Vico, Descartes, and decay of knowledge in software.

Cartesian project aimed at providing reliable foundations of sciences – that is, metaphysics. Sciences require metaphysics to define their subjects, so metaphysics must be rock solid.

In Meditations on First Philosophy, Descartes builds metaphysics as follows:

1. He starts from a skeptical stance of radical doubt, which treats everything uncertain as if it were false until proven otherwise.
2. Descartes can doubt everything, but he can not doubt that he exists. The very act of doubting is a sign of something happening, so something exists. Descartes concludes that "the thinker" exists³. This is the cogito argument.
3. Then, through a rather convoluted argument, at times reminiscent of Anselm's ontological proof of God's existence, Descartes proves that God exists and is benevolent.
4. God gave us cognitive faculties, and in virtue of his benevolence he does not want us to be deceived. Therefore, as long as we use our minds correctly, they will arrive at the correct conclusions – God makes good stuff.
5. How to use our minds correctly? We have to perceive our ideas clearly and distinctly. For Descartes, this is criterion of truth.

So, metaphysics rests on cogito, God as a guarantee of adequacy for our mental machinery, and reconstruction of the knowledge suspended during the radical doubt phase.

The clearest ideas are, of course, mathematical. Cartesian ego, detached from everything of the material world, can prove theorems just fine.

Unlike abstract mathematics, science has to deal with the material world, which is what exactly? In the famous wax argument from the second meditation, Descartes strips away all noisy signals from the senses – color, warmth, smell – until only one property remained: extension.

So, fundamentally, matter is simply stuff that takes up space, res extensa. That idea seemed clear and distinct enough, so it had to be true. It follows that the properties of bodies are geometric-ish, and science can be done with the precision of mathematics.

I am gladly engaging in Cartesian-flavored fantasies about software, imagining it as something with clean abstractions, implementing coherent formal models, with total specifications, no ad-hoc mess. After all, software is a product of the mind, substrate-independent, like mathematics. Such dreams are most pronounced in the mathematician-adjacent tribe who praises formal verification, abstract programming language theory, elaborated static typing to enforce correctness universally and ahead of execution. But real systems are large and complex, historical, contingent, patched, negotiated, and path-dependent, defying clean categorization and modeling.

Certainty is a key concept for Descartes. He sees ideal knowledge as universal, eternal, objective. Concrete, situated knowledge is unstable and often subjective. In opposition to it, Descartes strips away everything subjective by refining an imaginary ego disconnected from anything physical, a pure mind capable of clear and distinct thought.

The resulting framework aims to be devoid of anything opinionated. From one viewpoint, it is a victory: normally, humans disagree because they perceive the world differently, so a Cartesian framework allows people of different backgrounds and persuasions to discuss questions and converge on answers. This comes with a price: a Cartesian ego, cut from material world, does not have a clear path to know anything beyond itself.

Cartesian framework throws a situated, concrete knowledge into exile. The ten-argument function violates a general rule in any interface design textbook. History may reveal that it preserved ABI compatibility, avoided allocations, worked around a compiler bug, matched generated code, or protected a hot path. Can any clear, distinct theory account for all such situations? A critique ignorant of the context may be clear, distinct, and wrong.

The Cartesian journey starts with a radical doubt which arrives to cogito. The purpose of cogito is to provide a skeptic-proof foundation for metaphysics, and then, transitively, sciences.

First, Vico believes in two distinct modes of cognition: consciousness (coscienza), grounded in basic awareness of having a feeling, and causal knowledge (scienza). They map to conscience and science, to certainty and truth.

Anyone can read a class diagram and become aware of what are the classes; one can even read the code after that. But why did we chose to structure the class hierarchy this way? That question is harder.

Cogito further exemplifies this distinction and reveals a mistake of Descartes. We are aware of our existence, but we do not know it. We are certain that we exist but still oblivious to how we came to be, or why we are, or how thinking operates. We do not know our causes. Cogito is a sign of our existence, but not its cause.

Second, Vico remarks that when skeptics have a toothache, they do not doubt what they feel – they doubt the possible causes of the toothache. In contrast to what a reader of Descartes might expect, skeptics do not actually doubt any of the following:

• That they think.
• That they exist.
• That they perceive appearances.
• That they may cognize mathematics.

Skeptics suspend their judgement of causes. So skeptical attacks on cogito would attack the causes of our thinking, not the fact that thinking itself happens. Cogito says nothing about the causes, so it misplaces the line of defense.

Two things follow. First, cogito does not help building a skeptic-proof foundation for the natural sciences. Second, because 1) science studies causes and effects, and 2) cogito says nothing about our causes, we conclude that cogito is neither a model, nor a standard, of scientific knowledge.

The distinction between truth and certainty, as Vico understands them, applies to the mind/body separation. It is wrong to assume that thinking is the cause of our existence. We are embodied minds, therefore we think, not the other way around. Otherwise, thinking would have had to be the cause of a body too. Thinking is a sign of having a mind, giving us certainty that we have one, but not automatically enabling the knowledge of the mind.

So, the true is the object of science, the certain belongs to conscience and to particular human facts. The cogito gives certainty, therefore belongs to conscience. It reassures me that I think and exist, but does not reveal how my thinking came to be, what causes it, or what I am.

In software, reading code gives consciousness of behavior. Recovering rationale gives causal knowledge of design. I may know that the code takes this branch in a given situation, but why does the branch exist? I know what a test checks, but perhaps it tries to prevent a specific behavior somewhere else? I know how a type may enforce a constraint, but which tradeoff produced the constraint? I know what a feature commit changes in the code, but which alternatives were rejected?

Another line of attack is an overemphasis on the critical analysis which is only useful when our imagination sometimes feeds it with meaningful ideas.

Vico starts On the study methods of our time by endorsing Cicero's "Ars Topica", a book exploring how to be a good orator, and an adaptation of Aristotle’s topical method, but with a more practical, rhetorical, and juridical emphasis. Vico himself was a professor of rhetoric, hence the choice, I guess.

Cicero writes: "All methodical treatment of rational discourse involves two skills, invention and judgement". In today's terms, designing a speech involves brainstorming and critical selection. This is a common design strategy: divergent thinking followed by convergent thinking. Think about how diverse were early cars or mobile phones, and how similar they all look now: selection favored optimal shape, and now the diversity manifests elsewhere.

Then Cicero focuses on a method of invention: how to systematically explore the design space of possible arguments for a speech by using some anchor concept. For example, analyze the precise definition of this concept; decompose the concept; try to negate it or compare it with similar concepts, and so on. The gain is in systematic exploration which helps to produce better arguments and to unblock the author. The orator then critically selects the most coherent and suitable arguments for the speech.

Today, the closest exploratory approaches in engineering that I know of are Morphological analysis, TRIZ and Systematic inventive thinking. Neither are a common knowledge, I daresay.

Cicero shows a balanced approach. Contrary to him, the contemporaries of Vico overemphasize critical skills over invention, throwing students out of equilibrium. Descartes made a lasting impression on scholars by using radical doubt to reach the foundational principle of metaphysics (Cogito). For Vico, youth is the time when imagination and creativity blossom, and universities are neglecting their development in favor of analysis. Then, students are forced to concentrate on abstract logic⁴ to become good in critical selection, but they can't generate any original ideas to select from. Deductive reasoning becomes sterile, leads to abstract intellectualism and the overall progress in society stalls.

Additionally, Vico pushes an argument familiar to us, about the limits of control over our powerful tools. Modern science provides us with technology, but we need practical wisdom to use it for good. To Vico, the practical wisdom comes with the study of history, philosophy, poetry, law. A universal, refined, human-less rationality needs phronesis⁵.

I wonder, is it a danger for a new generation of engineers, who will attempt to create systems mostly by reviewing AI generated code? How will they be able to critique specific contextual decisions without sufficiently engaging in the generative practices in their careers? Will they be generally less capable to come up with novel solutions, architectures, technologies?

Another vulnerability originates in, perhaps, the most powerful part of Cartesian project: the Cartesian method aims at being objective, context-independent, universally applicable. But what about sense-data, phenomenal experience, or anything related to arts, history, human sciences in general? Cartesian method recognizes the first, but turns anything grounded in subjective experience into some kind of illusion.

Not only human sciences are left out without a solid foundation. Cartesian method is too rigid when it comes to stochastic arts ⁶ like medicine, where no amount of mastery guarantees the result. Everything a doctor does is, in a way, unreliable, doubtful, and can't be grounded in the Cartesian metaphysics. We can't build a unified, strict theory of healing people – it has to be constantly revised according to the experience. As Maurizio Esposito puts it:

The practice of medicine, as well as the art of navigation, requires either a flexible episteme or a systematic technē. Stochastic arts require a fine-tuned equilibrium between experience and theory. They imply some sort of practical wisdom.

Descartes and his followers treated nature as more certain than human culture. Vico turns it around, pointing that subjective has real power. External world always remains external to the human mind, only partially graspable. Humanly known truth is fragmented, for we know it by division, says Vico. But we make our own decisions, generating causes that are hard to doubt. If I cut a toy horse out of a wooden plank, I certainly generated the cause of that horse. And that's the key to verum factum – there is actually more certainty in our decisions than in anything we know about the world. Descartes, and sciences, deal in causality, which might be itself just a construct of human mind, posing for some objective aspect of reality. But our decisions happen in our mind and therefore are immediately accessible to it.

Software engineering is also a practical, stochastic art. Most of us are building systems in an open world and interfacing with other systems of which we have limited control. Developers may execute perfectly but the context shifts, some library becomes obsolete and abandoned, the requirements change so much you can't accommodate for them without rewriting everything from scratch. The user updates their OS, the world moves on and does not support the old version of the protocol you are using anymore. Try browsing modern internet from some old PDA which does not support HTTPS properly. No amount of expertise can zero the infant mortality rate in software.

And, most importantly, Cartesian quest became possible because Descartes came to be into an already existing culture. That culture was a necessary precondition to have any kind of discussion about Cogito and metaphysics in the first place. So which is primary: situated, practical knowledge, or abstract, universal one? In a closed loop, picking either side is a losing strategy.

Out of curiosity, I was looking for the first occurrence of Verum factum in Vico's works. I think it appeared in the first chapter of On the most ancient wisdom of the Italians. It begins like this:

For the Latins, verum (the true) and factum (what is made) are interchangeable, or to use the customary language of the Schools, they are convertible. For them, the verb intelligere is the same as "to read perfectly" and "to have plain knowledge." In addition, their cogitare was the same as our vernacular "to think" (pensare) and "to gather" (andar raccogliendo). And for them, ratio meant the reckoning of arithmetical ratios as well as man’s endowment by which he differs from and surpasses brute animals. Also, they commonly described man as an animate being, rationis particepem, that is, as partaking of reason, but not always having full possession of it. Moreover, as words are symbols and signs of symbols, so ideas are symbols and signs of things.

Hence, it is reasonable to assume that the ancient sages of Italy entertained the following beliefs about the true. "The true is precisely what is made" (Verum esse ipsum factum).

Comes as a wrecking ball with some rudimentary philological analysis and loosely related pieces of trivia. But I think this mess does not matter. I see it as an application of Ars topica that Vico likes so much. No matter what inspired an idea like Verum factum, even if Vico had misheard the lyrics of a song, the principle is then scrutinized critically and justified in later works.

The skeptical attacks on causes are themselves suggesting a line of defense. What are the causes that I would doubt the least? I generate the causes myself, I produce stuff, abstract or concrete. It is a practical, concrete certainty, my own awareness of my actions. And if others made something? I can recreate that, or at least study how I could, potentially, do it. I will not properly know what is a specific algorithm, an operating system, browser, game engine, unless I figure out how to implement them. Philosophers would say, that Vico establishes the certainty of technê, not of episteme.

Cartesian metaphysics implies that mathematics provides the most robust kind of human knowledge, because mathematical ideas are the most clear and distinct and can't be corrupted by our imperfect perception of res extensa. Then goes physics, a tier less robust. What will change if we part from Verum factum instead?

First, Vico underlines the distinction:

1. Science is based on causality is concerned with proper demonstrative knowledge, and we have only such knowledge of the objects we construct ourselves. It is a genetic knowledge: to know a thing fully is to know its causes, to possess the inner blueprint of its making, to be able to reconstruct it from its first principles.
2. Conscience is based on awareness. We explore the world this way – imperfectly, from the outside, through observation and classification.

Vico agrees with Descartes that mathematics (including geometry) is the apex of certainty, but for different reasons. All of mathematics is made: we define points, we generate lines, complex figures, statements about them, proofs. The human mind arranges elements of such knowledge, therefore causing it. Mathematical truths is what's made out of basic mathematical elements such as axioms, and the basic elements are themselves man-made. We do not need God to validate mathematics for us.

Like Descartes, Vico places physics lower than mathematics because it involves reasoning about the real world. We did not make the world, God did. Therefore we do not have the same grasp of it. To study the world, Descartes proposes to try internalizing it through our apprehension of clear and distinct ideas. Vico, on the other hand, points that we create a science whose elements are transparent to us. We experiment and in so doing we recreate the structures of nature; being more successful in experimental physics means imitating God better. For Descartes, experiments are checks of validity of our reasoning.

The further we go from mathematics, the less certain our knowledge becomes, and the moral sciences are at the opposite side of the spectrum. Descartes and Vico agree on that. However, where Descartes sees uncertainty in opposition to objective knowledge, Vico offers a redeeming foundation: even in stochastic arts, even in arts, history, moral sciences, we know what we made. Man-made things and man-guided processes are fully intelligible only to their maker or doer, or to others who have either made a similar thing or would be able to do it.

In The New Science Vico goes as far as to claim that the natural "science" is impossible. Science is more of a self-knowing process, where the knower explores what they have done, as both object and subject. Physics, from this angle, is a study of models that we create in experimental interaction with the world. So "science" should only be applied to the human world: history, for instance. Physics is then a systematic form of "conscience", an "elaborate witnessing of the world" as Donald P. Verene puts it. That's bold.

Cartesian metaphysics depends on benevolent God and the proofs of his existence. Vichean system demands a notion of "knowing with certainty", and assumes God as an omniscient maker.

Two things follow. First, we are also guessing the intention and the structure behind the design of nature. We can only hope to guess right. Second, if a human mind can't understand God, the diving knowledge about his work is not only unreachable but also ungraspable to us. In other words, even if God were to explain his plan to us, we wouldn't understand because we are limited. We can observe nature, measure it, build useful models, but that's it.

The act of making is how we mimic God. To a Renaissance thinker, the purpose of knowledge is to produce practical effects on the world, so its standard can not inhabit the world of pure minds. Knowledge enables us to transform nature, and in doing so to control and to understand it.

It is interesting to contrast Vico with Francis Bacon here, who also underlined the importance of connection between making and knowledge. For Bacon, making is aimed at the extraction of knowledge, but for Vico knowing equals making. For Bacon, the goal of knowledge (and sciences) is to make our lives more comfortable; for Vico, a more important aspect of knowledge is the enabled understanding. Technical knowledge without wisdom that draws boundaries of its meaningful, purposeful, ethical applicability, would be characteristic for the barbarism of reflection. Draw from the wisdom of Ancients, read poetry, don't neglect Ars Topica, listen to uncle Gio.

Software requires recreating world from a limited standpoint. When we automate processes of a shop, we think in ontologies to produce a model of the shop with its customers, employees, logistics and so on. The software itself must model whatever real world processes it automates, by Ashby's law: a controller must match the variety of the system it controls. Perhaps this is a part of the charm of our job: it's so much easier to be a software demiurge. For sure we understand the code that we wrote better than its dependencies. And just as physicists at some point can not penetrate further into the structure of the world, perhaps yet, we also need to accept the foundations – some abstraction layer – as a given. If we need to understand it better than the documentation allows, we experiment, extracting a hopefully sufficient synthetic knowledge of it. Perhaps this is why, in my experience, some physicists make really good programmers if they put themselves to it seriously.

As a software engineer, I found Verum Factum, as it appears in On the Most Ancient Wisdom of the Italians, lacking in two ways.

First, the maker is in a privileged epistemic position while they are making, but we forget things with time. My story precisely, as reading my code from way back is an exercise in remembering and reconstructing the rationale. I will refer to this as an epistemic decay.

Second, most of the time we either produce collectively, or base our effort on the work of previous generations. Sometimes we make without a plan, so the outcome emerges out of our loosely coordinated efforts. Most open source software fits this description, but the large movements of the industry as a whole are also emerging due to a combination of factors. The increase of complexity made hardware vendors specialize on different components such as memory or CPU, the market pressure pushed everyone to optimize and speed up their components independently of others, as the result systems acquired specific bottlenecks and we design software around them.

So, to apply Verum factum to engineering, we need to account for the epistemic decay and collective making. Vico has a lot to say about them in The New Science, as he turns towards humanities, which are collectively and historically constructed. His approach to history is not exactly modern historical science, but I find it insightful about making in general. Humanity as a whole is too messy, and complicated to satisfy Vico's model, but an individual engineered system might just be the right fit to apply his framework.

So let's start with how Vico treats collective making through human history. For Vico, civilizations move through recurring cycles – a divine age of myth and ritual, a heroic age of aristocratic codes, then, a human age of reason and democracy.