totalizing view of science would have it, man is to be the master of the world - not simply the lord and steward, but to be the master - then the researcher’s only option is to reject the world, because our world is created by God. It is easy to see the next step in this totalizing view of science being transhumanism: if man is the master of the world, then it is not only permissible but necessary for all natural constraints - including the body and what it represents - to be contingent and manipulable. To admit otherwise is to admit some part of the world that man cannot become master of.
The Christian vision, the vision of technology through the Mind of Christ9, is precisely the one Thaddeus lays out in contrast to the preconceived model. Technologies are a synthesis of the man-made and the God-made. A chair is designed by man out of wood designed by God. The line between the man-made and the God-made is the line of mastery and stewardship. We can construct, deconstruct, reconstruct the physical chair - one person made it, they can re-make it. The idea of a chair is a social thing, so one person cannot re-make it entirely- they can re-interpret it or offer a redesigned chair like we offer a redefined word in a conversation. On the scale of societies, we can construct, deconstruct, and reconstruct the idea of the chair. It's an invented thing, after all. But at any scale, we can't exhaust the knowledge and the possibilities of the wood the chair is made of, because trees have a nature set by God, and trees came before us and came into being without us.
To the modern, this fact is oppressive. To the Christian, this fact is humbling-- and therefore glorifying. We are called to lift up the tree into the worship of God through participation in the life of man.
We are stewards and lords, not masters, and we have the Lord to look up to as our master designer.
The use of the word “perfect” should feel weird given a modern definition of perfection as 100% / A+ / full marks, but that's because we've lost the meaning of what perfection is - consider the original Latin, per-facere i.e. "done through" or "completed". A perfect acorn is an imperfect tree, and that’s not a bad thing.
Fractality has a shot at being a better word to describe a medieval notion of hierarchy than "hierarchy" itself does. Thanks for ruining perfectly good words, Hobbes and friends…
Addison Del Maestro might need to check me on that one.
I hear explosions are also part of the draw, too, but that’s not what this post was about.
If only sin were that ugly…
Yeah, it's a puffy pseudointellectual title, but I was in a puffy, pseudointellectual mood when I first named this feeling a few years ago.
A quick Google seems to indicate that John McCarthy was the quotee, and that his published version of his 1971 Turing Award lecture (Generality in Artificial Intelligence) is likely to be the source I remember. In any case, it captures the idea I am going for.
Consequently, the recent era of data-intensive machine learning is a postmodern turn. On Computers and Cognition by Winograd and Flores is an excellent book if one wants to see explicit evidence of the turn from modernism to postmodernism within computer science.
It was really easy to slide in a reference to Machinae Ex Deo’s slogan - it must be a good slogan.
To your last section - Matthew Crawford in Why We Drive actually makes reference to the idea of "the world is it's own best model" and how this is influencing a new wave of AI research.