[FRIAM] Optimizing for maximal serendipity or how Alan Turing misdirected ALife

[uǝlƃ ☣]

For my part, I think both the "adjacent possible" and any distinction between inter- and trans-disciplinary are confused concepts. The clearer conception surrounding serendipity lies in the distinction between formal and informal (not formal vs. intuitive, but related). An equivalent (I think) conception is "flex and slop", which I think is attributed to Dennett, and (again [sigh]) Feferman's schematic axiomatic systems. I've been struggling to find a good way to express my contrary perspective to your post (On 5/26/20 7:20 PM, Jon Zingale) where you *seem* to go completely formal. My intention was to imply something *informal* ... sloppy, badly formulated, etc.

The inherent problem with the inter-/trans-discipline, Dave's polymath, and Jochen's fools outside one's house, concept [†] is that there does not exist a complete and consistent formalism capable of informal generation/construction. Only informal systems can do the constructing.

Of course, "informal" is ambiguous. Does it mean "only broken in one spot" ... "a slightly incongruous composition of formal systems" ... "integrationist Rube Goldberg assemblages of formal systems" ... "a radical resistance to all formalization" ... "a mishmash of confused nonsense"? Etc. The primary task of AI and ALife is to find out just how informal we *must* be to sit back and call something "intelligent" or "alive". And this lands squarely in the long, practical, tradition of finding out just how informal we *must* be to, say, build a good bridge, fly to the moon, or bake a tasty loaf of bread.


[†] Concept, singular. The implicit assertion being that all those are the same thing, the same mistake. I can reword my complaint as "There are no 'disciplines.'" "There are no houses." The concept of polymath *might* escape the category if Dave chooses to consider "learned"/"knowledge" as a massive noun ... where one can increase their learning by ε even as ε→0. But if there are disjoint domains, then polymath is in the same category.

On 5/27/20 3:51 PM, Jon Zingale wrote:
> In some ways, what gives material life a foot up on simulated life is a sense
> of /maximal serendipity/. To a large extent, I feel that this is the central
> argument of analog-high-fidelity loving nerds. There is a recognition that
> functions (plural) may in-fact follow from form. In those fiery digital vs. analog
> debates, it is easy to lose sight of the fact that transistors are by their very
> nature, /analog devices/. They are chunks of matter influenced by the world at
> large. The function space for how such matter is influenced is likely non-
> enumerable, much less what can be done with such influence. The imposition
> that a transistor /behaves/ digitally is an imposition demanding that the device
> act as a unit for symbolic manipulation, to act within clearly delimited bounds.
> 
> Consider, by analogy, the tails of aquatic mammals. Before they were tails,
> these appendages were evolved for walking. Later, they would be improved upon
> for swimming. Nature appears to work with what is readily at hand, and the space
> of possible functions is not likely to be concretely specifiable. To my mind,
> this is where the hypnotizing concept of a Turing test led the program to develop
> artificial life, astray. Here we set up a useless paradox. We demand that whatever
> system we design /forcibly/ participate in our investigation. We demand that it
> /behave/ like a good and servile device, and then we complain that we have failed.
> Perhaps, my chair is conscious in-part because it, like the sadists, says no. I am not
> necessarily committed to this position about my chair, but I do think it points
> to the self-defeating nature of Turing tests. In another post, Merle emphasizes
> the importance of identifying transdisciplinary research. In particular, she
> mentions its connection to the /adjacent possible/. Whatever will one day be called
> ALife, will only be interesting if it is capable of exploring such a domain.
> In an effort to contribute to this program, I advocate for taking seriously ideas
> like embodiment and potential for serendipity. Given consciousness, the question
> of /How do we know? /maybe the least interesting path of investigation.

-- 
☣ uǝlƃ