Yes, both Eric C. and Marcus have already answered this better than what I am about to say, because they have already abstracted it into concepts. But I will put only a particular.
I got this from the polymath Elwyn Berlekamp (who did run a hedge fund) in the kitchen on an Erdos-like visit by him to Santa Fe. The conceptually phrased problem is: what is the deterministic response (“a practice”) to a distributional input as object? Suppose you are playing a betting game, and you know the set of possible outcomes, and the probabilities of each. (Can be poker, though that is complicated enough to take work to describe. A simple double-or-nothing roulette would be a minimal model). You have some money. You can put some of your money as a bet for as many outcomes as you want. Bets have some smallest denomination, so it is possible for you to lose everything for some allocations. How do you allocate? 1. If you want to maximize your expected payoff in one play, you put all your money on whichever outcome is most likely. 2. If you wish to play repeatedly, and you want to both avoid going broke and to maximize your long-term payout, you bet a fraction of your money proportional to the probability of each outcome (or as close as you can get to that with your finite denomination), in every round. I think one of the Bernoullis solved this in 1700-something, and it is widely taught. I was just raised in a woodshed, so I didn’t know about it until Elwyn came by. Your expected rate of growth of your pot is proportional to the relative entropy between your distribution of bets and the distribution of probabilities of the various outcomes. I’m pretty sure this is in Cover and Thomas’s Elements of Information Theory. If your starting pot is finite, your probability of survival to round-N is some other function, now of both distributions and how finely you can divide the money, and there is some distribution of likely survival times that can also be computed. (Remember that there is no way you can be assured to survive; if you make bad distributions and lose more than you gain in too many rounds, you will eventually have too little to be able to place nonzero bets on every outcome with your nonzero denomination, and will start to run nonzero risks of losing all.) Ole Peters has built a whole privately-funded institute on this specific metaphor, which he packages in terms of expected utility: http://lml.org.uk/people/ole-peters-2/ <http://lml.org.uk/people/ole-peters-2/> But of course endlessly more sophisticated variants are developed in almost any field I can think of. So, for elections? Maybe something about investment in state and local organizing, how much attention the national committee gives to on-the-ground organizations and what they say their constituents are asking for, capture of congressional seats, a strategic view of what happens if legislatures won’t make laws so that court interpretations become the de facto origins of legal-precedent-as-new-law, feeding back to which elections are consequential. Stuff like that. Of course, for all of those things, more is always better. But on a budget, how you prioritize the time/attention/work you can get out of citizens to support your projects should depend on what the mechanics of the process is through time. Eric > On Apr 19, 2020, at 5:16 AM, <[email protected]> > <[email protected]> wrote: > > But Eric, > > If, over his career, Nate's site gives a 2/3 vs 1/3 split 1,000 ti mes, and > something near 333 times the 1/3 split wins, I think he gets to declare > himself accurate > > How is that practicial? I.e., how can we base a practice on it? Nate’s > career isn’t over yet? > > Nick > Nicholas Thompson > Emeritus Professor of Ethology and Psychology > Clark University > [email protected] <mailto:[email protected]> > https://wordpress.clarku.edu/nthompson/ > <https://wordpress.clarku.edu/nthompson/> > > > From: Friam <[email protected]> On Behalf Of Eric Charles > Sent: Saturday, April 18, 2020 1:59 PM > To: The Friday Morning Applied Complexity Coffee Group <[email protected]> > Subject: Re: [FRIAM] anthropological observations > > If, over his career, Nate's site gives a 2/3 vs 1/3 split 1,000 ti mes, and > something near 333 times the 1/3 split wins, I think he gets to declare > himself accurate. > > Similarly, a modern poker pro isn't trying to guess what the opponent has. > The modern player is trying to put the opponent on a spread of possible hands > under the circumstances. The outcome of any given hand doesn't matter, and > there is an expected amount of variance in performance even under a > game-theoretic perfect strategy. The question is whether the strategy pays > out in the long run, and whether the player has a deep enough bankroll (in > comparison to the stakes they are playing at) to ride out the variance. If > you think the pro is doing something else, you probably are still a very long > way from getting to that level. > > > > ----------- > Eric P. Charles, Ph.D. > Department of Justice - Personnel Psychologist > American University - Adjunct Instructor > > > On Sat, Apr 18, 2020 at 2:32 PM <[email protected] > <mailto:[email protected]>> wrote: >> So, Eric [Charles], >> >> What exactly were the practicial consequences of declaring that Hillary was >> “probably” going to win the election or that a full house was probably going >> to win the pot given she lost and the dealer held a strait flush? >> >> Nick >> >> Nicholas Thompson >> Emeritus Professor of Ethology and Psychology >> Clark University >> [email protected] <mailto:[email protected]> >> https://wordpress.clarku.edu/nthompson/ >> <https://wordpress.clarku.edu/nthompson/> >> >> >> From: Friam <[email protected] <mailto:[email protected]>> >> On Behalf Of Eric Charles >> Sent: Saturday, April 18, 2020 12:06 PM >> To: The Friday Morning Applied Complexity Coffee Group <[email protected] >> <mailto:[email protected]>> >> Subject: Re: [FRIAM] anthropological observations >> >> -------- Nick says --------- Nate constantly says that making such >> predictions is, strictly speaking, not his job. As long as what happens >> falls within the error of his prediction, he feels justified in having made >> it. He will say things like, "actually we were right." I would prefer him >> to say, "Actually we were wrong, but I would make the same prediction under >> the same circumstances the next time.” In other words, the right procedure >> produced, on this occasion, a wrong result. ----------------- >> >> Well... so this connects a lot with poker, which I am in the process of >> teaching the 10 year old... If I recall, Nate was giving Trump a 1/3 chance >> of victory, which was much higher than most of the other models at the time. >> You can hardly fault someone because something happened that they said would >> happen 2/3 of the time. >> >> If a poker player has a model that predicts a given play to be the best >> option, because it will work 2/3 of the time, and this one time it doesn't >> work, that isn't grounds to say the model failed. >> >> YOU want the modelers to have models that rarely give anything close to >> even odds. So do I, so I'm sympathetic. But the modeler might prefer a more >> honest model, that includes more uncertainty, for a wide variety of reasons. >> >> ----------- >> Eric P. Charles, Ph.D. >> Department of Justice - Personnel Psychologist >> American University - Adjunct Instructor >> >> >> On Sat, Apr 18, 2020 at 12:17 PM uǝlƃ ☣ <[email protected] >> <mailto:[email protected]>> wrote: >>> I think it's interesting that you seemed to have *flipped* your thinking >>> within the same post. You restate my point about conceptual metaphors by >>> saying models/computation merely *justifies* decisions/rhetoric. Then a few >>> paragraphs later, you suggest that's conflating language with thought. >>> >>> My diatribe to Nick was that he *uses* metaphors/models simply to impute >>> his conceptual structure onto Nate. Nick's decision is already made and he >>> wants Nate's work to justify it. And the way he *imputes* his conceptual >>> structure into Nate's work is through the sloppy use of metaphor. Then when >>> Nate tells Nick (indirectly) that Nick's wrong about what Nate's done, Nick >>> rejects Nate's objection. >>> >>> I'm picking on Nick, of course. We all do it. I wish we all did it much >>> less. >>> >>> On 4/18/20 6:14 AM, Steven A Smith wrote: >>> > But frankly as often as not, I saw >>> > them use our work to *justify* the decision they had already made or >>> > were leaning heavily toward, *apparently* based on larger strategic >>> > biases. >>> > >>> > [...] >>> > >>> > As for your gut-level (and often well articulated) mistrust of >>> > "metaphorical thinking", you may conflate a belief (such as mine) that >>> > language is metaphorical at it's base with being a "metaphorical >>> > thinker". Metaphor gets a bad rap/rep perhaps because of the >>> > "metaphorical license" often taken in creative arts (albeit for a >>> > different and possibly higher purpose). >>> >>> -- >>> ☣ uǝlƃ >>> >>> .-. .- -. -.. --- -- -..-. -.. --- - ... -..-. .- -. -.. -..-. -.. .- ... >>> .... . ... >>> FRIAM Applied Complexity Group listserv >>> Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam >>> <http://bit.ly/virtualfriam> >>> unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com >>> <http://redfish.com/mailman/listinfo/friam_redfish.com> >>> archives: http://friam.471366.n2.nabble.com/ >>> FRIAM-COMIC <http://friam.471366.n2.nabble.com/FRIAM-COMIC> >>> http://friam-comic.blogspot.com/ <http://friam-comic.blogspot.com/> >> .-. .- -. -.. --- -- -..-. -.. --- - ... -..-. .- -. -.. -..-. -.. .- ... >> .... . ... >> FRIAM Applied Complexity Group listserv >> Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam >> <http://bit.ly/virtualfriam> >> unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com >> <http://redfish.com/mailman/listinfo/friam_redfish.com> >> archives: http://friam.471366.n2.nabble.com/ >> <http://friam.471366.n2.nabble.com/> >> FRIAM-COMIC http://friam-comic.blogspot.com/ >> <http://friam-comic.blogspot.com/>.-. .- -. -.. --- -- -..-. -.. --- - ... >> -..-. .- -. -.. -..-. -.. .- ... .... . ... > FRIAM Applied Complexity Group listserv > Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam > unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com > archives: http://friam.471366.n2.nabble.com/ > FRIAM-COMIC http://friam-comic.blogspot.com/
.-. .- -. -.. --- -- -..-. -.. --- - ... -..-. .- -. -.. -..-. -.. .- ... .... . ... FRIAM Applied Complexity Group listserv Zoom Fridays 9:30a-12p Mtn GMT-6 bit.ly/virtualfriam unsubscribe http://redfish.com/mailman/listinfo/friam_redfish.com archives: http://friam.471366.n2.nabble.com/ FRIAM-COMIC http://friam-comic.blogspot.com/
