notes-philosophy-computationalPragmatism

Computational pragmatism

I am considering a view that i call "computational pragmatism". It seems related to "pragmatism" but perhaps not identical to it. It is also related to ordinary language philosophy. It is not well-formed and i do not know enough philosophy to know if it is unique.

It is not really a viewpoint with much actual content, rather it is a viewpoint on how philosophy should be applied, and perhaps also how it should be done.

It is, on the one hand, an acceptance of objective truth and a single, universal method of rationality, and an assignment of importance to precise, formal language and reasoning and to the results of such; while on the other hand recognizing that it can be extraordinarily difficult or even in many cases impossible to find out what is true, that we do not yet confidently have a precise consensus description of what it is to be rational, and that precise language and reasoning is extremely computationally expensive. This leads to a provisional, pragmatic acceptance of guessed truths, heuristic methods, and imprecise language, while at the same time to a committment to reject the results of these methods at some future time if a rational, formal method comes to different results.

Let me give some examples:

• computational pragmatism would say that "1 + 1 = 2" is objectively true, and would tentatively assume that the concept of the number "2" is 'real'
• computational pragmatism would say that there is a single universal formal 'logic', meaning a formal method for valid reasoning; however it is not clear yet (and possibly it never will be clear) what this method is
• when faced with tricky epistemological and ontological questions, computational pragmatism's response is often, 'since there is not a consensus among experts, i assume that there is a correct answer, however, i do not currently know what it is. Even if i were to become an expert myself and take a position, unless others were persuaded by my reasoning into consensus, i would have to assume that as i am fallible, i may be missing something. Perhaps someday there will be a consensus, at which point if i examined the reasoning and agreed with it myself, i would take a stronger stance (although, bearing in mind the past absence of consensus, indicating fallibility on this topic, not a 100% strong one). In the meantime, however, i will try to not take a position on things affected by this issue, or, if i must, i will provisionally adopt the 'naive' position.'

One way to put it is that computational pragmatism views a priori reasoning to be non-comprehensive due to the high cost of computation and to the fallibility of philosophers, and hence reliant on an unpredictable process of discovery. In this way it is, in practice, in a similar situation as science, which is unsettled because of the possibility of future discoveries. Therefore, just as one of the properties of the scientific attitude is being comfortable with saying "I don't know", computational pragmatism advocates an attitude similar to scientific skepticism (but not radical philosophical skepticism) applied to fundamental philosophical questions.

Also, science is applied to engineering even though science itself is unsettled; formalisms and sub-theories are developed to assist in these applications, even when it is known that the science they are based on is in fact incorrect (for example, various notations based on Newtonian physics), provided that that part of science tends to give approximately correct answers within some relatively predicatable set of contexts. Similarly, computational pragmatism advocates that, even while fundamental philosophical questions are unsettled, we should find ways to 'build on unsteady foundations' and go ahead and build pragmatic formalisms and sub-theories.

Techniques commonly used in computational pragmatism

I am not claiming that these are unique or that they are a core part of computational pragmatism.

Interpolation

If there are two opposed viewpoints on a controversial topic, for example realism and nominalism, then a future theory which is successful at achieving consensus must appeal to the sorts of people in both camps. Theories which are in some sense "in between" these are good candidates. Of course, "in between" is not well-defined in matters such as these, and in addition it's also possible that the future consensus theory will not be in between, but rather will be one or the other current contenders, after demolishing the opposition with some as-yet-undiscovered critical argument.

A deeper theory and a pragmatic theory

If there are two opposed viewpoints on a controversial topic, one resolution may be to accept one of those views to be correct at a deep, absolute, theoretical level, and yet to say that in most applications of the theory, it is better to act as if the other one were correct, for pragmatic reasons. However, when they conflict and when the actual truth is desired rather than a pragmatic decision, the deeper one always pre-empts the pragmatic one.

This is, for example, the position we take with regard to formal argumentation; in fact we view there to be objectively right and wrong answers to various philosophical questions, and we view formal argumentation to be infallible and to be the only standard of validity. However, because formal argumentation is very expensive, because we don't trust that we have yet discovered the complete method of formal argumentation, and because we don't trust our fallible minds to apply it correctly, for most practical purposes we advocate a reliance upon informal argumentation, and for the same reasons we advocate acting as if there is some subjectivity and validity of multiple viewpoints, even though in actual fact all but one opposing viewpoint must be incorrect.

Without committing computational pragmatism to the following, i note that another example of this technique is from ethics, namely a pragmatic form of "rule-based utilitarianism", wherein one says that some sort of (act-based) utilitarian view is true, however, because of the high utility of having individuals who can predict one another's ethics-related behavior and who can make credible committments to each other, one should act as if some sort of deontological theory is true.