Evolutionary Algorithms and ID

We’re not talking about how it evolved. We’re talking about how it creates new information in real time using variation and selection.

1 Like

Why? Because you say so? Aren’t those goalposts getting heavy?

Next, you’ll be equivocating on the meaning of “thing.”

There is no “ID theory,” Gil. A theory is a hypothesis with a long track record of successful predictions. You’ve merely got a hypothesis that you’re clearly unwilling to test rigorously.

1 Like

I don’t see how “function” is any less arbitrary than “thing.”

Gotta make sure those goalposts are very mobile, right?

The claim is that FI is lower for the 100 safe case. That must mean that you can calculate a value for FI. Just tell me what it is, for whatever search you think is appropriate, and how it’s calculated.

ETA: The target is clear: open all of the safes present, whether 1 or 100.

3 Likes

Then we can easily set the “thing” or “function” as the humoral immune response, because based on a huge amount of data, there are easily 100 clones doing this function together, in parallel.

There’s also the cellular immune response, which does a similar, yet different thing.

Come on, Jordan, you’re a scientist. These guys have absolutely no intention of testing a hypothesis. @gpuccio even told us that his goal was polemic.

1 Like

FYI: There is extended discussion of the Safes and Thieves example here:

And here:

1 Like

But as we’ve been over many times, this assumes the thing in question was produced in a single event, as opposed to an accumulation of smaller events, some of which receive feedback from selection.

3 Likes

Then all arguments about how difficult it would be to “find” functions with blind searches are irrelevant to evolution and therefore fail as arguments against it. You should contact Dembski and tell him you’ve found a fatal flaw in his ideas.

1 Like

As shown below by @gpuccio, the new information you are talking about, although random in some aspects, is in fact produced by « functional protein engineering embedded in a complex biological system ». IOW, design is involved at all stages in the production of new information by the IS.
https://uncommondescent.com/intelligent-design/antibody-affinity-maturation-as-an-engineering-process-and-other-things/

A)Search space for 100 1-bit safes:
Search space for 1 safe: 2
Search space for 2 safes: 4
Search space for 3 safes: 6


Search space for 100 safes: 200

B) Search space for a 100 bits safe: 2^100

Great. Spell it out, though. What are the 6 possible targets for 3 safes?

2 Likes

Water runs downhill in a blind, mindless and purposeless search to maximize entropy.

GA’s generate randomness in population of test solutions, and then expel excess randomness as determined by some fitness function, resulting in a population of overall higher fitness. GA’s search in random directions, but not all of those searches are allowed to continue. A human programmer may tune this to some useful purpose, but purpose is not a requirement for the GA.

2 Likes

So does that mean FI is not uniquely defined for a given sequence? If a protein can serve two purposes, two functions, then there are two FIs, right? The search space would be the same but the target space would be different. I find it hard to understand how we could use measures of FI as an argument, one way or another, if the first step is to define a singular function (so that we can get an unique FI value). Functions can and do change, right?

3 Likes

As usual this an interesting idea to ponder :slight_smile:

Oy vey! You’re killing me here Bill.

3 Likes

As far as two FI’s I don’t think so. There is one FI that can create a protein with two functions.

Cytochrome C is one of the proteins in the electron transport chain yet is also involved in skull development in mice. Two mission critical processes of animal development and sustainability. The function in this case would be both ATP production and embryo development. This requires a higher level of specificity than for a single function.

OK, I think that makes some sense, but doesn’t it make some of the criticisms about the lack of knowledge of the target space worse? To truly calculate the FI, wouldn’t we need to know every possible function? Or is the actual FI value not interesting to you, but rather you are using it to place something like a lower bound (as in, “the FI is at least X”)?

2 Likes

Yet I can shuffle two standard decks of cards together, deal them out, and the resulting sequence will have a probability of 1.03 x 10^166 which blows right past Dembski’s UPB.

3 Likes

Your point is well taken. Gpuccio and Kurt are estimating FI and its accuracy is assumption dependent. I think Gpuccio’s FI calculation is pretty accurate but that is because I think the design argument is right. If you live in the evolutionary paradigm where lots of function exists and we are observing lucky accidents when we see preservation you will see the estimate differently.

I see such a small window for evolution to work that I have discounted it as a viable explanation and so given the design hypothesis a protein that is preserved over 400 million years shows very high FI as it has to be in that configuration for a healthy functioning animal that can compete in the reproduction battle.

What makes you think that these are 2 separate functions, rather than the role in skull development being a result of its role in the ETC?