Slashdot Mirror
Still More Evidence for Evolution
Uche writes: "Biologists at the University of California, San Diego have uncovered the first genetic evidence that explains how large-scale alterations to body plans were accomplished during the early evolution of animals."
The article only refers to the repressor genes, (i.e. 6 legs instead of 12). But the creature still has to go through the slow process of developing legs itself in order for the gene to have some effect. It also doesn't explain how appendages like for instance wings on that fruit fly came along. They would have to start somewhere and I can't see how wings could be useful in any but their mature form. They wouldn't be needed to slow an insects fall(as they are small enoguh not to be hurt) and I can't see a pair of fans growing the muscle control and speed necessary to flight. What steered the evolution of the fruit flies to lead them to functioning wings?
I stole this Sig
Found an article that nicely describes antibiotic resistance and evolution:
From the FDA Web site The Rise of Antibiotic Resistant Infections:
The increased prevalence of antibiotic resistance is an outcome of evolution. Any population of organisms, bacteria included, naturally includes variants with unusual traits--in this case, the ability to withstand an antibiotic's attack on a microbe. When a person takes an antibiotic, the drug kills the defenseless bacteria, leaving behind--or "selecting," in biological terms--those that can resist it. These renegade bacteria then multiply, increasing their numbers a millionfold in a day, becoming the predominant microorganism.
[Insert pithy quote here]
Belief? I don't believe in evolution -- I wouldn't know how to do such a thing. Belief never comes into it.
The preponderance of the evidence leads me to an obvious conclusion -- changes in individual living things occur from generation to generation. Enough time and changes occur, and you have this thing called evolution. In some ancient businesses, it's just called breeding.
If that evidence wasn't there, I'd conclude differently...but not necessarily that a spirit or deity was the necessary other choice.
A firewall can not protect you from yourself. Turn off what you do not need. Do not use the firewall to do your work.
My friends and I have been batting this one around, maybe you can help. It concerns how one gets from a primordial soup full of replicators (see 'The Selfish Gene', by Richard Dawkins) to something like a cell, way before anything like a regulator gene.
.5%, you go through each of the 62 characters / genes in an agent, roll the dice, and if it comes up .5% or less, you mutate that character.
Every environment can be thought of as presenting a utility function to the organisms that inhabit that environment. Dawkins gives an example of the following utility function:
Try to see if a population of organisms can "discover" the line of poetry "This is the way the world ends, not with a bang but a whimper." You'll note that there are 29 possible choices for each letter (26 letters + commas + periods + spaces). And in the above string, there are a total of 62 characters. So, to present the power of evolutionary theory, Dawkins imagines a population of agents randomly initialized to 62 characters. One of these might be:
"jkdzcn43asdf lkjasdfhaokjshfla ksdhfoiuykjahs, asdasd. sdfsdf."
you can imagine that each agent reproduces unequally based upon how well it does given the utility function -- in this case, the utility function returns an integer from 0 to 62, where 0 indicates no letters match and 62 represents a perfect match for the entire sentence. Each generation is exposed to mutation in the Dawkins example, though one could easily add crossover (which implies sexual reproduction) and inversion. The code is roughly:
1) initialize X agents in a population to random strings of length 62
2) write a function where each agent reproduces unequally based upon how well it optimizes the utility function given above. This choice matters, but not a lot. For our purposes, imagine that every organism below some threshold X has a 10% chance per time period of dying outright. And every organism above this threshold has a 10% chance of replicating.
3) After step 2 (which represents one tick on the clock), expose each organism to genetic operators. Mutation is simple: pick a % chance Y (where Y is small; if it is too large, you lose information too quickly) for each character in an agent (or gene if you prefer) to mutate to a random character. Thus, if Y is equal to
4) repeat steps 2 and 3 until you see equilibration of your population.
After a bit, it should be obvious to you that most of your agents will approach the correct sentence, whatever their starting values. Further, not all of the organisms in a population will ever be at the "right" outcome, given mutation in step 3.
So what does this tell us? Simple math helps out. To optimize the utility function above is simple, and we know this because we can compute the number of steps it would take to optimize it. Couple of points:
1) the function Dawkins uses (outlined above) is separable. No character / gene depends upon any other character / gene to determine the utility of its expression. This is huge. Think about it until you get a smile on your face. For real organisms, this is NOT the case (i.e., genes are non-separable). This is why evaluating the results of the genome project is ugly. If we had, for example, one gene acting alone to determine intelligence, it would be easy to detect / modify. Sadly, multiple genes acting in concert determine intelligence, and modifying one gene in the set changes the value for the entire set.
2) The number of steps needed to optimize the above function is 29 * 62 = 1798, which is an extraordinarily TINY search space.
3) If the characters / genes were non-separable, as they are in real organisms, things are quite different. Worst case is completely non-separable -- i.e., every character depends upon the value of every other character for evaluation under the utility function. In this case, you have 29^62 (where the '^' represents the exponent function). Obviously, this is a freaking HUGE number. Even low levels of non-separability (e.g., pairs of genes that depend upon each other to produce a trait) generate huge search spaces.
The fraud of Dawkins is thus simple. He proposes a set of operators that
define his theory of evolution -- unequal reproduction, crossover, mutation,
and inversion, and illustrates their efficacy (i.e., the "success" of the
theory) on a simple toy problem. The ugliness, however, is that solving
separable problems, which is the class of utility functions Dawkins uses
to "test" his theory, is trivial. Everything / anything works well on them,
and there is no real way for any given theory to fail on this class of
utility functions. The other, more interesting class, which has the
property of being an analog to REAL ORGANISMS WITH REAL GENES is when the
utility functions are non-separable, and the theory / set of operators
Dawkins proposes has NO success searching the spaces induced by this type of
utility function.
It is as if I set up a craps game, you come to play, and the rules are, I
win all double sixes and you win everything else. You commence to roll
double sixes until I have all the money in the world. I assert that the
dice are not loaded.
The dice for complex life are loaded somehow, or we don't understand the
mechanisms of genetics. The existence of these regulator genes simply begs
the question.
None of this, of course, displaces evolution as the best fit for the
available evidence.
Karl
The second derivative of the space-luck curve is infinite at my nexus, at least on the pong axis.