When is reasoning with "unknown" knowledge permissible?

E.g. when can one "legitimately" formulate political beliefs or laws based on knowledge, which contains unknowable, unpredictable parts?

If one considers this strictly, then one could claim that "no, it's not possible to accurately infer with unknown knowledge". However, one could also argue that "some things will always contain unknown knowledge, should we therefore still be able to decide on it, even if it's based on subjective bias?".

I personally think that Occam's razor + some others would suggest that inferring with unknown knowledge is not a very good thing. That, while it might "by chance" hit the right spots, it contains too much risk about being very, very erroneous. And that it should be discarded as a methodology because of that.

Also, natural scientific "unknown data" is of different type as e.g. "unknown social data". Natural scientific portion of error can also often be measured or approximately known. Social phenomena on the other hand may well have components which are very, very vaguely understood, even if someone could suggest an explanation.

  • "inferring with unknown knowledge is not a very good thing." I agree. And doing so suggests some sort of delusional process, imho. – Bread Feb 12 '19 at 11:43
  • In this generality, yes it is easily possible to "accurately infer with unknown knowledge". I may not know what the current state of a physical system is, but as long as I know its energy and momentum I can accurately infer that they will remain the same as long as it stays isolated, however unpredictably it behaves. How far this extends into the political arena I would not presume to guess. But there is no problem, in principle, with formulating constraints that will have to hold, to some degree of confidence, regardless of the unknowables, and infer things from them. – Conifold Feb 12 '19 at 13:55
  • @Conifold How does one measure the confidence then? Also by "accurately infer" I don't mean that having a miniscule chance of hitting the right answer counts as being accurate, even if it's that miniscule chance. I would want that one can demostrate that one's claim works for at least 50% of similarly imaginable cases. Even more preferably, for over 95% of the cases. Or if it works for less, then it should be labeled as being applicable in just a minority of cases and with some degree of accuracy. – mavavilj Feb 12 '19 at 14:29
  • We do not. How sure are we that the Earth's population will be increasing, and no miracle source of cheap energy will be discovered? Pretty sure, I'd say, but slapping numbers on that is just a rhetorical exercise. What is "similarly imaginable"? Outside of mathematical models for very narrow context any determinate sample space is just a fantasy. For vague fields like politics confidence judgments are equally vague, and it is better not to pretend to quantify them. The only real measure is this: confident enough to act on it. – Conifold Feb 12 '19 at 19:37
  • @Conifold "Similarly imaginable" could mean "exact same", but I've found that e.g. mathematics is often generalizable at least up to some level. Thus it makes no sense to consider measuring a 1cmx1cmx1cm box and 4cmx4cmx4cm box without understand that some same principles apply, even if the scales differ. However this is not the point, the point is, when should unknown reasoning be permitted in the first place? – mavavilj Feb 12 '19 at 19:45

I'm going to assume that you don't mean permissible so much as advisable. There's nothing really stopping you reasoning.

In the right context, reasoning based on unknown data can be very useful.

A good example is in proofs by contradiction. Here you have a statement that you don't know is true or false. You follow through the logical consequences of the statement until you hit a contradiction to your assumption. If the logic is correct you now, somewhat magically, know your statement is false. Unknown data becomes known.

Similarly, in science, we often do something similar. We take something that may or may not be true. Follow through the consequences until we find some hypotheses that we can test. If the tests fail then our assumption is less likely to be valid. If they succeed, although it doesn't prove our assumption, it adds weight to its likelihood of validity.

| improve this answer | |
  • Proof by contradiction works, because it relies on foundational or axiomatic system in the background. Not all contexts have this. – mavavilj Feb 12 '19 at 14:29
  • @mavavilj I'm not sure I agree with you but i think it's beside the point. I'm not saying that all reasoning on unknown data is useful. Just that it can be. And i gave some examples. – Alex Feb 12 '19 at 14:54
  • How can you prove proof by contradiction without an axiomatic system? (BTW, this question should be unanswerable) – mavavilj Feb 12 '19 at 15:07
  • @mavavilj I mean the "not all contexts" bit. An axiomatic system is just the set of assumptions you apply logic to. I can't think of many (any?) cases where that's not possible. – Alex Feb 12 '19 at 16:13
  • But you need a bit more than "just decide on some set of axioms". Mathematical logic has developed to a direction in which a desire is to have so fundamental axioms that they would be unquestionably intuitive to a western mind. However, even that takes some developing such as discovery of Russell's paradox. Then one could still argue that taking natural numbers as more fundamental than complex does not make natural sense, even if it makes "human learning order" sense. A practically sound axiomatic system I think, requires perhaps as much as "scientific thought system". – mavavilj Feb 12 '19 at 18:04

...when can one "legitimately" formulate political beliefs or laws based on knowledge, which contains unknowable, unpredictable parts?

The most common method that people use is abduction. Abduction gathers observations that can be made, other information that can be relied upon, and assumptions that can be rationally drawn, and uses them to produce an inference to the best explanation.

Such conclusions never have as solid a base as a deduction, nor the probability of an induction. Abduction "yields a plausible conclusion but does not positively verify it." Wikipedia: Abductive reasoning. It provides a non-arbitrary way of dealing with the unknowable.

I agree with the useful observations made by Alex earlier.

| improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.