🚀 go-pugleaf

In article <7892@skye.ed.ac.uk>,
jeff@aiai.ed.ac.uk (Jeff Dalton) writes:

>>As I said before, for me "behavior" is not a sequence of actions, it
>>is a causal relationship between inputs and outputs. The code is the
>>specification of this causal relationship.
>
>So why didn't you agree with me any of the other times I've talked
>about such things as looking at source code? Or when I said we
>ought to look at more than the sort of behavioral evidence (is
>that still a term I can use?) that could be seen in the TT?

Let me try once again: What I mean by the behavior of a
(deterministic) system is the function giving the output as a function
of the history of inputs it has received. (It is more complicated to
say what behavior means for nondeterministic systems.) Observing a
system does not determine its behavior, but only determines bounds on
the set of possible behaviors. Looking at the code for a program, on
the other can in principle tell you everything about the behavior. So
I certainly agree with you that you can tell more about behavior from
the code than you can from any finite amount of observing the system.

However, when you were talking about "looking at the source code", I
thought you were talking about using the source code to distinguish
between behaviorally identical systems. For example, the lookup table
machine has the same behavior as a human (even if it doesn't execute
the same "code").

In article <7895@skye.ed.ac.uk>, you say:

>>The problem for me is that I don't know of an objective, coherent,
>>nontrivial notion of functional organization. To be nontrivial, it has
>>to be the case that two systems that are behaviorally identical can
>>have different functional organizations.
>
>But in another article you say:
>
>   "behavior" is not a sequence of actions, it is a causal
>   relationship between inputs and outputs. The [source] code 
>   [of a program] is the specification of this causal relationship.
>
>Now, it sounds from this like functional organization and
>"behavior" are almost the same thing.  A specification of
>a functional organization would also specify the behavior.

I wasn't saying what a functional organization is; I don't know.
But I was simply saying that any notion of functional organization
is trivial if it turns out to be the same as the behavior.

Daryl McCullough
ORA Corp.
Ithaca, NY