[FRIAM] Graal VM
Marcus Daniels
marcus at snoutfarm.com
Fri Feb 21 11:50:02 EST 2020
Glen writes:
< But the question remains *when* to increase the field. >
With thermal (simulated) annealing, there's a protocol that starts hot and gets cold over time. The inspiration comes from metallurgy where the range and schedule of the temperature sweep has to encompass the phase transitions of the material in question. If you want to see the ice melt, you better sweep above 0 C. It is less clear what the phase transitions of a population's `goodness' function would look like. For a given constituency, it could have many phase transitions and they could be hard to find. Quantum annealing is a similar protocol which is typically to start with all things be both true and false, and then over time (while tunnelling and entanglement occurs) one slowly reduces the field toward definite states.
Both are unsatisfactory because the `goodness' function is constantly evolving. People are born and die, for one thing. Taking an agent-centric approach would be one way to rationalize the literal use of these protocols because people often get more rigid as they age. (Pity the diachronic personality.) Liberals would run hotter and longer and conservatives would quench rapidly.
A related technique is parallel tempering (PT). Here there are many temperatures at once and replicas move between temperatures from time to time, temperatures can even be added to be hotter or colder as needed. PT avoids the possibility of having the optimizer mis-calibrated to the phase transitions of the referent. This would be social engineering, if purposely applied to human populations, but clearly it happens anyway since some people have traumatic violent lives (or just dynamic lives) and for others every day is a lot like the last. Swapping temperatures would be like the move Trading Places.
Another technique is population annealing where many individual solutions compete for survival during the cooling process. Roughly, an evolutionary technique or capitalism without a safety net.
Then there is reverse quantum annealing, where one starts out with some solution and then perturbs it to make it better. The size of the perturbation is simply how big of the field is activated, and for how long. I think this the kind of protocol most people want from governance. Small experiments that may improve things in their neighborhood. Trump and other fascists like to identify some subspace, fix everything not in that subspace, and then erase it with massive uncontrolled perturbations.
It would better to have a bottom-up approach where people would learn to recognize being stuck-in-a-rut and subject themselves to a higher perturbation, rather than appealing to a monster to scramble other people who are already operating at high temperature. That requires emotional intelligence that demagogues don't nurture.
Marcus
________________________________
From: Friam <friam-bounces at redfish.com> on behalf of uǝlƃ ☣ <gepropella at gmail.com>
Sent: Friday, February 21, 2020 8:37 AM
To: FriAM <friam at redfish.com>
Subject: Re: [FRIAM] Graal VM
OK. But the question remains *when* to increase the field. Steve's answer (if I understand it -- maintain a quasi-periodic pulsing field) is *generally* good, but particularly bad. I think one of the issues the Right has with things like Political Correctness has to do with our inability to maintain a high swappability between true and false. Witness those of us who spend a *lot* of time referencing science fiction (particularly authors like Heinlein) or fantasy (particularly like the Lord of the Rings). Our metaphors carry us away into an inability to distinguish what-is from what-if. The "adjacent possible" smacks of that same inability. In may be just dandy for super-intelligent god-people with disposable income, time, and cognitive power, lucky enough to be "paid to think". But normal people have things that need to get done ... like ... NOW. And tomorrow comes with a whole new set of things that need to get done, now.
Inspired by this post: https://www.lesswrong.com/posts/wJutA2czyFg6HbYoW/what-are-trigger-action-plans-taps, what we normal people need are specific and particular *triggers* which, when identified, launch the field/heat increase to re-dimensionalize the discussion. Similarly, I'd argue we need specific and particular triggers to launch a dampening effort to cool down, reduce, the discussion from a high dimensional riot into more gelatinous cohesion.
On 2/20/20 1:11 PM, Marcus Daniels wrote:
> Steve writes:
>
> < "But when and how do we force the system into an anarchist mode
> of exploration?" I don't know the term off the top of my head, but I think there is one
> which fits a similar role to that of annealing (both in materials
> science and computer simulation) where the dimensionality is "pulsed" or
> "phased". >
>
> With quantum annealing, one distinguishes between the energy of the problem (goodness, defined somehow) and the energy of a transverse field which is used to conduct the search for solutions. The two are different axes of angular momentum. When the transverse field is high, proposition are both true and false, when it is low they must be true or false. One can perform this procedure from things that are true or false toward things that are true and false or the vice versa.
>
> The anarchist in this metaphor turns up the x axis amplitude like the Men in Black would activate their Neuralyzers. Classically, one might turn up the temperature to get propositions bouncing between true and false at different rates.
--
☣ uǝlƃ
============================================================
FRIAM Applied Complexity Group listserv
Meets Fridays 9a-11:30 at cafe at St. John's College
to unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com
archives back to 2003: http://friam.471366.n2.nabble.com/
FRIAM-COMIC http://friam-comic.blogspot.com/ by Dr. Strangelove
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20200221/a7defd81/attachment.html>
More information about the Friam
mailing list