Imitating biology

[rossum]

Evolution is a hill-climbing algorithm, so it will tend to stop at the top of the nearest hill.

Evolution is not a single algorithm. It is a massively parallel series of near-copies of the same algorithm with some random input. The human evolutionary algorithm has 7 billion parallel copies running. Bacterial algorithms have trillions of parallel copies. With all those parallel algorithms and some random input, more than one hill will be climbed.

The landscape is changing as well; what was once a hilltop may now be a valley and vice versa. Passenger pigeons had parasites: fleas, tapeworms and the like. That was once a nice large hilltop when there were immense population of the pigeons. That hilltop disappeared.

Evolution works in populations, so it has multiple starting points. Hence it can climb multiple hilltops. Some humans have climbed the "digest milk as an adult" hill while others have not, for example.

[TheLurch]

Evolution is not a single algorithm. It is a massively parallel series of near-copies of the same algorithm with some random input. The human evolutionary algorithm has 7 billion parallel copies running. Bacterial algorithms have trillions of parallel copies. With all those parallel algorithms and some random input, more than one hill will be climbed.

I think that's the fundamental gap here. Lee views evolution as random stumbling in the dark. People who actually understand evolution realize it's a massively parallel problem solving approach that, thanks to the randomness, explores solutions that designers would never consider.

It's not clear how to bridge that gap when Lee's worldview clearly demands he not appreciate the power of evolution.

[rogue06]

I think that's the fundamental gap here. Lee views evolution as random stumbling in the dark. People who actually understand evolution realize it's a massively parallel problem solving approach that, thanks to the randomness, explores solutions that designers would never consider.

It's not clear how to bridge that gap when Lee's worldview clearly demands he not appreciate the power of evolution.

Like so many who reject the evidence for evolution Lee is among those who focuses on the randomness aspect while determinedly ignoring the selection part.

[Leonhard]

Common design patterns, would be the reasoning here.

That only explains commonality, it does not explain why those patterns show ancestry the way you'd expect if evolution is true. So this explanation is weaker than the one from the Theory of Evolution, only the latter can explain the bio-geographical distribution of features without begging the question.

[Leonhard]

Evolution is a hill-climbing algorithm, so it will tend to stop at the top of the nearest hill.

Source: Towards Data Science

One of the main problems faced by evolutionary algorithms is the presence of local optima in the fitness landscape. Local optima, can, in fact, mislead our algorithm to not reach our desired global maxima in favour of a less optimal solution.

Source

© Copyright Original Source

Now he goes on to describe ways to alleviate this problem, but it's still a problem.

And I don't claim that we seek only to imitate global optimums, I only claim that biological designs are remarkably good, better even than humans have been able to do, in various cases.

One of the simplest explanations is that the "fitness landscape" continually changes, which follows the evidence we see that most evolution occurs during climate changes.

Blessings,
Lee

You do realize that your post already has a signature line, your username and an avatar attached right? Why do you insist on writing these posts as if they're letters? It's silly.

[lee_merrill]

That's an evolutionary algorithm. it's not evolution. Try again.

That's evolutionary algorithms in general, which would include evolution itself. But for a quote about evolution per se, how about this:

Source: Wikipedia

An evolving population typically climbs uphill in the fitness landscape, by a series of small genetic changes, until -- in the infinite time limit -- a local optimum is reached.

Source

© Copyright Original Source

Hill? Are you referring to an energy hill.

No, a fitness peak, a hill in the fitness landscape.

The human evolutionary algorithm has 7 billion parallel copies running. Bacterial algorithms have trillions of parallel copies. With all those parallel algorithms and some random input, more than one hill will be climbed.

The landscape is changing as well; what was once a hilltop may now be a valley and vice versa. Passenger pigeons had parasites: fleas, tapeworms and the like. That was once a nice large hilltop when there were immense population of the pigeons. That hilltop disappeared.

Evolution works in populations, so it has multiple starting points. Hence it can climb multiple hilltops. Some humans have climbed the "digest milk as an adult" hill while others have not, for example.

All this, however, does not change the fact that evolution tends to find local optimums. Massively parallel, yes, and multiple hilltops, but still evolution will climb to a local optimum as a rule.

One of the simplest explanations is that the "fitness landscape" continually changes, which follows the evidence we see that most evolution occurs during climate changes.

Well, if the fitness landscape continually changes, that shortens the time available for searching considerably.

Blessings,
Lee

P.S. I realize I have a signature too, I just keep signing my posts partly to remind myself of why I am here, to try and do some good.

[lee_merrill]

Like so many who reject the evidence for evolution Lee is among those who focuses on the randomness aspect while determinedly ignoring the selection part.

Well, no, evolution couldn't find an optimum without selection.

Blessings,
Lee

[shunyadragon]

[QUOTE=lee_merrill;772479]That's evolutionary algorithms in general, which would include evolution itself. But for a quote about evolution per se, how about this:

Source: Wikipedia

An evolving population typically climbs uphill in the fitness landscape, by a series of small genetic changes, until -- in the infinite time limit -- a local optimum is reached.

Source

© Copyright Original Source

In the history of life there are millions fitness hills local, regional and continental in history with differnt species, subspecies, varieties and related species..

No, a fitness peak, a hill in the fitness landscape.

As objectively demonstrated there are millions of interrelated fitness hills.

All this, however, does not change the fact that evolution tends to find local optimums. Massively parallel, yes, and multiple hilltops, but still evolution will climb to a local optimum as a rule.

There are millions of opitimums, as the history of life on earth they are not isolated, but involved in a world in interelated regional and continental multiple related hill tops as we see today, and not iaolated local optimums, which are not the what is opjectively observed as regionally interreated regional and continental evolution, especially with migrating populations. Evolution as objectively observed is regional and continental relationships of variations in populations sub-species, related species and their varieties. In fact human evolution has been demonstrated as regional and continental genetic relationships between populations accross Africa and Asia with interbreding between different variations of varieties and sub-species at least several times with regional and continental migrations.

The driving force in evolution is regional, continental and eorld changing and cyclic climate. Nothing to do with local hilltops.

Well, if the fitness landscape continually changes, that shortens the time available for searching considerably.

Simply by the known fossil and genetic evidence this is false the answer is no. As is the witness today the continually changing environment encourages the best populations to adapt to change. Those that do not adapt through evolution go extinct.

[lee_merrill]

There are millions of opitimums, as the history of life on earth they are not isolated...

Well, how do you know this? as I have seen, for example, protein space is sparse.

Source: Wikipedia

Most sequences in sequence space have no function, leaving relatively small regions that are populated by naturally occurring genes.

Source

© Copyright Original Source

Blessings,
Lee

[shunyadragon]

Well, how do you know this? as I have seen, for example, protein space is sparse.

Source: Wikipedia

Most sequences in sequence space have no function, leaving relatively small regions that are populated by naturally occurring genes.

Source

© Copyright Original Source

Blessings,
Lee

So what? This does not remotely address the issues. There remain trillions if not quadrillions of optimums over the billions of years of life.

[rossum]

Well, how do you know this? as I have seen, for example, protein space is sparse.

You are confusing two different things. The fitness landscape is a part of the environment. Protein space is the set of potentially effective proteins. You are comparing apples and oranges so your point is irrelevant.

[lee_merrill]

So what? This does not remotely address the issues. There remain trillions if not quadrillions of optimums over the billions of years of life.

That may be, but is the fitness landscape sparse?

You are confusing two different things. The fitness landscape is a part of the environment. Protein space is the set of potentially effective proteins. You are comparing apples and oranges so your point is irrelevant.

I was using protein space as an example of the sparseness of what evolution has to examine. A fitness space could be put on top of protein space, such that all proteins have a fitness value for a particular purpose.

Blessings,
Lee

[shunyadragon]

That may be, but is the fitness landscape sparse?

No, and it has been made clear by [quote=rossom=You are confusing two different things. The fitness landscape is a part of the environment. Protein space is the set of potentially effective proteins. You are comparing apples and oranges so your point is irrelevant. [/quote]

I was using protein space as an example of the sparseness of what evolution has to examine. A fitness space could be put on top of protein space, such that all proteins have a fitness value for a particular purpose.

The genetic diversity developes many large populations over a long period of time. There is no sparseness involved. In fact if sparseness was an issue than the population would likely go extinct.

[lee_merrill]

The genetic diversity developes many large populations over a long period of time. There is no sparseness involved. In fact if sparseness was an issue than the population would likely go extinct.

I need a reference for this claim! A sparse fitness landscape does not mean the population will be unfit, it only means the population will tend to stay on a local optimum. Thus the surprising discovery of exquisite designs in nature.

Blessings,
Lee

[rossum]

A sparse fitness landscape does not mean the population will be unfit, it only means the population will tend to stay on a local optimum. Thus the surprising discovery of exquisite designs in nature.

That is what evolution predicts. If there only a few local optima then populations will tend stay on those local optima -- the distance to the next optimum will be too large. You need to find a design prediction that is different from the standard theory's predictions.

The number of optima has little to do with evolution; different optima are part of the environment.