KarlPopper's notion of how "close" to reality a theory is :
- your theory describes the world in a language
- assume there's a language which describes how the world "really" is
- assume there's a MetaLanguage which can describe how to translate between the two
- the degree to which your theory approximates the truth is proportional to the description length of a text in the meta-language which would explain / correct the errors in yours.
I find this pretty plausible and useful.
The Hegelian counter is this.
- Kant says we can't know the world as it "really" is. So we don't have access to this "true description". So we can't measure it. We can't even know that the "reality" even exists, so it's meaningless to talk about it
there is only your theory.
This is another justificationist CantGetThereFromHere argument. It's based on the assumption that items of knowledge can't succesfully refer to things (or be had) unless they've been reached by correct series of
inferences. As a CriticalRationalist who believes knowledge is conjectural I don't have this problem. I conjecture there's a reality, I conjecture there's a linguistic description, I conjecture there is a quantifiable gap between my model and it. These conjectures don't have to be justified to succesfully refer. So this description of the quantifiable gap, and this notion of degree of approximation are valid. Sure, I can't, in practice measure the gap. But I can succesfully talk about it and, more importantly, appeal to it as explanatory eg. why Einstein better than Newton? Because gap between Newton description and the reality is smaller than the Einstein one. The text to say what's wrong with Einstein is shorter than the text to explain what's wrong with Newton.
See also :