Free will
in [Lisp]
|
From: Don Geddis on 28 May 2010 18:10 Raffael Cavallaro <raffaelcavallaro(a)pas.despam.s.il.vous.plait.mac.com> wrote on Fri, 28 May 2010: > 1. a chain of deterministic physical events *does not* equal the sort of > thing traditionally considered free will in western philosophy and theology. > 2. unconscious processes that precede and predict our subjective experience > of choosing, often by several seconds, give the lie to our subjective > experience of choosing. Only if you choose to interpret the data in that way. But that is not the only conceivable interpretation. There are others, just as reasonable, which have the added benefit of maintaining our intuitive sense of choice and responsibility. -- Don _______________________________________________________________________________ Don Geddis http://don.geddis.org/ don(a)geddis.org Our brains live in a dark, quiet, wet place. That is the reality. It is only by means of our senses that we get the illusion of being out there in the world. In a way, our bodies are a form of telepresence, operated by our brains, huddling safe in their little caves of bone. -- Hal Finney
From: Don Geddis on 28 May 2010 18:20 RG <rNOSPAMon(a)flownet.com> wrote on Fri, 28 May 2010: > Among the many positions that Ralph (and Don and Dawkins) seems to > subscribe to that I don't is the idea that if everyone just abandoned > their silly superstitions and turned to science for moral guidance > then we would enter an era of harmonious coexistence with everyone > singing kumbaya around the campfire. For the record, I don't believe that, if everyone dropped faith and "converted" to science, that it would lead to humans having peaceful and harmonious coexistence. I also agree that "seeking Truth" is not always a top goal for everyone. It is certainly not the same as seeking happiness, for example. If a lifelong religious believer is on his deathbed after eight or nine decades, and peacefully looking forward to the afterlife, nothing good comes of trying to convert him to atheism during those last days. -- Don _______________________________________________________________________________ Don Geddis http://don.geddis.org/ don(a)geddis.org There's nothing I like less than bad arguments for a view that I hold dear. -- Daniel Dennett
From: Bob Felts on 28 May 2010 20:31 Don Geddis <don(a)geddis.org> wrote: > wrf3(a)stablecross.com (Bob Felts) wrote on Thu, 27 May 2010: > > Don Geddis <don(a)geddis.org> wrote: > >> Or, by analogy: you can keep patching epicycles upon epicycles in an > >> attempt to save your favorite theory, but at some point you need to > >> realize that planets probably just orbit in ellipses, not circles. > > > > But ellipses are not circles and observation can show the difference > > between ellipses and circles. > > Do you not know what an epicycle is? Are you unfamiliar with the > transition in astronomy between the old Ptolemaic system of circular > orbits centered on the earth, vs. the Copernician revolution of orbits > centered on the sun, and then Kepler's transition to ellipses? > I know what an epicycle is, and I am familiar with the development of orbital calculations. > The important part of the story, which seems to have escaped you, is > that the predictions of the various theories all had about the same > amount of accuracy. It was not observation (at the time) which > distinguished the theories. > I agree. Not about the claim that your point escaped me, but everything concerning the theories. > [...] > So you missed the whole moral of my example. Which is that, sometimes, > experimental data is not sufficient to distinguish between two theories. And you missed the whole moral of my rebuttal. Since we're in comp.lang.lisp, let me use a kludge as the software equivalent of epicycles. None of us like kludges. It offends our sense of esthetics. We much prefer elegant software to kludgy software. But esthetics simply isn't sufficient to determine whether or not a kludge is necessary. As just one example, perhaps the replacement for the kludge changes timing which causes things to break. Or maybe the compiler is broken and the kludge is a necessary workaround. > Nonetheless, science does not come to a halt in those cases. > There are still other methods for ranking theories, besides just data. > Then it's philosophy, not science. One of the main reasons why science has succeeded is its ability to remove the subjective, i.e. our sense of esthetics. That's one reason why appeals to Occam's Razor are so problematic; it soothes our sense of esthetics, but it isn't a _scientific_ principle (in fact, this post claims that there is scientific evidence against it, at least in some domains: http://apperceptual.wordpress.com/2008/06/14/ockhams-razor-is-dull/) [...] > Wave function collapse is completely unmotivated, unsupported by > experiment, and basically unfalsifiable. Every data point can be > accommodated by carefully modifying the claims of the collapse > interpretation. > Just curious, but how is MW falsifiable? [...] > > > Without experiments, it's not science -- it's philosophy. And usually > > bad philosophy at that (e.g. dependence on Occam's Razor, which ends up > > depending on our notions of good and evil). > > What does Occam's Razor have to do with good and evil? They seem > totally orthogonal to me. > Because of it's correspondence with our esthetic sense. > >> > I never said that intelligence was a simple binary property. Are you > >> > intelligent? Is Ron? > >> Are those yes-or-no questions? > > > > For you and Ron, I would have thought yes. > > You're the fourth person I've asked this question, and I'm 4 for 4 on > > not getting an answer. > > (http://stablecross.com/files/category-dialogs.html) > > This is just absolutely fascinating. One would think that you wouldn't > > hesitate a bit in answering "yes" to the question "are you > > intelligent?" > > Well, aside from the obvious fact that you're trying to trick me and > play "gotcha!", I don't have a problem at all. I'm not trying to trick you. I'm not above "gotcha!" if I think logic warrants it. ;-) > Yes, Ron and I are intelligent. > > But you missed the whole point of my response. It wasn't at all > reluctance to answer the question. Do you see what you yourself wrote? > "I never said intelligence was a simple binary property. Are you > intelligent?" > > Does the contradiction not leap out at you? You are asking the very > binary question that you claimed you never say. One can ask binary questions about non-binary properties. I have never claimed that intelligence is a binary property. [...] > >> > At some level, intelligence includes the ability to refuse to submit > >> > to experiments. I look forward to see how science deals with that. > >> Observation in the wild, natural experiments, etc etc etc. > > So if you were going to answer the question "are you, Don Geddis, > > intelligent" what observations or natual experiments or etc. etc. etc. > > would you use? > > I would observe problem solving in novel situations. Watch the entity > fail to solve some goal it has originally, then watch it learn over time > and get better and better at solving the problem in similar situations. > So, are computers intelligent in the same way you are intelligent? If not, what tests would indicate this? > >> (E.g., how does macroeconomics test their theories? National economies > >> also "refuse to submit to experiments".) > > National economies don't have free will. > > True, but you again missed the point completely. Macroeconomics is a > reply to your confusion about "how can science deal with an intelligence > that refuses to submit to experiments". National economies also refuse > to submit to experiments, but macroeconomics is an example of a science > that does just fine with that topic. > > By analogy, having an intelligence that refuses to submit to experiments > is not an absolute barrier to investigation by science. > You, Ron, and I are sitting before three computer terminals. None of us can see another's terminal. All of us use our terminal to conduct a Turing test of whatever the terminals are connected to. Afterward, we meet and Ron and I claim that the "computer" passed. You, however, received no response whatsoever. No matter how often we perform the test, while sometimes Ron or I will get no response, you never get one. Is the "computer" intelligent? Why, or why not? > > 1) There is some true knowledge which is inaccessible to science, or > > No. What ethical system can be used to judge all other ethical systems? [...]
From: Don Geddis on 28 May 2010 23:38 wrf3(a)stablecross.com (Bob Felts) wrote on Fri, 28 May 2010: > Don Geddis <don(a)geddis.org> wrote: >> wrf3(a)stablecross.com (Bob Felts) wrote on Thu, 27 May 2010: >> > Don Geddis <don(a)geddis.org> wrote: >> >> Or, by analogy: you can keep patching epicycles upon epicycles in an >> >> attempt to save your favorite theory, but at some point you need to >> >> realize that planets probably just orbit in ellipses, not circles. >> > But ellipses are not circles and observation can show the difference >> > between ellipses and circles. >> Do you not know what an epicycle is? > I know what an epicycle is, and I am familiar with the development of > orbital calculations. > I agree [...] about [...] everything concerning the theories. Well then it's odd for you to have written "observation can show the difference between ellipses and circles", because the significant part of the story was that observation could not distinguish epicycles from ellipses. So your comment really makes no sense. Yes, eventually, centuries later, other observations were detailed enough to distinguish the theories. But in the time of Copernicus and Kepler, observations were not sufficiently details. Apparently, you would have gone on using Ptolemaic astronomy, because you can't seem to conceive of any reason aside from observational data, to prefer one theory over another. >> Nonetheless, science does not come to a halt in those cases. >> There are still other methods for ranking theories, besides just data. > > Then it's philosophy, not science. Says you. I say that science is broader than you appear to conceive. > One of the main reasons why science has succeeded is its ability to > remove the subjective, i.e. our sense of esthetics. There are objective reasons, besides data, to prefer one theory over another. Just because it isn't based on data, doesn't suddenly make it subjective. Consider this puzzle: for any finite data set, there are ALWAYS and infinite number of theories that exactly predict that particular data set. (Plus a whole bunch more theories, which allow for varying amount of error in the original measurements.) Do you really believe there is absolutely no way to have any (objective) preferences over the infinite set of matching theories? From your perspective, any one would be just as good as any other? I think you would make for a poor scientist. Among other problems, you would fine that the random theory you often chose from the infinite set, usually was immediately falsified by the very next observation. Whereas I, able to use information in addition to observational data, would typically choose scientific theories that just so happen to continue predicting new future observations correctly. At the end of the data, that's what matters, isn't it? One of our procedures "works", in practice ... and one doesn't. > That's one reason why appeals to Occam's Razor are so problematic; it > soothes our sense of esthetics, but it isn't a _scientific_ principle > (in fact, this post claims that there is scientific evidence against > it, at least in some domains: > http://apperceptual.wordpress.com/2008/06/14/ockhams-razor-is-dull/) Occam's Razor is a shorthand phrase for the idea of having a prior bias based on the structure of proposed theories. There are lots of similar metrics (Kolmogorov complexity, minimum description length, shortest Turing machine, etc.), each putting a slightly different ordering on possible theories. The point is not that we know the exact right preferential ordering between any pair of theories. The point instead is that when you have two competing theories, generally one is completely dominated by the other according to all of these metrics. It's rare that it matters exactly which metric you choose to use. > Just curious, but how is MW falsifiable? Multiple Worlds basically says: we know exactly how subatomic particles work. We are not aware of ANY data that shows that ANYTHING different ever happens, no matter how many particles you put together into a big system. Therefore, the most reasonable extrapolation from subatomic scales to human-level scales is ... the exact same thing keeps happening. This is pretty easy to falsify. You just need an experiment that shows that something other than superposition (for example, "wavefunction collapse") definitely happens somewhere, at some scale, in some experiment. Then, Multiple Worlds would be falsified. >> What does Occam's Razor have to do with good and evil? They seem >> totally orthogonal to me. > > Because of it's correspondence with our esthetic sense. You seem to think that Occam's Razor is some simple, "I like it!" subjective feeling, similar to "I like chocolate!" It's far more justified than that. As I say, if we both try to form theories to explain a series of data points, and I use (something like) Occam's Razor to guide my search of theories, and you refuse to use anything except "falsified by the data" ... then I'm going to get to the right answer long, long before you will. We can even bet on it. A black box spits out numbers. Our goal is to predict the next number. Every time we predict correctly, we get paid $10. Run this experiment with many such black boxes, and with you and me using our different procedures of "science", and (in this real universe) I'll wind up with a whole lot more money than you. It's got nothing (directly) to do with "esthetics". Nor, therefore, with good and evil. > >> >> > I never said that intelligence was a simple binary property. Are you >> >> > intelligent? Is Ron? >> >> Are those yes-or-no questions? >> > This is just absolutely fascinating. One would think that you wouldn't >> > hesitate a bit in answering "yes" to the question "are you >> > intelligent?" >> Does the contradiction not leap out at you? You are asking the very >> binary question that you claimed you never say. > One can ask binary questions about non-binary properties. I have never > claimed that intelligence is a binary property. It betrays a lack of understanding of the concept, to ask binary questions about non-binary properties. "Is the moon hot?" "Is a mouse big?" These are not yes-or-no questions. They only make sense in some context, for some purpose. When you ask, "Are you intelligent? Is Ron?" without providing any context at all, the question is either meaningless, or (more likely) you just don't understand that intelligence is a spectrum. Without context, the moon is neither hot nor cold, it simply has a particular (average) temperature. Other entities are colder, and others hotter. A mouse is neither big nor small, but simply a given, measured, size. And similarly, Ron and I are neither intelligent nor not, but simply have a certain level of intelligence. > So, are computers intelligent in the same way you are intelligent? If > not, what tests would indicate this? Yes, they are. Or at least, can be. (Only a few computers have been programmed to exhibit even a small degree of intelligence.) Just not nearly to the same degree as I am. (Orders and orders of magnitude less, in fact.) > You, Ron, and I are sitting before three computer terminals. None of us > can see another's terminal. All of us use our terminal to conduct a > Turing test of whatever the terminals are connected to. Afterward, we > meet and Ron and I claim that the "computer" passed. You, however, > received no response whatsoever. No matter how often we perform the > test, while sometimes Ron or I will get no response, you never get one. > Is the "computer" intelligent? Why, or why not? (All three terminals are connected to a single computer?) Sure. Because we have examples where it demonstrated intelligent behavior (with you and Ron). Since I got no data, I have no information to contribute to the judgement. I'm reminded of Lord Acton's famous quote: Judge talent at its best and character at its worst. Intelligence is something that you judge based on its very best performance, not the average or worst case. >> > 1) There is some true knowledge which is inaccessible to science >> No. > What ethical system can be used to judge all other ethical systems? This looks like a non-sequitur to me. I don't seen any connection between "true knowledge", "science", and "ethics". You seem to be asking a rhetorical question, but I don't even understand the point. -- Don _______________________________________________________________________________ Don Geddis http://don.geddis.org/ don(a)geddis.org A fool does not learn from his mistakes. A normal man does learn from his mistakes. But the exceptional man learns from the mistakes of others.
From: Bob Felts on 29 May 2010 13:43
Don Geddis <don(a)geddis.org> wrote: > wrf3(a)stablecross.com (Bob Felts) wrote on Fri, 28 May 2010: > > Don Geddis <don(a)geddis.org> wrote: > >> wrf3(a)stablecross.com (Bob Felts) wrote on Thu, 27 May 2010: > >> > Don Geddis <don(a)geddis.org> wrote: > >> >> Or, by analogy: you can keep patching epicycles upon epicycles in an > >> >> attempt to save your favorite theory, but at some point you need to > >> >> realize that planets probably just orbit in ellipses, not circles. > >> > But ellipses are not circles and observation can show the difference > >> > between ellipses and circles. > >> Do you not know what an epicycle is? > > I know what an epicycle is, and I am familiar with the development of > > orbital calculations. > > I agree [...] about [...] everything concerning the theories. > > Well then it's odd for you to have written "observation can show the > difference between ellipses and circles", because the significant part > of the story was that observation could not distinguish epicycles from > ellipses. > > So your comment really makes no sense. Yes, eventually, centuries > later, other observations were detailed enough to distinguish the > theories. But in the time of Copernicus and Kepler, observations were > not sufficiently details. > > Apparently, you would have gone on using Ptolemaic astronomy, because > you can't seem to conceive of any reason aside from observational data, > to prefer one theory over another. > I can conceive of a lot of reasons to prefer one theory over another. "Prefer" means "to like one thing _better_ than another"; "better" means "more good", and that's puts it straight in the realm of morality. I'll use my wetware to make a choice from imagination space (what I think ought to be). But if we're going to do _science_, individual preference has to be removed. If I have to compute orbital mechanics, and I have the choice between pages and pages of calculations and a simple formula, I'm going to take the easy way out. But that's just me. I want time to do other things besides expend labor on math. On the other hand, there are people who get their kicks doing such things. But you're confusing computation with description. Both epicycles and ellipses give the same answers, but only one matches the full gamut of what we observe today. Now, you might say, "the simpler calculations eventually turned out to correspond to reality, so we should always choose the simpler." That's not a bad rule of thumb, but it isn't proof that reality always corresponds to the simpler theory. Now, I'm a working man, so I haven't had time to scan the literature to see if this has been done. But suppose you have calculations X and Y where both give the same result. X is computationally simpler (for some definition of simpler) than Y. What you want is proof that the universe works according to X and not Y. Now, I'm not up on quantum computing, but perhaps some metric of "simpler" could be tied to conservation of engergy, which would bias the universe toward that form of the computation. > >> Nonetheless, science does not come to a halt in those cases. There are > >> still other methods for ranking theories, besides just data. > > > > Then it's philosophy, not science. > > Says you. I say that science is broader than you appear to conceive. > No, you don't and I'll show you, _using your own words_, that you don't consider this true. > > One of the main reasons why science has succeeded is its ability to > > remove the subjective, i.e. our sense of esthetics. > > There are objective reasons, besides data, to prefer one theory over > another. Just because it isn't based on data, doesn't suddenly make it > subjective. > > Consider this puzzle: for any finite data set, there are ALWAYS and > infinite number of theories that exactly predict that particular data > set. (Plus a whole bunch more theories, which allow for varying amount > of error in the original measurements.) > > Do you really believe there is absolutely no way to have any (objective) > preferences over the infinite set of matching theories? From your > perspective, any one would be just as good as any other? > There you go, again, injecting "good", which is a property of our imagination, back into the discussion. > I think you would make for a poor scientist. Among other problems, you > would fine that the random theory you often chose from the infinite set, > usually was immediately falsified by the very next observation. > Right. If it was _falsified by observation_ I'd change horses. I've never wavered from this position. In fact I've been the one insisting on it as a means to judge between competing theories. But we haven't been talking about theories that have been falsified by observation. Were talking about theories that have appeal to our intuitive sense of "fitness" -- where "fitness" happens to vary between individuals. > Whereas I, able to use information in addition to observational data, > would typically choose scientific theories that just so happen to > continue predicting new future observations correctly. > Where do QM and MWI differ in their predictions? [...] > > > Just curious, but how is MW falsifiable? > > Multiple Worlds basically says: we know exactly how subatomic particles > work. We are not aware of ANY data that shows that ANYTHING different > ever happens, no matter how many particles you put together into a big > system. Therefore, the most reasonable extrapolation from subatomic > scales to human-level scales is ... the exact same thing keeps > happening. > > This is pretty easy to falsify. You just need an experiment that shows > that something other than superposition (for example, "wavefunction > collapse") definitely happens somewhere, at some scale, in some > experiment. > > Then, Multiple Worlds would be falsified. > Then do the experiment. Until then Copenhagen and MWI return the exact same results. It doesn't matter how "pretty" a theory is until there is experimental evidence that the universe does, in fact, conform to our sense of "prettiness" (or vice versa). > >> What does Occam's Razor have to do with good and evil? They seem > >> totally orthogonal to me. > > > > Because of it's correspondence with our esthetic sense. > > You seem to think that Occam's Razor is some simple, "I like it!" > subjective feeling, similar to "I like chocolate!" > > It's far more justified than that. As I say, if we both try to form > theories to explain a series of data points, and I use (something like) > Occam's Razor to guide my search of theories, and you refuse to use > anything except "falsified by the data" ... then I'm going to get to the > right answer long, long before you will. > Then take a look at the expermients that show when Occam's Razor fails as a guide to theory selection. > We can even bet on it. A black box spits out numbers. Our goal is to > predict the next number. Every time we predict correctly, we get paid > $10. > > Run this experiment with many such black boxes, and with you and me > using our different procedures of "science", and (in this real universe) > I'll wind up with a whole lot more money than you. > We do have such black boxes. How many times have you played the lottery? How many times have you won? ;-) > It's got nothing (directly) to do with "esthetics". Nor, therefore, > with good and evil. > Ok, then why the opposition to my theory that humans are driven by (quantum) randomness and, to test that, to construct an AI using quantun randomness, instead of pseudo-randomness, or some deterministic algorithm? It has nothing to do with ease of calculation; rather, I suspect it's the philosophical implications. > > > > >> >> > I never said that intelligence was a simple binary property. Are you > >> >> > intelligent? Is Ron? > >> >> Are those yes-or-no questions? > >> > This is just absolutely fascinating. One would think that you wouldn't > >> > hesitate a bit in answering "yes" to the question "are you > >> > intelligent?" > >> Does the contradiction not leap out at you? You are asking the very > >> binary question that you claimed you never say. > > One can ask binary questions about non-binary properties. I have never > > claimed that intelligence is a binary property. > > It betrays a lack of understanding of the concept, to ask binary > questions about non-binary properties. "Is the moon hot?" "Is a mouse > big?" These are not yes-or-no questions. They only make sense in some > context, for some purpose. > Sure. But the context is our everyday experience. Every mathematical system is based on "undefined" concepts, and unproven assertions (axioms). Yet we work with these undefined concepts, since we have common shared experience with them. Geometers don't define "line", but we work with them anyway. > When you ask, "Are you intelligent? Is Ron?" without providing any > context at all, the question is either meaningless, or (more likely) you > just don't understand that intelligence is a spectrum. > > Without context, the moon is neither hot nor cold, it simply has a > particular (average) temperature. Other entities are colder, and others > hotter. A mouse is neither big nor small, but simply a given, measured, > size. And similarly, Ron and I are neither intelligent nor not, but > simply have a certain level of intelligence. > How can you have a "level" of something if you don't already have an idea of what that something is? > > So, are computers intelligent in the same way you are intelligent? If > > not, what tests would indicate this? > > Yes, they are. Or at least, can be. (Only a few computers have been > programmed to exhibit even a small degree of intelligence.) > I didn't ask "can be". > Just not nearly to the same degree as I am. (Orders and orders of > magnitude less, in fact.) > Fine. Wonderful. How do you know this? > > You, Ron, and I are sitting before three computer terminals. None of us > > can see another's terminal. All of us use our terminal to conduct a > > Turing test of whatever the terminals are connected to. Afterward, we > > meet and Ron and I claim that the "computer" passed. You, however, > > received no response whatsoever. No matter how often we perform the > > test, while sometimes Ron or I will get no response, you never get one. > > Is the "computer" intelligent? Why, or why not? > > (All three terminals are connected to a single computer?) > Maybe it's a cloud. > Sure. Because we have examples where it demonstrated intelligent > behavior (with you and Ron). Since I got no data, I have no information > to contribute to the judgement. > > I'm reminded of Lord Acton's famous quote: > > Judge talent at its best and character at its worst. > > Intelligence is something that you judge based on its very best > performance, not the average or worst case. > So you're willing to take Ron and my word that there does, in fact, exist an intelligent entity for which you have no direct evidence? > > >> > 1) There is some true knowledge which is inaccessible to science > >> No. > > What ethical system can be used to judge all other ethical systems? > > This looks like a non-sequitur to me. I don't seen any connection > between "true knowledge", "science", and "ethics". > > You seem to be asking a rhetorical question, but I don't even understand > the point. If science is going to discover an objective moral standard then it has to be able to compare between two different moral systems. But the decision whether M1 is better than M2 is itself a moral decision, which requires M3, limited only by the "size" of the imagination. |