Hello DoctorDick. I reread the thread hoping to gain a firmer footing. Unfortunatly, like quicksand, I think I just got swallowed up a little deeper
I'm exaggerating
hopefully
Quote:
Originally Posted by Doctordick
It is quite simple actually, anytime one has a theory one must be able to compare the consequences of that theory to the actual information on which it is based. To the casual observer, it might appear that the “i” index is sufficient to that purpose; however, that is not quite true. The problem with the “i” index is that it is defined by the specific theory being examined and that theory could be incorrect. The fact that the theory might be incorrect implies that the “i” index identification could also be incorrect. It should be held in mind that being correct is actually quite a different classification than is “flaw-free”.
If we are to include all possible “flaw-free” theories, we must allow for all possible mappings from each of those theories to the actual information which is to be explained. Our model must allow such a comparison without specifying what the actual labels are.
|
Yes, I understand the difference between being correct and being flaw-free. Actually, you'll see I introduced the issue in this thread in post 18
Quote:
Originally Posted by modest
So, I'd say it's true that "given an arbitrary prediction, an acceptable explanation could tell us if that prediction were acceptable or not", but that would not necessary mean the prediction would be correct.
|
I do understand that an explanation may be flaw-free yet incorrect. What I was a bit unclear on was the need for 2 labels for an element ("x" and "i") and also two lists. If the information list is the past, the actual information, then I don't see why it needs or why it should contain "i". On the other hand, if the explanation list contains "x" then I see no purpose for the information list at all. It's already represented in the explanation list.
Quote:
Originally Posted by Doctordick
You are correct; there are indeed going to be three different lists here serving three very different purposes
|
Oh, my.
Quote:
Originally Posted by Doctordick
The “knowledge list” (as I intended to use it in the quotation you referred to) is an excerpt from the information list which is used for the purpose of analyzing the actual use of the “explanation”.
|
Ok. Would you say the knowledge list is part of the past while the information list is the whole past (where the past is defined as what is known)?
Also, does the knowledge list contain unknowable information as implied here:
Quote:
Originally Posted by Doctordick
From that perspective, we can conceive of “having the explanation list” (no mention of how that result was achieved) without actually having the “knowledge list”.
|
It's also difficult for me to imagine what the information list and the knowledge list are going to be practically used for if their information will always be indeterminate.
Quote:
Originally Posted by Doctordick
The fundamental purpose of an explanation is to allow one to deduce the constraints imposed upon the answer to a question when given only small part of the relevant information. The essential situation is that if we actually knew the entire “information list” an explanation is rather beside the point; we could simply refer the questioner to the ”what is”, is “what is” explanation.
What is important here is that other, quite simple, explanations exist. Explanations which do not require memorization of the entire “information list” and it is these explanations which are of interest to us.
|
Right. I agree, a rather small explanation can explain a rather large set of information.
Quote:
Originally Posted by Doctordick
In the absence of persistence, I suspect the simple ”what is”, is “what is” explanation is the only possibility: i.e., we have the information list and nothing more.
|
But, I can conceive of information where there is no persistence such as (x)
t = {(1)
1, (2)
2, (3)
3...} yet an explanation such as dx=dt is possible. By my thinking this would be like a timer—not persistent, but explainable by means other than a list of each tick.
Quote:
Originally Posted by Doctordick
So, presuming you understand everything up to this point, that brings me back to the question I asked earlier. The one I referred to in post #35.
Quote:
Originally Posted by Doctordick
Let us go back to that original question, suppose we are given a set of numbers (a supposed “present”) and are asked, “what is the correct t index assigned to that set?” The possibility certainly exists that there are multiple presents with exactly the same set of reference numbers associated with different “t” indices. In that case, our ”what is”, is “what is” explanation (as currently defined) will fail to provide us with an answer.
|
This thread has now exceeded fifty posts and I was unable to find the post where I originally asked that question which is somewhat troubling. Nevertheless, this is a very important question and resolving it will lead to more subtle issues. We need to be able to handle this kind of question. Let me know that you understand the issue I am bringing up. |
I suspect this is where tau gets introduced. To be honest, the development of this model is not feeling like a deductive method to me.
The "t" index distinguishes different presents. Only by arbitrarily ignoring the value of "t" for a given set does one have trouble telling in which of two otherwise identical presents it belongs. Arbitrarily ignoring one index doesn't seem like a good reason to add another.
~modest
