At 04:58 PM 3/7/2007, Michael Poole wrote: >Randomness is not identical to noise. Stochastic computing methods >use randomness to get "good" results (according to the method's >definition of good) -- in many cases, much faster than naive methods >reach comparable results.
Gad, this is irritating. I gave examples of where randomness is useful as described, and "noise" is, quite clearly, not identical to "random." However, random number generators can use electrical noise as a seed, because some kinds of noise are random. Others aren't. Technically, noise means an undesired "signal," though the term "signal to noise ratio" is used. If my radio receiver is hissing at me because of thermal electrons, that hissing is pure noise, if it is garbled because of other radio signals interfering with the one I'm trying to receive, that is also noise. I'm using noise to mean data that is used by an election method that is not related to the goal of the election method, it is not designed to produce good results. If you want to distribute property by lottery, then the random number generation used is not noise, but it may come from a "noise source." However, if you want to maximize collective wealth by rewarding successful enterprise, using lotteries can be noise. Depends on what you are trying to do. So when I'm claiming that using random distribution of "victory" in an election method is introducing "noise," I am claiming that, in general, random distribution does not further the underlying goals of election methods, which is not to "distribute victory fairly" so that every neuron feels that it has "won" a fair number of times. Uh, every person, that is. Imagine programming your computer so that every analysis element in it gets its fair share of 'victory.' Do you think this would improve results? Depends on what you want, doesn't it? Poole writes as if I have denied the existence of stochastic methods, as if using random numbers can never be a process for improving efficiency. That's what is irritating, because I never claimed that, nor do I believe it. > > How evenly benefit is to be spread is a choice. How is this > choice to be made? > >This is closely related to the question I asked but that you did not >answer: How do you define the proper outcome of an election? If you need to know it, the voters deliberate the question and vote on the questions that arise in the process. That is one solution. Another is that you have a dictator who defines it. You could define it as being proper if the outcome meets some "election criterion," such as the Majority Criterion. Problem is that I don't think that satisfying the Majority Criterion is the reason we hold elections. We hold elections because we imagine that we have a better society because of it, and it is the better society we desire, not the Majority Criterion as such. The Majority Criterion *looks* like something that would be desirable in a good society. It turns out that this is inaccurate. It's *often* true, but it is also true, apparently, that satisfying the Criterion can be suboptimal, even seriously suboptimal, and that all members of society would agree as to this. That, by the way, is another way of defining the goal of elections. If everyone agrees that outcome A is better than outcome B, after the fact, and particularly after there is time for consequences to unravel, then we would say that the goal of the election is to select outcome A. Or we could simply define outcome A by majority rule. However, I think that most of us would agree that supermajority definition would be much more desirable, if attainable. And I do think that it is possible. Without deliberation, many people would consider the Majority Criterion as highly desirable. But show them real-world examples where the Criterion *should* fail -- they would agree that it failed -- their opinion would change. It would change because the Criterion was not their goal, it was what they understood as a mark of the goal, a measure of the goal. >[...] > >You have written a great many paragraphs here that only repeat the >same assertion that I questioned: namely, that elections using >stochastic methods or history introduce noise. I thought I was rather >clear in saying that you have not defined "noise" in this context. And I thought that I've been reasonably clear in defining it. It's actually quite simple. Noise is a signal that is unrelated to the goal. If you want to "fairly distribute victory," then whatever randomness you use, if you do it that way, is not noise. But if your goal is, for example, to estimate the pennies in a jar, randomly distributing "victory" -- i.e., which individual's guess is used -- would be. The general deviation of the results would be far greater than by amalgamating and using the amalgamated result. Simple averaging might be the best method, but I certainly would not claim that there is no superior method. Discarding outliers might, for example, improve accuracy. To really know, one would need to experiment. > > Now, in such a situation, you could imagine that there is a mushroom > > "faction" and a pepperoni faction and a tomato faction. In reality > > these are not factions, they are simply *votes* or individual > > preferences. > >Do you have a better word for groups whose votes are equivalent under >the election system? "Faction" implies more than being one who voted with another. It implies a much stronger association. But you could use the word technically to simply refer to a group who voted in an election identically. The error would be in assuming that therefore these voters would vote identically in another election. Even if the same question were asked, they might vote differently. My comment assumed, I'd have thought, that "faction" *was* a usable term -- after all, I used it -- but that there was a discrimination to be made, because of other meanings of the term. This is getting tedious and I really don't have time for it. I've got two small children who have to go off to preschool at a fixed time in the morning.... Mr. Poole has not convinced me, and apparently I have not convinced him. I can speak for myself and state that my goal is not to convince anyone, but to examine issues, expose what I know about it (and thus likewise what I do not know), and to benefit from the response of others, as well as to develop information and analysis useful for social action. I'm not here for the pure theoretical love of election methods. I'm here because I was sent here by people who thought it might be useful. Presumably both for me and for certain others. With the organizational concept, with political implications, that I'm promoting, I'm not seeking to convince everyone, nor even a majority. I'm looking for a few people who understand the concepts, and I've been finding them here and there. That's what's needed now. Later, much further down the road, there will be more general outreach. I consider that my work here has been successful, and I want to thank all that have contributed to this success. In some cases this includes "critics." And in another sense, it includes *all* critics, even the "idiots." Sometimes explaining things -- or attempting to explain things -- to an "idiot" is an efficient way to get clear about it, whether or not the "idiot" ever gets it. We all have our roles to play. ---- election-methods mailing list - see http://electorama.com/em for list info
