Comments on “Rationality, rationalism, and alternatives”

I happened to run across a quote today, attributed to Albert Einstein (admittedly I haven’t checked because who cares?), that fits nicely with the theme of this page:

So far as the theories of mathematics are about reality, they are not certain: so far as they are certain, they are not about reality.

It seems to me that most of controversies in philosophy of mathematics are at root about this mismatch and what to do about it.

You mention that rationalism and metarationalism are incompatible explanations about how and why rationality, metarationality, reasonableness work.

You also mention that rationalism is normative. Telling us how we ought to use rationality.

What about meta-rationalism? Is it also normative by your definition?

My preference is that meta-rationalism seems to suggest what you should do, but is actually not. By understanding its definition clearly and then applying it, you automatically understand how to use rationality etc as it is best designed for.

Is it normative to use things the way they are best designed for? Maybe a new adjective needs to be invented. Normative might be too broad for my own taste.

SIM KIM SIA,

I’m not David, but my own view is that it’s normative, but it’s a nebulous norm: “Given an X-ish situation in which Y-ish things are important, do Z-ish things.”

THanks James for response. Given X probably a good idea to do Y is similar to what i was suggesting about doing things based on what they are best designed for.

So I suppose, in metarationalism, there are no universal context-independent normatives.

By plural nouns, I mean specific normative goals or values.

I suppose it will be like rationalism is normative (in the attitude sense hence the adjective).

Metarationalism is non-normative (as in silent in the attitude sense)

But what metarationalism will likely focus on are normatives (in the plural noun sense) that are context-dependent (as in your Given When Then). Any context-independent expressions of normatives are actually also context-dependent.. just that the context is yet to be discovered or purposely omitted for brevity’s sake during communications.

Thoughts?

SIM KIM SIA,

So I suppose, in metarationalism, there are no universal context-independent normatives.

That’s my take as well. To me the Platonic conception of the Good is a millstone around the neck of ethics and norms generally. Just like it makes no sense to do evil for the sake of doing evil, it makes no sense to do good just for the sake of its own abstract goodness.

A few ethicists of a more Aristotelian bent have argued that actions and things are not good or bad tout court but good or bad in a certain way, good for this or bad for that. I take that approach while avoiding both the Scholastic mistake of thinking of good in terms of conformity to fixed, well-defined essences and the existentialist mistake of thinking that terminal values are subjective or arbitrary.

Metarationalism is non-normative (as in silent in the attitude sense)

I have to admit, I’m not sure what “silent in the attitude sense” means here. Could you elaborate?

As in agnostic about whether capital S Should exists in the eternalistic sense or does not exist in the nihilistic sense or that it’s arbitrary in the existentialist sense

Hence silent. Makes no position.

Also not saying the position towards normatives is no position which is self defeating. Just simply no position.

At least that’s how I read it. Thanks for writing. Your phrase about “existentialist making values arbitrary” is useful for me.

As I understand it, Yudkovskyan rationalism disagrees with meta-rationalism in claiming that
4. You should conform to the criteria as neatly as you can.
is true, and of the 7 claims, it only disagrees with meta-rationalism about that claim.
(With Yudkovskyan rationalism I mean the “LW rationalism” expounded in the Sequences)

I don’t think Yudkowsky neccesarily believes in an One True Rationality.

However, by “system” I mean, roughly, a set of rules that can be printed in a book weighing less than ten kilograms, and which a person can consciously follow.11 If a person is an algorithm, it is probably an incomprehensibly vast one, which could not written concisely. It is probably also an incomprehensibly weird one, which one could not consciously follow accurately.

But Yudkowsky is primarily an GAI researcher, focusing on GOFAI presumably due to the reasonable assumption that one can hardly align an implicit consequentialist if one doesn’t know how to align an explicit consequentialist.

On the other hand maybe Yudkowsky occasionally belives in a One True Rationality, as long as you qualify it by “but it doesn’t exist yet” due to failure to keep apart his FAI research and “the art of human rationality”.

Yudowskyan rationalism cannot claim that meta-rationality is important due to being mostly unable to distinguish it from rationality, tho.
But what I would most want is discussion of whether 5: “Conform to the criterion!” is true, while agreeing that 1&2 are true and that 4,6&7 are false.

To that end I will quote 12 virtues of rationality:

As with the map, so too with the art of mapmaking: The Way is a precise Art. Do not walk to the truth, but dance. On each and every step of that dance your foot comes down in exactly the right spot. Each piece of evidence shifts your beliefs by exactly the right amount, neither more nor less. What is exactly the right amount? To calculate this you must study probability theory. Even if you cannot do the math, knowing that the math exists tells you that the dance step is precise and has no room in it for your whims.

The bolded text I take to mean that you should conform to the criterion (i.e. 5), with the criterion being probability theory, which you admited is true as far as it goes.

(This is still half a draft, but I decided to post it before I ran out of steam and left it as a draft, in the end leaving what I’d written to disappear when I shut of my computer. Sorry if I’m rambly.)

Hi,

You mention that the scientific method can hardly be defined and is therefore implausible. I am by no means a rationalist who wants to believe in it, but I’ve caught glimpses of a certain “vastness” to the scientific method that I encountered in University, that goes beyond what they teach you in high school, and makes me wish I had learned more about it. It’s something about how you don’t just make a hypothesis, but you make two: a “null” hypothesis and a “positive” hypothesis, and you check to see how they interact before concluding anything. Also, making hypotheses is not the first step you do, but the last one, after much careful consideration. Something about how before that, you’re making a theory, but not applying it to the world and/or checking to see how it interacts with what you see. I might not be remembering it very well since it was a long time ago, but it seems like there is an ideal there that goes really deep.