Richard Dawkins speaking on 'The God Delusion'

[EricL]

Richard Dawkins site

Richard Dawkins reads excerpts from The God Delusion and anwsers questions at Randolph-Macon Woman's College in Lynchburg, Virginia on October 23, 2006. This Q&A features many questions from Jerry Falwell's Liberty "University" students. In Richard's tour journal he says:

"Many of the questioners announced themselves as either students or faculty from Liberty, rather than from Randolph Macon which was my host institution. One by one they tried to trip me up, and one by one their failure to do so was applauded by the audience. Finally, I said that my advice to all Liberty students was to resign immediately and apply to a proper university instead. That received thunderous applause, so that I almost began to feel slightly sorry for the Liberty people. Only almost and only slightly, however."

[shvarz]

Thanks for the link!  A bit long, but quite interesting

[Jez]

Ah well, oh wow!   Thanks for the link, that site goes on my favourites list.

To quote; (Chance and natural selection and intelligent design (Part 2 - Q & A - 65+ min))

"One of the biggest fallacies in popular understanding of darwinian evolution by natural selection is that it is a theory of random chance, it is not, it is the very opposite and this is one of the most important things to understand about it. there is a certain chance element in it. Mutation is a process of random chance, it's random with respect to improvement, things don't tend to get better as a result of mutation, the important step in the darwinian theory of evolution is natural selection, natural selection is a non random process, natural selection is the non random survival of randomly varing genetic codes, the reason why some genetic codes survive better than others is their phenotypic effects via the process's of embryogenisis on phenotypes, on bodies, which make them survive or not survive, reproduce or not survive and the ones that do survive and reproduce pass on the genetic coded instructions that built them, and equiped them, and made them good at surviving and reproducing. That's the idea, that is the explanation for the apparent adaptive design, the illusion of design, which all living things show, it is a non random process, it does not involve design of any sort, it produces an illusion of design. It is hard for people to grasp for various reasons and one reason the questioner has pinpointed is the sheer length of time involved."

So DB has it wrong by relying so much on mutation to replicate natural selection?

"Geological time is larger than most human minds are capable of grasping, one of the various metaphors that have been used to convey the sheer magnitude of geological time, one that I like that I didn't invent is that you hold out your hand to represent the length of geological time and if the middle of my tie is the origin of life and the tip of my finger is the present; then the dinosaurs, which went extinct 65 million years ago lived about there (points at wrist) most of this is bacteria (shoulder to wrist) you have multicellular life evolving about here (points a little below wrist) dinosaurs about there, humans at my fingernail, and the whole of recorded human history, everything from the Egyptians in biblical times, the Romans, the Assyrians, the Greeks, all of human history, disapears in the dust from one stroke of a nail file."

MB's are going to perhaps take longer than a few million cycles to evolve then.  

[EricL]

So DB has it wrong by relying so much on mutation to replicate natural selection?

Um, I don't think you understand the difference between mutations - a random process - and natural selection - a non-random process.  Random mutations create diversity within a population.  Selection operates on this diversity.  Mutations are random (although as an aside, there is some recent evidence that many mutations may not be quite so random.  Hot spots for mutations in the genome may be directed by the genome itself.  That is, a genome that is better at coding itself such that mutations are more frequent in places where having more frequent mutations is benificial - such as sequences for coding for specific bio-toxins in poisonious species co-evolving with preditiors which produce anti-toxins - and coding itself such that mutations are less frequent in places where mutations would be deleterious can be said to be fitter than one which has an equal probability of mutation across it's entire genome.  But I digress.)  

Natural selection is non-random in that a non-random subset of the popuatlion survies to reproduce I.e. those which are fittest for their environment.  Thus, the two things go hand in hand, mutations and natural selection.  If there were no mutations, there would be no diversity within a population and selection woudl have nothing to operate upon.   Mutations create diversity and thus that random process is essential to the non-random process of evolution by natural selection.

MB's are going to perhaps take longer than a few million cycles to evolve then.  

More than a few million cycles yes, but not as many as you might think.  DB isn't starting from scratch.  We are bootstrapping evolution by builidng into every bot the potential for various capabilities such as eyes, ties, shots, etc.  We aren't waiting for those things to evolve, we have just decreed 'There exist ties.  There exist shots. There exist eyes.  There exist means to share nrg via ties.  There exist means for bots to cooperate.'  All we have to do is wait for bots to evolve the means to use these mechanisms.  We don't have to wait for the mechanisms themselves to evolve.

My mental model of DB has it set right in the pre-Cambrian period - lots of one-celled animals on the brink of cooperating.  All we need to do is provide the right consistant set of physics and an environement such that selection favors the evolution of mutlibots and I predict we will see our own a Cambrian multibot explosion.

[Numsgil]

 Dawkins is a jerk.  So much wasted energy battling the IDers.

In my experience, natural selection isn't so much survival of the fittest as non survival of the non fittest.  Giraffes didn't have selection for long necks as much as selection against small necks.  In general evolution seems great at breaking things.  Any limitations coded into the genome get removed pretty quickly.  Novelty is orders of magnitude slower to develop.

[EricL]

I have met Richard Dawkins.  He is not a jerk.  He is in fact, very personable and about as unjerk like as I can imagine someone being.  You can argue that beating one's head against the ID wall of ignorance is fruitless, but doing so does not make you a jerk.

I may not see your point regarding non-selection of non-fittest but I think I disagree.  Sometimes selection favors something.  Sometimes it disfavors something.  Both are types of directional selection.  But favoring something is not the same as disfavoring it's opposite.  There are many more ways to not have something than there are to have it.

Contrary to popular beleif, Giraffes long necks are the result of sexual selection, not foraging ability.  Male giraffes use their necks in competition for herds of females.  Males with larger necks tend to fair better in such contests and win and breed.  Those without tend not to.  You can say that is selection *against* short necks but what it really is is selection *for* winning male-male contests.

Your novelty point escapes me completly.  Are you arguing for punctuated equalibrium?

[Numsgil]

Very much so.  I would say 99% of successful mutations I've seen either turn a gene on all the time or turn a gene off all the time.  The number of really neat and new behaviors I've seen are very scarce.  In my experience evolution is very good at destroying non optimal solutions but much less better at rewarding good solutions.

For instance, lets say that a mutation lets you have 10% more successful kids.  That gives you a 10% advantage more or less over another critter.  There's a strong chance that that 10% won't even matter until several generations down the road.  That nice mutation can easily be lost when you die from an accident.  This nice mutation is already extremely rare (the number of useful mutations is very small when compared with neutral or harmful mutations).  Even when it shows up the odds aren't that great that it will make it to the next generation.

On the other hand, a bad mutation that renders you sterile gives you an infinite percentage disadvantage.  Your bad mutation has 0 chance of making it into the next generation.

I'm probably fudging numbers a bit, but yes, in my mind evolution is far better at lancing off unnecessary or harmful things than it is adding new things.  Novelty is a rare and beautiful thing.
----------------------------------------------------------
Saying Dawkins is a jerk is probably too strong.  He's like a kid that pokes the hornet's nest because it's fun.  Naming your book "The God Delusion" isn't going to be mending any fences.  Too much effort is wasted from both the religous right and the intellectual left fighting each other.  In the end it poisons both the religous and scientific wells.  Religions come to despise science, ostracising any potentially controversial free thinkers, and science refuses to give money, time, or attention to anything that might be contrived as having a religous agenda.

If God would just come down and tell us that He uses Lamarckian evolution, everyone would just be better off

[Jez]

Um, I don't think you understand the difference between mutations - a random process - and natural selection - a non-random process.

There is a very good chance I understand a lot less then you realise, I'm just an uneducated person with a big mouth.  
So thank you for taking the time to explain it a little more clearly.
The reason I posted that is because what I see Dbots doing is fighting a generally losing battle against the mutations. 'Intelligently designed' bots devolving into lazy and inefficient parodys of their ancestors. I have to say that the results coming back from the evosimmers are, seemingly, producing better results than they used to but the fact remains; if I try to evolve a top bot that I have written it devolves.

So what am I doing wrong? Is the eviroment not tough enough or to small? Are the mutations set to high? Is the sample size to small? Am I expecting to much?

I like your pre-Cambrian model btw.  

***

When it comes to Dawkins wasting his time butting his head against the big brick wall of religion, good on the guy I say. I used to argue with religious people and used to be able to get them to retreat back to the meta physical standpipe of faith and at that point I just used to give up.
I don't think we should leave religion alone, it has spent it's history fighting scientific advancement, it is, generally, morally wrong, it has caused pain and suffering, wars and death across the ages. (One of the three biggest causes of human death is religion I believe).
One of the greatest things, IMO, to happen in England was the creation of the C of E. The removal of the Catholic church, that organisation that is still known to actively cover up priest pedophilia.
Being Christian does not make you good, it means you can appologise on your death bed for your sins, be forgiven and go to heaven. What other 'get out of jail free card' do you need?

I'm gonna shut up in a minute! My standpoint is that many religions are morally reprehensible and the fact that someone has taken on the 'great blinkers in the sky', the fact that science is, at last, trying to correct these age old self replicating organisations should be lauded not booed.

I appologise for having such a bee in my bonnet about the whole issuse but I believe in science.

P.S. Religion always despised science, science for a long time tried to be religously acceptable. It never worked so I don't see a problem challenging religion, they are hardly going to start hating scientists more than they already do.

[EricL]

I think we are agreeing for the most part.  It is true that deleterious mutations are orders of magnitude more common than benificial ones.  To quote Dawkins, (though he was making a different point) "there are many more ways of being dead then of being alive."  Most mutations result in lower fitness, which tends to kill you relative to those without the mutation, if not in that generation than in the next and thus most mutations don't get passed along to future generations.

But every once in a while, a mutation is incrementially benificial.  It still may not stop that specific individual from dying in an accident or something, but over time, lots of time, probability will rule, benificial mutations will covey increased survivial probability, the mutation will fixate in the population and the species will have evolved.

DB is pushing the envelope with respect to trying to get results by speeding up time.  The mutation rates are too high for one thing (and the populations too low).   If everyone is mutating, complexity will suffer.  It's equivalent to trying to evolve bots inside a nuclear reactor.  It is not surprise that complexity gets destroyed in a high mutation rate environment.  

But personally, I think we are still missing a key evolvability mechanism in the program and that is the means for the mutation rate attached to variable sized sequences of base pairs to itself mutate.  We need a way for the genome to evolve stable sections at various levles of granulatiry which can remain stable even in the face of high mutation rates.  Nature has evovled this mechanism over the eons but we are missing it.  I think it is crucial for letting complexity survive and flourish.
_______

I admire Dawkins for taking a stand on the ID debate.  It may be an age thing.  I've had more time to get frustrated with the willfull ignorance of the masses and witness the exploitation of the faithful by people in power.

[Numsgil]

I think the primary missing function is sexual reproduction.  Eukaryotes, multicellularity, sexual reproduction, and chromosomes all appear at almost the exact same moment in the evolutionary record.

That can't entirely be on accident.  They must all be entertwined in some way.  I think if we implement a working sex system, we'll start to see some improved stability in our bots over time.

I also disagree on the idea that evolution will continously find incremental improvements.  From my experience it tends to be an asymptotic approach to a stable point.  This stable point might be less fit than where it's coming from.  The population dynamics tend to be alot less clear cut than "better" and "worse".  However, if the organism is actually reacting to changing stimuli, this won't hold true anymore.  Then the stable point will be moving, and the organism will constantly be changing.  In this situation, though, "better" and "worse" are impossible to define, since the rules of the game are changing as the game is being played.  If the organisms changes move the stable point of another species, you might be able to create a constantly changing feedback arms race between the two.

However, I haven't actually seen this occur, and I'm assuming that there doesn't exist a mutually stable point for both of them to exist in at the same time.

----------------------------------------------------------------------

I would say equally horrendous attrocities have been perpetrated in the name of science as religion.  Remember the Holocaust and eugenics?  The robber barons defending their positions of power as "Social Darwinism".  Surely these aren't somehow less attrocious than clerical hypocracy and inquisition.

Ultimately people are going to be unethical and willfully ignorant wether they're religous or not.  I've seen plenty of close minded bigots who also happened to be atheist.  They toted science as the penultimate Truth, yet it didn't make them nice people.  People are people.

Religion and science are orthogonal.  Religion allows us to understand and accept our position in life, our relationship with the rest of the universe, and our relationship with the rest of humanity.  Religioin is ultimately a symptom of the greatest gift man has: sentience.  Science seeks to understand the nature of reality, the rules by which we all play.  Science tells us nothing about the reality of human existance, and religion tells us nothing about the workings of the universe.  It's only when people think they do that there's problems.

Traditionally, science and religion have been inseperable intertwined.  The witchdoctors of our tribal past were as much doctors as priests.  It's only been in the last 500 years that the pace of science has exceeded the ability of the staunchier religions to adapt, and this has fueled the majority of problems between the two ever since.

I'd like to see a world where Religion recognizes that it doesn't hold the answers to the universe, only the human awareness, and science recognizes that religion has more validity to the human existance than science ever can.

[EricL]

I think the primary missing function is sexual reproduction....  ...I think if we implement a working sex system, we'll start to see some improved stability in our bots over time.

You may vary well be correct, but not all multi-celled biological organisms reproduce sexually (though most do) so it is not at all clear to me at least that sexual recombination is the 'primary missing function' in enabling multibot evolution or even stable complex single bot organisms.  My own opinion is more along the lines that few sims have been run for long enough with an evironment that would favor singlebot cooperation for complex multibots to evolve.   In fact, we may not have rich enough environmental capabilites in the program to even create a sim that truly favors multibots.  I also think that the reasons why complex behaviour seems to degenerate have their roots in the (too) high rate of mutation and the missing mutation locality functionality I describe above not to mention the fact that hand-coded bots are not well engineered to survive mutations and thus quickly fall off their peak in the fitness landscape, giving the impression of complexity instability.  We have yet to really see any *evolved* complex beahviour, so it is perhaps premature to talk about the stability of such.

Don't get me wrong, I would love to see sexual reproduction emerge.  It's not clear to me we should build it into the system.  As has been discussed in other threads, there are in fact arguements not to do so, not the least of which is that it would subsume aspects of recombination from chromosome structure to recombination granulatiry into the simulator, not in the bot DNA.  I'm not necessarily opposed to this, as it may be one of those things we need to bootstrap.

I also disagree on the idea that evolution will continously find incremental improvements.  From my experience it tends to be an asymptotic approach to a stable point.  This stable point might be less fit than where it's coming from.  The population dynamics tend to be alot less clear cut than "better" and "worse".  However, if the organism is actually reacting to changing stimuli, this won't hold true anymore.  Then the stable point will be moving, and the organism will constantly be changing.  In this situation, though, "better" and "worse" are impossible to define, since the rules of the game are changing as the game is being played.  If the organisms changes move the stable point of another species, you might be able to create a constantly changing feedback arms race between the two.

I am not claiming that evolution continiously finds incremental improvements since the term 'improvement' implies some sort of external frame of reference which does not exist.  I am claiming however, that evolution does favor incremental improvements in fitness realtive to an environment.  That environement may be a 200 degree hot spring where being simple and tough is the definition of fitness - I am not at all claiming that macro multi-celled organisms are always fitter or favored.  Quite the opposite in fact.  Multibots will not emerge until selection favors their emergence.

I fully agree that in a changing environment, selection is always chasing a moving target and that in relatively static envionments, long periods of 'stasis' do occur, but selection favors the prepared genome when conditions do change.  The worse mistake a genome can make is not to mutate.

I would say equally horrendous attrocities have been perpetrated in the name of science as religion.  Remember the Holocaust and eugenics?  The robber barons defending their positions of power as "Social Darwinism".  Surely these aren't somehow less attrocious than clerical hypocracy and inquisition.

There is strong evidence that Hitler was a praticing Catholic and the Holocaust was at least partially religiously motivated, but arguing where fault lies in historical attrocities is not very useful IMHO.  Let us agree that whackos use whatever means they can to build and hold power.  Religion is a powerful means but not the only one.

Ultimately people are going to be unethical and willfully ignorant wether they're religous or not.  I've seen plenty of close minded bigots who also happened to be atheist.  They toted science as the penultimate Truth, yet it didn't make them nice people.  People are people.

No argument here.

Religion and science are orthogonal.  Religion allows us to understand and accept our position in life, our relationship with the rest of the universe, and our relationship with the rest of humanity.  Religioin is ultimately a symptom of the greatest gift man has: sentience.  Science seeks to understand the nature of reality, the rules by which we all play.  Science tells us nothing about the reality of human existance, and religion tells us nothing about the workings of the universe.  It's only when people think they do that there's problems.

Here I strongly disagree.  If your religion includes faith in supernatural things for which there is no evidence (and often for which there is evidence against) if it encourages you to accept easy answers of explantion along the lines of "because god wanted it that way" rather than seek out the actual causes, then it is entirely at odds with science.

Traditionally, science and religion have been inseperable intertwined.  The witchdoctors of our tribal past were as much doctors as priests.  It's only been in the last 500 years that the pace of science has exceeded the ability of the staunchier religions to adapt, and this has fueled the majority of problems between the two ever since.

Agreed.  Religion provided answers to omportant questions.  Where do we come from?  Why do things happen the way they do?  500 years ago it made sense to accept supernatural answers to those questioins.  There were no alternatives.    It does not however, make sense today.  There are better answers to those questions and all the other questions religion purported to answer, scientific answers meaning there is verifable evidence to support them.  If you are educated, beleiving in supernatural explanations for the natural world is inexcusable today.  Note that I did not say being religious is inexcusable.  My beef is with the supernatural part.  If your religion accepts the big bang, the age of the earth, evolution via natural selection, discounts supernatural explanations for happenings be they today or 2000 years ago and mostly provides you a social network and personal framework for interpeting the world, then more power to you.  I might not call it religion anymore though....

[Numsgil]

I have a very narrow definition of religion, which I think everyone in general would be better off defining it as such.  Religion to me is specifically about coming to terms with ourselves and our place in our community.  The creation myths, the rules, the rituals, the belief systems, are all very pointless from outside this frame of reference.  Any of these religous adornments that do not create personal growth or bind us to the community are superflous.

I suppose in the end I'm pretty much a Humanist.  Religions have value because they provide a sevice for the people in them.

[Numsgil]

Now that I think of it, most of the interesting adaptations I've seen, in fact maybe all of the adaptations I've seen, were directly caused by a predator/prey relationship.

For instance, veggies learning to wiggle randomly to throw off my enitor comesum.  Bots learning to be cannis.  etc.

Maybe the idea incremental development needs to include a caveat about interacting, mutually evolving populations.  A single species in a fairly static environment seems extremely limited.

[Jez]

That's an interesting point Nums, predator prey relationships causing the more positive mutations.

Also the addition of shapes to an enviroment soemtimes leading to the circular behaviour that has been seen.

Perhaps it would be safe for me to say that in an enviroment that provides no challenge to a bot the negative mutations aren't 'punished' in the same way and that natural selection is therefore more likely to take a back seat.

Perhaps many of the DB sims are too friendly to the bots to actually promote serious evolutionary pressure!

**

I agree with Eric on pretty much everything he has said about religion and he has said it much better than I might have.
I have nothing against belief per se, it is the religous structure and its imposition on others that I have problems with, in the same way that I am vehemently against anyone using the scientific label to commit attrocities such as forced eugenics.

[EricL]

I discoverred quite by accident that as of today (December 27th, 2006) Dawkin's 'The God Delusion' is the #4 best seller on Amazon (#1 is a diet book, #2 is on dog training and #3 is Barack Obama's political manefesto).   I found it surprising and I must say, somewhat encouraging to see a non-fiction book dealing primarily with evolution so high on the list right after Christmas (although no doubt it's title gave many buyers reasons for purchase that probably had little to do with scientific interest).

There are an amazing 305 (305!) reviews at Amazon, predictably bunched into polarized groups either heaping praise or scornful attack.  Most are positive.  There is also a lively discussion forum.  Browsing the posts is nothing if not entertaining, most are predictable as you would expect though a few such as the one following are eye opening.  


Reviewer: T SANTOSO (Surabaya, Jatim Indonesia) - See all my reviews
     
I live in a country where you have to have religion. You somehow can't choose not to have a religion. We only have major religions, and sects are not allowed. I belive that this book will not be alowed to be published here.