Against Moldbug’s Reservationist Epistemology: Reason Alone Is Not Reasonable

Against Moldbug’s Reservationist Epistemology: Reason Alone Is Not Reasonable

For most, Mencius Moldbug will need no introduction; for those who do not know him, this review of his work is succinct (which Moldbug never was) and fair.

We’ll not discuss his political ideas today per se, but instead tackle his thoughts on what he called his reservationist epistemology.

The central dogma of reservationism is that reason is irreducible and untranscendable. Reason is no more and no less than common sense. It is not possible to construct a useful definition of common sense, nor is it possible to construct a system of thought that improves on common sense. Any system that purports to do so is either (a) bogus, or (b) justifiable via common sense, and thus a special case of it.

For example, mathematics is a special case of reason. Mathematics proves theorems by reducing complex formal propositions to a series of obvious steps. Since there is no mathematical definition of obviousness, there is no mathematical definition of proof.

My disagreement begins at the beginning. Reason is not irreducible, nor is any system of epistemology relying solely on reason reasonable (Dear Internet: this last is a joke). Reason is no more than a tool—a necessary tool—to work with ideas. Ideas themselves can and are produced by reason, but not often and usually only in specialized contexts, like mathematics. Ideas—concepts, universals, call them what you will—more often are the product of insight and intuition, which is to say, some form of inspiration, insight, instinct; i.e. a case of induction (see this article for the various types).

Moldbug is right that mathematicians prove theorems, which is to say, that they demonstrate the truth of certain ideas. Reason is used to build the ideas, the end result of theorems, and can even generate them in some contexts. But usually the mathematician’s intuition—or insight, or inductive powers—provides him with the goal. Reason fills in the blanks between ideas known before, and the new inuited idea. Do not forget, what is often forgotten, that the proof supplied by the mathematician does not make the idea true. It is, was, already true. The proof is only a demonstration of the truth.

Plus, no mathematical theorem could ever get off the ground without first assuming truths (ideas) that cannot by definition proved by reason. These are the axioms, which are supplied and massaged and put into readable form in some small part by reason, but again the bulk of the work is done by intuition (induction). Axioms are incapable of proof by strict reason. Yet everybody believes them (or most of them), because, of course, they are obviously true. Faith (in intuition, in the unproved axioms’ truth) is just as necessary as reason.

The limitations of reason are not restricted to mathematics, but apply everywhere. Since reason is just a tool, it has to have material upon which to work. Think of reason as a lathe, a complex apparatus which turns a shaft of wood into a newel post. The newel post, which like all famous artists we can say was always “in” the wood but needing drawing out, is not the lathe’s idea, either. It must be directed. We couldn’t get to the post without the lathe, so neither is reason dispensible.

This goes for morals, politics, logic, everywhere, including, in part, religion. Religion provides a different source for ideas beyond, or rather underpinning, reason, which is revelation or inspiration. There are many ideas which are held (such as the Eucharist) which can only have been proved by divine authority. Reason, as ever, is able to take these inspirations and push them forward, but only incrementally. We cannot do without intuition or inspiration.

Moldbug says that the “The great enemy of the reservationist is the automatist. An automatist is a small, grubby person who believes he can reduce or transcend reason.”

Automatists tend to fall into four camps. The stupidest are literalists, who believe that instead of thinking, we should accept the literal text of some holy book or other. The most dangerous are officialists, who believe that truth is whatever the government says it is. The most annoying are popularists, who believe that the most fashionable thoughts, as of right now, are the most likely to be true. And the most pernicious are algorithmists, who believe they have some universal algorithm which is a drop-in replacement for any and all cogitation.

I’ll man the barricades side by side with Moldbug against officialists, popularists, and algorithmists, though I would fend off the last with compassion, since most of these folks are nothing but overly earnest nerds too in love with their toys. Officialism and popularists need to be bayoneted mercilessly—intellectually speaking, of course (or in hope).

Bayesians, at least many serious ones, are inveterate algorithmists, as Moldbug rightly notes. Too many believe they have discovered in Bayes’s rule a formula for, well, everything. But Bayes’s rule, as I have pointed out before, is not needed. It is only a computational aide. It is the mathematical equivalent of the lathe, accepting numbers as input, massaging them a certain way, and spitting them out. It can even be skipped, bypassed, as we can figure probabilities without it. Bayes needs input, as all probability is conditional, and that input can’t be provided by the rule itself, but must come from outside. (Moldbug, like many, writes the theorem wrong; e.g. writing P(B), when no such object can exist. We can only have P(B|E), where E is the evidence probative of B we assume or believe.)

Moldbug chooses as his representative literalist a dogmatic chemist who believes “the CRC Handbook is the literal word of God.” Such a man does not exist, except in mindless computerized form (as an algorithm), for every human chemist knows how the tables in the Handbook were generated, and knows therefore of the potential for changes and error.

Perhaps Moldbug thought religion too easy a target, or that because of his sympathy for the religious (he admits he’d prefer an orthodox Catholic over a Bayesian as dictator) and desire not to cause them explicit grief, he instead constructed his lignin-cellulous man. Moldbug is forced into condemning literalists by his worship of reason and eschewing inspiration. Yet the ideas produced by inspiration, unfortunately for Moldbug’s theory, cannot be proved true or false by reason. Not wholly, that is; some claims of inspiration are subject to debunking, but only when those claims intersect the contingent world. For instance, we can test prophets and would-be prophets. Anyway, it is unreasonable to reject the possibility of inspiration. Any argument saying inspiration is impossible is, or would be, circular (try it).

“Reservationists,” Moldbug says, “are fascinated by the interpretation of human affairs. In human history, politics, and economics, we observe patterns which appear to be patterns of cause and effect.” Human affairs are a fine thing in which to take an interest. And it is good to hope the patterns we see really are patterns caused by forces we think we have identified, and that therefore can be used as analogs for forecasting. But how to do this?

Moldbug sees that Bayes theorem cannot provide the structure or foundation of this system of interpretation. Bayes is rejected because it is a mere algorithm, but then so is reason. Reason takes what is given and acts on it, producing output, just like Bayes. Therefore we must reject reason as our epistemological basis, too. Which means we’re right back to the hard problem of figuring out how inspiration intersects with intuition and reason. As for that, I have nothing new to offer.


  1. JH

    Axioms are assumed true and serve as groundwork in a mathematical system. Which doesn’t imply that everyone believes them.

    If you assume mathematical truths exist in who-knows-where and is awaiting to be discovered, why do they need to get off any ground? They can just sit there and look pretty. 🙂 Reason is a craft to who-knows-where, just as a thinking boat is needed to reach an island. Sometimes it gets lucky and runs into an island that may not be the original destination.

    I think that Bayes is an algorithm. However, reasoning and critical thinking would be required to properly employ the algorithm. Depending on the definition of Bayes, one can program a Bayesian framework, but cannot program reasoning and critical thinking (I don’t know how, at least.)

    I never heard of Mencius Moldbug. I like the name!

  2. Joy

    “Mencius Moldbug” Briggs made it up on the fly.
    and when he’s calmed down he might consider posting my comments again.

  3. isn’t the whole gist of the first quote exactly to propose that reason is but a special case of common sense?

  4. Ray

    “Axioms are assumed true and serve as groundwork in a mathematical system.”
    Thermodynamics is also based on axioms. These axioms are as follows.
    1. You can’t win.
    2. You can’t even break even.
    3. You can’t get out of the game.
    4. The perversity of the universe tends towards a maximum.

  5. Oldavid

    That’s beautiful, Ray.
    It should be regarded as the definitive exposition of Murphy’s Law.

  6. Mactoul

    “Bayes is rejected because it is a mere algorithm, but then so is reason. Reason takes what is given and acts on it, producing output, just like Bayes”

    An algorithm is a well-defined procedure. Bayes is certainly an algorithm but “Reason” is not and can not be.
    For if reason were an algorithm, machines would think.

  7. Oldavid

    Reason, by itself, is a metaphysical undirected missile. It need not have any secure and certain launch point (certain premises) or a trajectory guided by the scientific rules of logic. The science of logic is nothing but the metaphysical rules for coherent, consistent reasoning.

    Reasoning is just the (metaphysical) extrapolation of ideas by deduction, induction or pure convenience. It does not contain within itself any guarantee that it is not a fanciful error.

Leave a Reply

Your email address will not be published. Required fields are marked *