My MemesJohn Newbury 5 October 2009 |
Home > My Memes
Some of my memes (thoughts and beliefs)
I am specifically interested in robust, self-aware, self-improving systems for representing and learning patterns, concepts, analogies, protocols and procedures. I am not specifically interested in the human brain, except that it holds one solution to the problem. (Birds hold a solution to flight, but flapping wings are not the best way to make a plane.) I am more interested in symbolic than sub-symbolic systems, being potentially better suited to my (and the systems') reasoning about the system itself essential for conscious thought, which, I guess, is my ultimate interest. Meta concepts, are surely vital.
I believe that a concept needs to be represented by a fuzzy "cloud" of symbols (and perhaps sub-symbols), many with similar interpretations, representing nuances of meaning; that is, concepts are not, and cannot be properly represented by, point objects: they need a "shape" that can be deformed according to context. Certainly, conventional systems of (point-like) symbols are evidently inadequate by themselves: they are too brittle, and rigorous logic is generally impractical. Systems of pure sub-symbols; for example, neural nets, tend to be too opaque. In my opinion, other AI and non-AI techniques (such as theorem proving, neural nets, number crunching or databases) would be useful tools to be incorporated where appropriate; for example, for performance, but will not provide the best foundations for what I want to see in AI.
My long-term project (that I call MEME) is to develop an AI engine to acquire, invent, organise and interpret a potentially unlimited collection of arbitrary memes. Still on the foundations. In the process I have been extracting a library of general C++ tools.
As a digression, I am currently exploring the possibility of producing an AI bot to play the game of Diplomacy. I had considered the possibility and many of the issues involved several years ago. Now I find that the donkey work of producing a good framework for this has recently been developed, leaving mainly the juicy development of AI players to be done. Sounds fun! (See Diplomacy AI Centre, where I am registered as johnnewbury1.)
The quadrillionth decimal digit of pi could be odd or even. When we calculate it there will be a bifurcation of the universe analogous to a quantum measurement. Discuss!
[Watch this space for something more earth-shattering in due course.]
Science is a "belief system". Discuss!
[Watch this space for something more earth-shattering in due course.]
MAD lives up to its name. Never sensible!
To be at all effective, each side must believe that the other is irrational enough to "push the button", triggering mutually assured destruction, if sufficiently provoked; certainly if threatened with destruction, or even just a take over. But, hopefully, not irrational enough to push the button just for the sake of it, of course! Nevertheless, to be fully effective, it is important to actually approach total irrationality, otherwise your chink of rationality may be discovered, so your bluff may be called. (E.g., command structure and training, at all levels, must be such that a full strike will, in fact, actually be launched if ordered, and leaders must be such that they will actually command it when their declared policy demands it.) So, in fact, the button may well by pushed, even when silly (and it always would be silly!), and, in due course, just for the sake of it!
If player A were perfectly irrational (i.e., random), there would be a finite chance of him appearing to be rational to opponent B in any finite period, such that an B would call his bluff, hence causing mutual destruction, because, in fact, A would not actually tend to behave rationally and avoid such a silly resolution. As the period during which a MAD strategy is used approaches infinity, the probability of not having mutual destruction approaches zero. Hence MAD is not of value in the long run. (Indeed, it approaches certainty that it will produce the worse outcome possible for any MAD user; -infinity utility, as no recovery is possible except, in a sense, by future evolution afresh.) QED!
(It does not help if players play the opposite of what is rational instead of
merely being random. In that case, instead of normally not pressing the button
every day, they would normally press it every day! Of course they need to be a
bit rational, otherwise each day it it would be a random choice, giving humanity
a half-life of two days!)
Indeed, MAD cannot be value even in an arbitrarily short period. Was it ever really better to be dead than Red? Surely:
u = g*p[g] f*p[f],
where u is expected utility; g is gain; f is loss; p[x]
is probability of x.
g is finite, f is infinity (worse possible outcome; impossible to recover);
probabilities are in range [0,1]. So u = -infinity. QED again!
Yet we, on average, still vote for policies that include MAD? How come we
call ourselves homo sapiens (wise man)?
(NB: It may seem obvious that MAD cannot be sensible, since it depends on
irrationality; perfect irrationality at that and, as shown above, even
that is insufficient. But it can be of value to use MAD of a part, for
example only for some individuals of a group, provided not the whole gene or
meme-pool. But nuclear MAD strategy threatens the whole human gene and meme
pool and much else besides that we may value. Maybe not everything we may
value, for example, if we value cockroaches, which are apparently very
resistance to radiation. So maybe the utility of MAD is not quite -infinity
it just excludes any component value that we may believe is due to man!)
It is often said that at least MAD has kept the peace for over 50 years, so has proved itself worthwhile. Not so, I believe, even ignoring the multitude of minor wars and other conflicts; though I agree that it has been, and probably always would be, effective in preventing take-over of any nuclear power by another power. By the Weak Anthropic Principle, of course we, as observers, were not annihilated. But it could well be (most likely I believe) that, in a very important sense, most of our "existence" was destroyed since the beginning of our MADness, in the sense that most world-lines, or most universes in the multiverse, have no mankind in our time, even when they did in the past. If anything can be deduced, perhaps it is that, because (on on what we call our world-line) there have been no limited nuclear wars since the introduction of MAD, where and when there is a nuclear war it tends to lead to total destruction of the "intelligent" population on that planet. (Maybe, but not necessarily, also on any other planets that are at least part owned by the powers concerned.)
The partial loss of existence of mankind is the same as I believe happens to you if you survive anything that was likely to have killed you if anything had gone wrong, such as overtaking on a blind corner. It is not sensible to say that you have often overtaken on blind corners in the last 50 years when late, and thereby always kept your appointments, so the strategy must have been a good one; one that should be continued. You should not think that you (the whole you) luckily got away with it. If it was, indeed, dangerous, that is, it did, in fact, have a high probably of causing death, suffering and mayhem, to yourself and others, then that did, actually, occur to most counterparts of yourself and others on other world-lines that were indistinguishable from (but evidently not identical to) yours earlier. There is total symmetry, in the sense that there is no way to say that one world-line is more real, say, than any other. (Most people on each world-line argue that theirs is the real one!) Nor can you say that the real you is in any specific world-line: you do not know exactly what the future holds for you, nor what the far past held. That is, you do not and cannot know which world-line you are on, so it is only meaningful to say you are on them all! (The asymmetry in timescale is probably due to the second law of thermodynamics, that is, to the generally low entropy of the past.) All other world-lines were, are and will be just as real as what you call this world-line (or reality). The reality, stupidity, death and suffering caused by the act is just as real in those unfortunate world-lines as the sense of relief is in other, fortunate, world-lines, such as what you (in this world-line) call this one, which just happens, by definition by the Weak Anthropic Principle to be OK. The risky, and hence morally reprehensible, act was exactly as bad in all the world-lines, since in all of them you (including your counterparts) deliberately took the same risk. (The only reason that we cannot always adequately punish such a surviving culprit is that we cannot calculate the risk as accurately as the culprit.) You would hate to be in their world lines; survivors in theirs would love to be in yours! But most of even these currently fortunate world-lines will not be OK if you do such a thing again which you probably will do (that is, will do in most world-lines) since you have experienced no negative feedback (in any currently fortunate world-line)! (In that sense, only, "it is written!") It was just as much a bad idea in the past as such will be in the future or anywhere else in physical, or or any other, space.
Note that world-lines (or a multiverse) are required for any meaningful interpretation of probability. They are not just a requirement of quantum mechanics; they are also required even for a deterministic Newtonian world. Even in the latter case, there is a continuum of possibilities at any point in space and time, past present or future. As above, there is total symmetry between all the possibilities; it is not possible to meaningfully say which one is the real one. The people in the alternatives do not think, "Shucks, we are not real!" They are just as happy and sad, an so on as we are. OK, some of them think silly things like we do, such as, "Cogito ergo sum," and refuse to accept that anyone else is necessarily as real as they are. In fact, all attempts to say which is real are obviously just pointless, meaningless, playing with words.
There is obviously an uncountable number of "universes of discourse", which, when wide enough, we sometimes abbreviate to "the universe". But "the Universe" is, by definition, unique, and contains everything that there is, even things yet to be discovered or forgotten. It is embodied in "Uni". From what I have said above, it must also include not only everything that was, is or will be, but everything that could be (since probabilistic entities exist in the Universe, along with past present and future entities). It is therefore synonymous with "the Multiverse", so the latter is redundant and better avoided. Indeed, that term is even confusing and illogical, since it is trying to say that it contains multiple Universes, which is meaningless. (It is OK to say that the Multiverse contains multiple universes [of discourse], but the Universe already includes that idea.) Perhaps it might be useful to use "a multiverse", to mean "a set of universes [of discourse]", but it would be clearer to use that phrase explicitly (especially as "multiverse" is often used inconsistently or meaninglessly), and even such a set is just another universe [of discourse] (of wider scope).
Therefore, where important, I prefer to use the term "world-lines" or, when not wishing to emphasise the coherent (roughly speaking, causal or time) element, "worlds", since "the multiverse", or even "the universe", are often not used to mean just "the multiverse [or universe] that we are discussing", but mistakenly or confusingly, "the Multiverse", that is, "the Universe".
Some memes are unhelpful, to say the least. Below are some that I have now (finally!) totally and forever expunged from my mind, but in some cases only after years of fierce competition with incompatible memes. (I never did believe some of these things, I hasten to add, but I am too embarrassed to say which were which.) Most are downright, silly, misleading or even downright harmful, but nevertheless widely believed. All is clear to me now my memetic stress is minimised! But some of these memes are very tenacious (especially some in the religion collection, even to an agnostic, which I was until mid-thirties) and so all-embracing that they can screw up a mind's evaluation of itself and other memes (Keep an open mind, of course, but not so open that your brains fall out!) I would be happy to expand and defend my views.
READ ON AT YOUR OWN RISK!
MANY OF THESE IDEAS ARE VERY CATCHING; ONCE CONTEMPLATED THEY CAN BE DIFFICULT
TO CURE!
DO NOT BELIEVE ANY OF THEM FOR A MOMENT!
I totally reject the following ideas and all that follows from them (and recommend that you do too):
ReligionI mainly read popular science books and Web articles. I will not attempt to list them all, but here are some that profoundly influenced me. In many cases they are just samples of those I own by given authors, and with whose views I tend to concur.
Gφdel, Escher Bach (Douglas R. Hofstadter)
This book seems too amazing to exist! Not present in the vast Library of Babel, but I have a copy. Sod the Bible and Shakespeare on my desert island!
The Selfish Gene (Richard Dawkins)
The foundation for understanding Life. Even my (non-technical) mother understood and enjoyed it. The meme and word, meme, was coined here, and, as you can see, it has certainly infected me. If atheists needed a pope, it would, of course, be Richard Dawkins.
Computers and Thought (collected essays)
Dated now, but was my introduction to AI.
Consciousness Explained (Daniel C. Dennett)
Addresses the ultimate challenge to anyone (like me) as they try to convince themselves that they really believe that their own mind experiences are just states of a machine.
How the Mind Works (Steven Pinker)
Not precise enough to program one, of course!
Chaos (James Gleick)
Well organised!
Game, Set and Math (Ian Stewart)
A winner!
Synch (Steven Strogatz) and Six Degrees (Duncan J Watts)
I usually read two or more books contemporaneously, partly for convenience of location around the house, and sometimes as relief from a worthy but difficult tome. But I have also found that this often leads to a marvellous synergy of often unrelated ideas from the several books. (One then also tends to confuse who said what, but ideas are more important than who espoused them.) It was most weird when I happened to intersperse reading the above two books. Unexpectedly, I discovered that Watts had been a student of Strogatz; part of each book describing their experience of the other. Small world! (which was also a phenomenon covered in both books).
The Emperor's New Mind (Roger Penrose)
Penrose really knows quantum mechanics and explains it very well to the informed layman. Just ignore the rubbish about microtubules and quantum mechanics having anything to do with intelligent thought, let alone endowering the human brain with fundamental abilities beyond any machine.
Where is Everybody (Steven Webb)
A balanced book, but which overall reinforces my long-held view that there almost certainly are no other civilisations as advanced as ours in the galaxy. (I also believe that no significant inter-galactic interaction is likely. On consideration, it is not plausible that "they" are already here but unobserved, or only observed by UFO freaks, US military, or similar.) Drake's equation is a truism, but is of no value, other than to highlight (with each term) the many relevant factors we are almost totally ignorant about. Defensible values of its terms can lead to almost any probability. The proof-of-the-pudding is that "they" are not here, yet in about 100000 years, say, we (or our cybernetic descendents) will probably have colonised the galaxy (or died off), yet we should expect some of "them" to be millions (maybe billions) of years more advanced than us. My belief is not for naive chauvinist reasons. Furthermore, I think many of those who believe in ETs do so (at least subconsciously) for quasi-religious reasons: hoping that there is someone brighter that us to sort out our messes. No! While we cannot even estimate the truth, we should play safe and face up to our responsibilities: assume that if we kill the earth we kill the galaxy.