>The purpose of step 3 is to compare the second set of information in the computer to our experimental observations.

Good - I was also getting there, to the good old "and then we predict a result based on our theory and then see whether the experiment yields it or not". The part I didn't get was, well, all of this - what was it that was supposed to happen, and where was I supposed to look to see whether it did.

>If the results don't match, the hypothesis (the computer program's rules and initial conditions) is disproven.
>
>You could look at this like a scientific solipsism.
>
>Or a scientific existentialism.
>
>Or a scientific idealism.

Which doesn't say much - even sociology can boast that it's applying the scientific methods, but it still doesn't have any universally applicable laws except "people will behave".

>But how we talk about the mathematics isn't quite as important as the mathematics themselves.

And whenever we apply maths to something, we never really know whether the model is true to the reality or not. We can only say that if the model does so, then it will happen like this. Note the use of word "reality" here - it may make the whole issue irrelevant (if we define reality to include this mathematical model, then whatever the model shows is part of that reality), or it may make it irrelevant (if our reality is defined so that it is unrelated to maths) or anywhere in between (if you assume the former were opposites and you've advanced from judeochristian dichotomy to gray scale linear model).

>Yes or no questions can be answered in an epistemological context such as a set of axioms.

Which is why I allowed for "explain why question is wrong" - i.e. what's your set of axioms, and what's your logic, i.e. rules of inference.

>Notice in figure a there is absolute reality and relative reality.
>
>Both have their own version of truth. There is absolute truth and there is relative truth.
>
>Your question asks about the relationship between two kinds of truth.
>
>So the question refers to an area of knowledge where our standard axioms aren't very applicable.

Well bring yours forth. Never mind if they aren't sufficiently normalized, formalized and if your rules of inference can poke holes in your axioms. It can all be polished later.