Mt. Improbable?

There's an interesting discussion going on between Larry Moran and Richard Dawkins. The subject is the title of Dawkins' 1996 book Climbing Mt. Improbable. It started with this post over at Larry's blog. He included Dawkins in his list of good science writers who were nonetheless excluded from Dawkins' recent anthology of science writing. Along the way, Moran offered this thought:

Dawkins is also a master of metaphor but, sometimes the metaphors are misleading and can give an incorrect view of evolution (e.g. Climbing Mt. Improbable).

Personally, I loved the metaphor of Mt. Improbable. We shall return to this point momentarily.

Dawkins' website reposted Larry's blog entry. Dawkins himself turned up in the comments to offer this thought:

I am interested in the suggestion that Climbing Mount Improbable might not be an ideal title. Interested, because I regard it as the most under-rated of my books. It sells FAR fewer copies than The Blind Watchmaker although I think it is a better book. Perhaps the reason is that the title is not so good. It covers some of the same ground as The Blind Watchmaker, but -- even though it is not really for me to say -- I like to think the two chapters called 'The Museum of all Shells' and 'Kaleidoscopic Embryos' are genuinely novel and original, where The Blind Watchmaker is mostly popularizing stuff that is already well known to professionals. Well, as I say, it is not really for me to judge. But my own opinion is that, if anybody is thinking of reading The Blind Watchmaker, they might do better to read Climbing Mount Improbable instead. Of course, I wouldn't want to STOP anyone reading BOTH!

I found this interesting, since I preferred Climbing Mt. Improbable (CMI) to The Blind Watchmaker (TBW).

My introduction to Dawkins' writing came during my graduate school years at Dartmouth College. The student newspaper published a very enthusiastic (to put it kindly) op-ed defending young-Earth creationism. Since I was feeling a bit burned out on mathematics at that time, I used the opportunity to start educating myself on the subject of evolution and creationism. Of course, at that time it never occurred to me that this topic would occupy any significant part of my life.

Anyway, at that time I knew of only two evolutionary biologists. One was Charles Darwin. The other was Stephen Jay Gould, who I took to be the voice of establishment science. It never even occurred to me that anything he said about evolution was contrversial. I began my program of self-education with several of his essay collections. I enjoyed them immensely, but decided nonetheless that it was time to get someone else's view of the matter. So I went to the public library, located the three shelves or so of evolution books, and noticed a rather attractive volume called The Blind Watchmaker by Richard Dawkins, of whom I had never heard.

I was disappointed with the book on my first reading. The main reason was the subtitle: “Why the Evidence of Evolution Reveals A Universe Without Design.” The book, you see, presents very little in the way of evidence for evolution. Really it's main purpose is to dispel certain common misconceptions of evolution. This it does very well, but it was not what the subtitle had led me to expect.

Dawkins devotes the early chapters of the book to explaining how natural selection builds up complex structures by accumulating small, random variations. (No doubt everyone will recall his Weasel experiment and the biomorphs.) I felt he was belaboring the obvious. Was anyone really confused on these points?

Today, a decade later, after reading countless inane criticisms of these chapters not just from creationists but even a few from actual scientists, I realize that Dawkins had good reason to include this material. I can't tell you how many creationists I've met at various conferences who seem genuinely mystified about how evolution is said to work. I have experienced the frustration of trying to explain what I regard as perfectly obvious points to people who, in perfect sincerity, find them baffling. Mind you, I was not trying to convince them of the truth of evolution. I was just trying to get them to understand what was being claimed.

So my opinion of TBW is higher today than it was after that first reading (though I still hate that subtitle). That aside, I loved CMI. It provided exactly the sort of thing I was looking for. You see, at that time I was sympathetic to the creationist argument that complex structures could not evolve gradually. I understood Dawkins' Weasel argument, and his related arguments about his biomorphs, but that was all very abstract. In principle it was a viable argument, but could it really be applied to actual complex structures?

CMI went a long way to convincing me that it could.

And I really loved that metaphor! It perfectly captured the idea that natural selection changes the probabilities of everything. (Indeed, a failure to understand this basic point has foiled nearly every creationist mathematical argument in history, from the misapplication of basic combinatorics found in YEC volumes, to the murkier adumbrations of William Dembski.) Sure, if you have to get to the top of the mountain in one step there is little hope of getting there. But if you can find a gradually sloping path to the top, you could, indeed, get there step by step.

In one simple metaphor it cut to the heart of what I believe is the major stumbling block for people who find evolution hard to accept.

On the other hand, maybe this is just my mathematical bias showing. Responding to Dawkins' comment, Larry offers some criticisms of the metaphor:

To begin with, you use the Mt. Improbable image as a metaphor for evolution. This is misleading since evolution encompasses more than just adaptation. It would be difficult to apply the “Climbing Mt. Improbable” metaphor to the organization of our genome, for example, since it's clearly not well-designed and could never be characterized as the peak of an adaptive landscape.

I don't have the book handy at the moment, and it's been some time since I read it. But this is not how I remember the metaphor at all. Mt. Improbable was not a metaphor for all of evolution, but simply for the part of evolution that explains how complex structures evolve gradually over time. The top of the mountain was not the peak of an adaptive landscape, but simply some particular complex structure (an eye, a wing) whose origin we are retroactively trying to explain. I could be remembering this wrong, but I think that was what Dawkins had in mind.

But even as a metaphor for adaptation the image is less than perfect. Most readers will see the peak of Mt. Improbable as a goal of adaptation, implying that evolution somehow recognizes that there is an ultimate perfection that all organisms seek to achieve by reaching the summit. As you well know (I hope) there are very few (any?) species that are perfectly adapted to their environment. If this were true, adaptation would cease because the species resides on the summit of Mt. Improbable.

Once again, I didn't get this at all from what Dawkins said. The peak of Mt. Improbable is not a goal of adaptation, just some complex structure whose origin we are trying to understand. Vertebrate eyes are mainfestly imperfect, but they are nonetheless the sort of thing Dawkins placed at the top of Mt. Improbable. Notions of ultimate perfection or of species trying to attain anything (even metaphorically) had nothing to do with Dawkins' metaphor, as I understand it.

At any rate, Moran goes on a bit longer, but I will stop here. For now, I still think Mt. Improbable is a pretty nifty metaphor. And by “pretty nifty” I mean that I wish I had thought of it.

Tags

More like this

A few posts back I engaged a discussion about Richard Dawkins' metaphor of “Cimbing Mount Improbable” for the proces by which evolution by natural selection crafts complex structures from simpler precursors. Since I did not have his book in front of me at the time, I was working from memory. As it…
A number of years ago I attended an ID conference near Kansas City. One of the breakout sessions featured a fellow from the Heritage Foundation (an ultra-right-wing political think tank) presenting a lecture about probability and evolution. His talk was mostly just a watered-down version of…
Pulitzer Prize winning novelist Marilynne Robinson wrote this lengthy review of The God Delusion for Harpers Magazine. She was unimpressed. The review weighs in at 4599 words, but you'll find yourself almost a thousand words in before hitting anything substantive about Dawkins' book. Prior to…
When I got interested in evolution, one of the first books I read was The Blind Watchmaker, by Richard Dawkins. I had never heard of Dawkins before reading that book. I read it simply because I happened to notice it a the public library and thought it had a cool cover. The book's third chapter is…

An explanation of any event or occurrence of mysterious origin always involves choosing from a hierarchy of improbabilities.

The Genetic Algorithm is an excellent optimization algorithm, and in that sense the metaphor of a mountain is perfect.

Reiterating the possibly obvious: 'random' algorithms are so powerful, in fact, that the solution of many of the hardest problems are possibly only by random algorithms.

Looking for the solution of a hard problem is the very opposite of what creationists imagine the process of Evolution would allow.

Whether it's a good metaphor seems to depend on whether you see the mountain as a peak of excellence, or merely something seemingly unatainable.

I enjoyed CMI, but most of the ideas were ones I was familiar with from other sources. TBW the same: a good read, but preaching to the converted. Of the books of Dawkins' that I have read, the two which have had the most impact on me were The Selfish Gene, and The Extended Phenotype. Perhaps because I came to them fairly fresh - I was 15 or 16 when I read The Selfish Gene - the ideas were new and startling. They offerred new paradigms for considering the workings of genetics and evolution, and everything made a lot more sense afterwards.

John Landon has looked at the hierarchy and picked the next to last of the improbabilities therein.

Right, the mountain is not a metric of fitness in an adaptive landscape, it's more like a metric of probability -- like Ramachandran plots for structure and thermodynamic probability.

I agree with Dan. Genetic algorithms see fitness as a landscape, possibly including many hills or mountains. Assuming a static landscape (which isn't true in biology), they climb a mountain, improbably rising to a great height, but they're not guaranteed to climb the highest one. (No Free Lunch, you know.)

By Pseudonym (not verified) on 23 Jul 2008 #permalink

Yeah, the Eonic effect is some grade-a crankery. Not quite as insane as the Time Cube, but absolutely vapid in it's description of an ordered march of history.

I went and saw the spam - it might help to drop the creationist "Darwinist" and "Darwinism". Is it supposed to be scientific, or is it some new-age crankery?

Other than that, I liked the "Mt Improbable" metaphor, even if I don't think I read that book.

Landon chose a form of teleology that sees natural laws as being intrinsically purposeful in situations where purpose seems required, without the need for a creator that can intentionally manipulate these laws for its own (and possibly intermittent) purposes.
It would appear he was motivated to find some middle ground between accident and design. He hasn't found any solid enough to stand on, however.

Mt. Improbable was not a metaphor for all of evolution, but simply for the part of evolution that explains how complex structures evolve gradually over time.

Actually, it's not a metaphor for the structures, it's a metaphor for the *probability* of the structures evolving gradually over time, right?

I found Landon's criticism to be a poor attempt at rescuing Gould's Wonderful Life argument from banality.

I left my substantive comments on Dr. Moran's referenced post.

Im a big fan of Dawkins' writing, and I like the Mt Improbable metaphor. It explains certain things very well, like how evolution is a slow gradual process instead of a big jump. But I can sort of see where Larry is coming from.

I think it was in the God Delusion where he mentioned the Mt metaphor again, and also spoke about how there was no such thing as half a wing - the structure would have had a use the whole way through its evolution (fall arrest, then gliding, then eventually proper flight). In my eyes, the metaphor doesnt work so well when you look at it this way. where is the peak? at the fully formed wing or along the way somewhere?
If each stage is useful, it may be more useful to look it as a neverending climb, with a landing on each step where the species rests at each generation..

Interesting post, Jason. I liked Mt. Improbably better than BW as well.

By John Farrell (not verified) on 24 Jul 2008 #permalink

er...I meant Improbable. Improbably is whether Manny Ramirez lets his 'sore knee' keep him from playing this weekend against the Yankees.

:)

By John Farrell (not verified) on 24 Jul 2008 #permalink

I also found "Mt. Improbable" that to be a confusing analogy, when the argument is that there is nothing at all improbable about ending up with an improvement when there is selection. The argument that the particular improvement is improbable is meaningless until a more probable solution is demonstrated.

Jason writes,

The peak of Mt. Improbable is not a goal of adaptation, just some complex structure whose origin we are trying to understand. Vertebrate eyes are mainfestly imperfect, but they are nonetheless the sort of thing Dawkins placed at the top of Mt. Improbable. Notions of ultimate perfection or of species trying to attain anything (even metaphorically) had nothing to do with Dawkins' metaphor, as I understand it.

I don't agree. Think of enzymes and proteins as an example of the sort of evolution Dawkins is addressing. There are billions of ways of combining amino acids to make a protein that's the size of hemoglobin. As Dawkins points out on page 78 the Hemoglobin number is 10^190.

All of these combinations are equally improbable. If you were to represent this as a three-dimensional view of probabilities then it would be a large flat plateau rising above the surface. Most of these combinations are completely useless, however, so we want to depict the actual hemoglobin structure in such a three-dimensional image. To do this we adopt the "adaptive landscape" of Sewell Wright as Dawkins explains on page 135. That's how we get the peak of Mount Improbable. It represents a particular improbable event--the one that produces the perfect hemoglobin.

Dawkins describes this on page 77 when he says, "The height of Mount Improbable stands for the combination of perfection and improbability that is epitomized in eyes and enzyme molecules ...."

In case anyone misses the point, he emphasizes it two pages later when he says,

Eyes, ears and hearts, the wing of a vulture, the web of a spider, these all impress us by their obvious perfection of engineering no matter where we see them: we don't need to have them presented to us in their natural surrounding to see that they are good for some purpose and that, if their parts were rearranged or altered in almost any way, they would be worse. They have "improbable perfection" written all over them. An engineer can recognize them as the kind of thing that he would design, if called upon to solve a particular problem.

The peak of an adaptive landscape is the ultimate goal of natural selection. It is the perfect solution. Yes, it is a highly improbable state but it is the particular highly improbable state that characterizes the best solution to the problem.

Earlier in your posting you say, "The top of the mountain was not the peak of an adaptive landscape ..." but that's not what Dawkins says. The metaphor definitely conveys the impression that evolution strives to reach the peak of perfection--and usually succeeds.

Larry,

What's the fuss? Evolution is a process, and Climbing is a process. It happens that the latter is a good metaphor for the former. Metaphor doesn't mean 1:1 correspondence! Metaphor is an aid, but you cannot reason through metaphor alone.

Explain, please, as if you were telling it to a mathematician, what the problem with the metaphor is.

Thanks.

I don't recall the title of the essay, but I'm fairly sure that Garrett Hardin developed a very similar metaphor in the '60s or '70s.

By Pierce R. Butler (not verified) on 24 Jul 2008 #permalink

Larry says: "The peak of an adaptive landscape is the ultimate goal of natural selection. It is the perfect solution."
But natural selection doesn't have a goal, and especially is not a calculative "problem solving" process. The process is adaptive, reactive, one of avoidance rather than of seeking.
There may be a perfect avoidance or a perfect adaptation, but perfection per se was not the goal.

Drs. Moran and Rosenhouse can tell me where I've screwed up the following.

It seems quite simple to me why "climbing Mt. Improbable" is misleading as a metaphor for evolution, and it's capsulized for me by this quote from Jason's post:

Sure, if you have to get to the top of the mountain in one step there is little hope of getting there. But if you can find a gradually sloping path to the top, you could, indeed, get there step by step.

The activity Jason describes - mountain climbing - is goal-directed, but evolution is not. The products of evolution, ISTM, are more like the location of the Mississippi River shoreline - ever-changing, in large part due to random, contingent inputs, but obeying laws.

That is also IMO a prime reason why so much of the probability math flung around by creationists of the YE or ID varieties is garbage. So what if the current conformation of the Mississippi River shoreline is one among billions of possibilities? The improbability of any specific conformation is meaningless unless one contends the river was somehow attempting to arrive at that conformation as a particular goal. The same with evolution. The fact that a particular genotype is one among endless possibilities can't be considered a violation of probabilities unless you posit that particular genotype as a goal - the "peak" of the mountain. There are no peaks in the sense of goals (though of course there are in the sense of fitness landscapes), and the products of evolution, though certainly contingent, ought not thereby to be considered improbable.

Jud:

The activity Jason describes - mountain climbing - is goal-directed, but evolution is not.

That's only true if you try to anthropomorphize evolution.

If I define an algorithm for you:

1. You wander randomly, but prefer steps that locally lead you up, rather than down.

2. Repeat.

we can agree that this algorithm will lead you upwards, possibly to the top. Is it goal oriented? Does that even matter? It is, however analogous to evolution.

(NB: An ensemble average should of course be considered. And after an appropriate change of coordinates, recombination and a whole host of complicated mutations can also be seen this way.)

Improbability assessments apply to the question of whether evolution was in fact goal directed. So they have or had their place in determining that it isn't. Because at bottom the creator theories represent the epitome of improbability.

Larry Moran -

Perhaps we can get Richard Dawkins himself to clarify things for us!

I don't believe the quotations you provide support your argument. In each case Dawkins is pointing to some complex structure that exists in the present (hemoglobin in the first quote, vulture wings and so forth in the second) and is observing that their combination of functionality and complexity requires a special sort of explanation. It is these structures that are at the peaks of Dawkins' metaphorical mountains, precisely as I said before.

You can put this in the language of Sewall Wright and adaptive landscapes if you want to, but I don't think there's is any need to be so fancy to understand Dawkins' point. I also don't see how you're getting any more general statements about what evolution is striving to accomplish, or usually accomplishes, from the quotations you provide. His metaphor captures an important aspect of the evolutionary process, an aspect that many people find difficult to comprehend. I see no reason to read anything more into it than that.

I know you have disagreements with Dawkins about the relative importance of various evolutionary mechanisms, or about the usefulness of adaptationist thinking, and things like that. But I see those disputes as ancillary to the point Dawkins was making with his metaphor.

As long as you agree that complex adaptations like vertebrate eyes evolve gradually over long periods of time, and that one effect of natural selection is to make it possible to produce structures that would be far too improbable to explain by blind chance, I don't see why you should have a problem with his metaphor.

Jud -

I am not arguing that evolution is a goal-directed process. The issue here is how we explain specific complex structures that exist in the present.

The general sort of Darwinian explanation is that these structures evolve gradually over long periods of time under the auspices of natural explanation. That sort of explanation is viable only if there really is a plausible series of evolutionary intermediates to get us to the finished structure. To use a variation of Dawkins' analogy, we can get to the top of the Washington Monument because there is a staircase that breaks up its awesome height into smaller more manageable steps. Take that staircase away and we would have no way of getting to the top of the monument.

If we could show that there simply was no gradual trajectory to get us to the top of the mountain, and that the only possible naturalistic explanation for the vertebrate eye (for example) was some massive macromutation that brought it into being in one fell swoop, then I would not object to arguing against such a thing on probabilistic grounds. The ID folks claim (wrongly) to have shown precisely that. So I don't think it's inherently wrong to use probabilistic language in discussing these things. It's just that the creationist calculations are invariably based on the assumption that there is no gradual path for evolution to traverse.

Why does the progress of evolution involve taking a path up a mountain any more than it does an obstacle course on relatively level ground?

Why does the progress of evolution involve taking a path up a mountain any more than it does an obstacle course on relatively level ground?

It doesn't necessarily. All you need is a direction. But there are good reasons why the gravitational potential on a mountain or in a valley is a very useful analogy for the local behaviour of any fitness landscape. (Think Taylor expansions.)

We are all familiar with the Earth's tendency to 'equilibrate' us toward the ground...

An obstacle course could work too, but it's more complicated. I wonder if a pinball machine metaphor would work.

It seems to me that using the mountain metaphor implies that there's a search on for perfection, or at least a higher order, when it may be only ourselves who assume we are at the apex of some such natural hierarchy.

I can't lay my hand on my copy of CMI right now (I might have loaned it to my brother), but my recollection of the metaphor differs subtly to both Jason's and Larry's.

Jason,
I think the metaphor was indeed a metaphor for evolution, or more specifically a metaphor for the way a species changes over time through mutation and natural selection. All evolved features are characterized as peaks of an adaptive landscape, with some of the most visible being eyes and wings.

There are limitations in the metaphor (the changing shape of the landscape is difficult to portray, predator-prey arms races are a stretch, and it doesn't really apply to sexual selection), but I think it was intended to apply (and does so very well)

Larry,
The peaks in Dawkins's metaphor are not necessarily *perfect* solutions - they are local peaks. I recall that Dawkins stresses that a species can never backtrack - that a species at a local peak (such as the organization of the genome) can not climb a neighboring higher peak if it means climbing down even a little way.

To put it another way, natural selection can not "see" neighboring peaks. Natural selection can only see as far as a step or two beyond the current species gene pool, and will always select the upward steps, even if a downward step might be better in the longer term.

The metaphor might be easier to apply in reverse - to think of water descending rather than a climber ascending.
Water runs downhill, but doesn't have the lowest ground as a goal - it will stop in a pothole rather than climb out to reach a lake.

By Peter Byrnes (not verified) on 24 Jul 2008 #permalink

http://darwiniana.com/2008/07/24/more-on-mt-improbable/
I am not surprised at the incomprehension of the eonic effect at a site such as Evolutionblog. First point, Darwinists, yes I insist on the term, have been lowballing so long around Darwin, Dawkins and natural selection, etc, that they have lost all perspective on what evolution is, at least for the descent of humans.
The point is that we can closely observe mans evolution in a macro sense by looking at a very late phase of it, in its overlay with world history and the emergence of civilization. And it has nothing to do with natural selection climbing Mt. Improbable. There is an explicit evolutionary directionality at work, which suddenly stands out if we apply careful analysis using periodization.
Forget fitness landscapes and all of that. To grasp evolution you must have closely tracked data, and a method that looks at the full spectrum of human culture, in all its aspects. Then you will see a dramatically coherent pattern stand out, one that can only be called evolution.
Darwinism makes people lazy and careless, because it offers an oversimplification that appeals to a mass public. Do the work on world history indicated, and the result will become clear.
You can cheer for Dawkins all you please, but you will never get evolution straight using the current paradigm.

Landon takes history which evolves in apparent patterns as evidence of some evolutionary imperative in nature when instead this imperative or purposefulness can easily be explained by noting that once life was formed, an element of purposefulness was formed as integral to that life's function. Humans of course have had a sophistication of purpose that has affected, or effected, their own history, and patterns stemming from that ability are much more readily discernible. Such patterns are relevant to the theory of evolution only as they show the complexities of adaptation that are relative to the complex natures of the various species.

They don't reflect any evolutionary purpose that conflicts with the mechanism of natural selection as posited by Darwin.

I've just read several pages of Landon's eonic-effect website, and tried quite seriously to understand what it's about. I can't say I disagree with it, because the text is to me so very low on content. It's one of those situations where the speaker goes on and on, and yet nothing becomes clearer. Either the text describes something so deep that I can't even scratch the surface, or it's something very banal dressed up in lengthy prose. Perhaps it's just a sales site to get you to buy the book.

baboo's explanation seems to make the most sense (congrats on parsing that from the website!), in which case 'eonic theory' falls in the banal category.

Dan, as I recall I parsed some of that from a favorable review of the book, obviously a plant, that inadvertently gave some of the game away.

Dan finds the material on the eonic effect 'low on content'. Hardly! Keep in mind these web materials are intro pieces for the text of World History And The Eonic Effect, and pull their punches before the punchline.
Baboo's remarks miss the point, I think. I talk very little about purposefulness in general, and demonstrate that directionality is present in world history, no more, no less.
The wrong application of Darwinism to human evolution and history results from the imaginary potential of NS to 'climb Mt. Improbable'.

Directionality is goal oriented purpose. So the central question here is whether human goals are an extension of evolutionary goals, or a product of an otherwise "goal-less" mechanism.

We can't ascribe teleology to a process in time since we must be at the end of the process (and maybe out of time??) to estabish what it's 'telos' is, if any.
But we can detect 'directionalith' looking backward. Directionality is a relative transformation that shows changes in direction. That's the most we can detect,with respect to the past.
The interaction of human agency and a macrohistorical directionality is a tricky issue, carefully dealt with in the model of the eonic effect.
Check it out.

Landon,
If you don't concede that your directionality implies a teleological process, then you have demonstrated nothing that contradicts Darwinism, either in its original or modern versions.

And to say that no such process is implicit in your theories because you can't know the "telos" by looking backwards is just silliness.

It's like saying evolutionary theories that don't allow for a goal are wrong because it's clear that there's some sort of a goal being sought even though we don't have any way of knowing what it is.

John Landon:

Keep in mind these web materials are intro pieces for the text of World History And The Eonic Effect, and pull their punches before the punchline.

Ah, so it is there to sell the book.

Honestly, you can't expect people to understand and believe your thesis if it require buying a book just to get some of the main points.

And to say that no such process is implicit in your theories because you can't know the "telos" by looking backwards is just silliness.

It's like saying evolutionary theories that don't allow for a goal are wrong because it's clear that there's some sort of a goal being sought even though we don't have any way of knowing what it is.