Mutations cannot add information.

From RnRWiki

Mutations cannot add information they can only destroy it.

The Short Smackdown

"Ooh, interesting point. How do you define and measure information?"

Watch the creato squirm. How, indeed, are they supposed to know information is only destroyed when they don't know how to measure it?

The Long Smackdown

Those who propose "mutations cannot add information" are conflating two things. We say, for ease of explanation, that DNA is like text. This is an analogy. Numpties confuse the analogy with the actuality. DNA is not text. Their entire argument is based on the fact that if I mistype a word I'm far more likely to get nonsense than another word. This is pushing the analogy too far. More useful would be to say DNA is analogous to numbers. If I mistype a number, I just get another number.

The more significant mistake made by proponents of the information argument is to conflate fidelity of the signal passed by DNA with fitness of the organism to its environment.

Information theory tells us that noise inevitably and permanently degrades a signal. But 'degrades' applies to the accuracy of the signal; the degree to which the information remains unchanged. In those terms, a beneficial mutation and a harmful mutation are both degradations to the original signal: they make it not the same as it was. Whether that change represents an improvement to the organism's fitness in its present environmental context (or more usually a later one) is a completely seperate issue.

An illustrative example: imagine a bank transfer of £100, which through noisy interference to the signal is accidentally 'mutated' to a transfer of £1000. Is that good, or bad? Depends upon whether you're paying, or being paid! Either way, it's a degradation of the intended signal. Same applies if the error made it £10 - again, a degradation of the signal which might be helpful or harmful depending on the context.

Let's see if we can push this a bit further. Suppose you work on contracts. You get paid by the job. So sometimes money comes into your account very frequently, sometimes not very frequently. But your bills go out at a steady rate. Suppose also that you have the ability to block transfers of a certain amount.

At the times when you are getting lots of contracts, £100 lots will be flying into your account, and occasional hundred pound lots will leave. You will want to kill all £10 mutations, and allow all £1000 mutations to stand, as you will want to take the occasional £1000 bill in return for lots of £1000 incomes.

At the times when you are getting no contracts, £100 will be regularly leaving your account, and not much will be coming in. You will want to kill all £1000 mutations, and allow all £10 mutations to stand.

I am your accountant. I come along and review your transaction history. I can work out during what periods you were working, and during what periods you were not by seeing which mutations were allowed at what point. This means you've coded information about your environment (financial status) into your history.

When we talk about it being impossible for noise to 'improve' a signal, all that means is that it can't make it a more accurate representation of what was intended. Information theory says nothing, nothing at all, about whether the degraded signal will be of greater or lesser value to the reciever.

Ducking Hell

Birds produce a protein called BMP4 which induces apoptosis in their feet. Ducks have webbed feet.

How do we get from birds to ducks?

The answer is not that ducks stop producing BMP4, they produce normal amounts of it. They also produce a protein called Gremlin which blocks the BMP4. So now we have two proteins and no loss of flesh.

This is a suppressor mutation, and, however you cut it, either the mutation that meant birds started producing BMP4 or the one that gave ducks the Gremlin must be "adding information", as one reverses the other.

Therefore the "no new information" canard is demonstrated to be wrong by cute little ducks.