Review for Submission 134
Accept. Clear, well-written, interesting.
A likely framework for the description of intelligence will involve
agents that self-organize, approaching a phase transition criticality
where reorganizational avalanches occur.
Although this paper does quite not use these thermodynamic words,
it lays much of the background conceptual framework for such a
description.
-----------
One suggestion for changing the formula to use more conventional
notation for the concept:
Page 6 Section 4.2
A more common way of expressing the relationship here is as a kind
of generalized enthalpy. The conventional notation would be
"change in task progress" = Delta P
"amount of communicated data" (change in communication) = Delta C
and thus
$ Delta P = U + R_CAN Delta C $
is the formula in thermodynamic form. It avoids the need for epsilon
and for a "scaling constant" and most importantly, eliminates the time
derivative, making it clear that this would happen anyway, even if
the CPU's and communications channels were slower/faster.
This form makes contact with existing literature in the calculus of
communicating processes, pi-calculus, PEPA, the theory of traces
(trace and history monoids), etc.
----
Below follow some typographic corrections.
Page 3
> Limited memory, free energy, time, information, and other re-
sources are indispensable; without such limits, the reflexive cycle underlying
adaptation cannot start.
The opposite of "limited" is "unlimited". Perhaps this means to say
"available" or "greater than zero".
Page 6:
The formula at the top of page 6 is a bit opaque; First, it is written
that R_i={r_i,1 ... r_i,N} implying that r has two subscripts: r_i,j
and that r is a rule. But then, they symbol r with one subscript
appears, and it is called a reward, not a rule. Further down, there
is a bold-face 1{j = a_i(t)}; the meaning of the bold-face 1 is unclear.
Is this some indicator funtion? Then a_i(t) is called the "chosen rule"
even though at the start, the symbol used for rules was "r" not "a".
The idea that there needs to be some update is clear enough. The
specific example seems quite interesting but is incomprehensible.
Page 6 section 5
"In table roles and limits of different agents was presented."
This is grammatically awkward. Perhaps
"Roles and limits of different agents are presented in the table below."
Next paragraph: the comma after CANs not needed.
Page 8 Table 1
" For provided example semantic rule"
should be
" For the provided example semantic rule"