[FRIAM] it's world logic day!

[uǝlƃ ↙↙↙]

Both EricS and Marcus launched in the direction I want to go, which I *think* ends in parallel Turing machines. Modern AI/ML is clearly a success. And I'd argue that its success depends fundamentally on its multi-paradigm modeling (or as it makes more sense to me, multiple "models of computation" -- MoCs). Some MoCs target ordering directly (to relax from or tighten to total ordering). Some MoCs attempt to ignore as much ordering as possible and satisfy some set of weakest constraining predicates.

But the important thing (I think) is how the MoCs are [dis]assembled into and out of working things. 

We have process calculi that handle parallelism well enough. But, in my ignorance, they seem "classical" (for a lack of a better word) rather than constructive. What I think we need are constructive models (like Turing machines) for the [de]composition of cooperating collections of logics. And I speculate that a proto use case for such are inconsistency robust IT systems. But perhaps a better one would be how human sensorimotor systems handle contradictory streams of information (e.g. the physiology of synaesthesia).

I tried several months ago, in my inadequate way, to surf whatever work I could find on parallel Turing machines, but got distracted. If anyone has any short-cuts to hand my lazy @ss, I'd appreciate it.

Regardless, Merle's objection sounds like *any* one logic will (of course) fail to represent everything. And the extent to which we can change logics so that we choose the one fit to purpose, what we *want* of Logic is a way to construct executives that can do the fit-to-purpose selection given whatever context *and/or* compose encapsulated logics so that the composition is more expressive than the components.

Sorry if that's incoherent. I'm trying to watch talks on logic while writing this.


On 1/14/21 12:29 PM, David Eric Smith wrote:
> I also found the branch to normativity odd.  Took listening to several rounds of this joust to find quite the right metaphor to express why.
> 
> If I suddenly became Chaitin, and cared more than anything else about which numbers were random and which weren’t, I would be saddled with a need for criteria by which to assign such attributions.  That would blossom into a whole system to study kinds of order and what-else there might be where they are absent, and there goes that lifetime.
> 
> If I wonder whether a series of thought events is ordered, and whether the way it is ordered would also be shared by some other set of thought events so that they could be understood as some kind of family, I would then need to pursue a theory of what kinds of order thought-events can have.  Since I am NOT a logician, and look in through the window as the elves work, I can only say that that is what the work looks like to me.  The question whether conditions of order can be worked out seems to me much wider than an effort to develop true positions, and also not dependent on values of rightness.
> 
> If someone asserted to me that a capacity for ordered thought, which one might recognize and be able to use deliberatively even if it were not one’s first impulse, I could not help but hear such an assertion in context of 
> https://twitter.com/i/status/1346919171595137025 <https://twitter.com/i/status/1346919171595137025>
> 

On 1/14/21 11:40 AM, Marcus Daniels wrote:
> Unless one believes AI isn’t possible, or that machine learning doesn’t make holistic assessments, this seems pretty ridiculous.  It is all implemented on logic.


-- 
↙↙↙ uǝlƃ