« first day (1979 days earlier)      last day (2279 days later) » 

18:37
Hey moderators
I think this question should just be deleted meta.mathoverflow.net/questions/3981/…
It's becoming a big argument
and it has the potential to blow up and etc
 
4 hours later…
22:13
Hey guys. I've got a question I'm thinking about fleshing out and posting here... mind if I run it by y'all?
I've been thinking about computer proof assistants. Any given proof assistant will have a certain logical theory which it implements.
The problem is, it seems like it's impossible for any logical theory to be "satisfactory".
In order for a logical theory to be satisfactory, it should be possible to study it and determine that it has these two properties:
1. Each of the axioms of the theory is a true statement.
2. Every statement which we can possibly prove (using our intuitions and natural language) is a theorem of the theory.
Let me modify that second criterion, actually.
2. Given any statement, if it's a statement whose truth we can be convinced of, then it's a theorem of the theory.
Suppose we've studied a theory and found it to be "satisfactory" in this sense. Well, each of the axioms of the theory is a true statement, so that convinces me that the theory is consistent. So, since we've been convinced that the theory is consistent, criterion 2 requires the theory to have "the theory is consistent" as a theorem.
But that's not possible if the theory actually is consistent.
22:36
So, this seems to present a paradox. Either our intuitions about mathematics can't be formalized (which can't be right), or our intuitions about mathematics are contradictory (which also can't be right), or I've made a mistake somewhere.
And the question is: what's the resolution?
Of course, if I do write this up as a question, I'll make it a lot clearer and less confused-sounding!
23:28
Tanner, don't ask it.
It will be closed mercilessly.
There are a bunch of 90 year old results like the undefinability of truth
and Gödel's incompletness theorems
Awodey has a nice talk about this youtu.be/alLgEf0uVkg

« first day (1979 days earlier)      last day (2279 days later) »