[FRIAM] A Question For Tomorrow

[uǝlƃ ☣]

Very interesting.  In the last post, I deleted a paragraph where I analogized the human population to a swarm intelligence optimization problem, each human being an ant pursuing her own little solution, but the whole circumscribing (up to a convex hull) the solution space. I deleted it because I was being too glib with the analogy. Humans are very complicated organisms, not zero-intelligence agents.

But we do have *modes* where we behave very depth-firsty. My depth-firsty methods don't kick in in very many contexts. Others tend to use them in more places. E.g. I have a friend who really dug deep into ferementation food and drink. I limit myself to beer. But he launched into everything that involved any type of fermentation to get a deeper (hands-on) understanding of our symbiotic relationship with those little bugs. I struggled to stay interested in any context but that of beer ... though the bread phase was interesting. In any case, even *that* relatively narrow aspect of "food and drink" is pretty diverse, almost fractal. We can find a connected path from any part of fermented food and drink to pretty much any other aspect of humanity that I've ever encountered, from domesticated animals to humans visiting the far side of the moon.

All that text is merely to provide context that my guess is your depth-firsty commitment to a reasonably trustworthy reductionism isn't as depth-firsty as you think it is. It's more like those massive muscles in your back or leg that attract all the attention, but that are useless without the thousands of little control tissues providing the context that allows the big guys to do their work. The real weight is being pulled by the infrastructure, not the rock stars. Anyone whose suffered from Tennis Elbow will attest. 8^)

On 5/1/19 10:29 AM, Marcus Daniels wrote:
> One recollection from many years ago was debugging a generational garbage collector (GGC).  The program with the GGC would crash after hours due to a memory corruption that manifest itself via multiple layers of indirection.    C programs often have memory overruns that create similarly baffling outcomes, but this was worse due the complexity of the algorithm.   The advice I got from one expert was to ratchet it down on degree of freedom at a time.  It was incredibly tedious, days of work, and required systematic bookkeeping.   I eventually found the problem.   That reductionist approach from experiences like that, is burned into my psyche and has paid-off many times.    The alternative is suspect to me at a primal level.    Pulling up stakes and trying something else only slightly different is wasted motion.   There has to be some clear stopping evidence to show an approach is flawed before one pulls up stakes.   Otherwise it is just a game of musical chairs.   So to me jumping between different modeling approaches or "views" speaks not to plasticity but a lack of commitment; it is an act of desperation.
> 
> Obviously this is not a justification.

-- 
☣ uǝlƃ