Here I'm going to show you just how unutterably dumb ID/creationist arguments against genetic change adding information can be.
Oh, one thing to keep in mind. Genes are not "always on". Sometimes they're doing stuff, but for most of them they're only on some of the time. Some only for a very short time. The activity of genes are controlled by other genes (and sometimes by epigenetics). Most of the time, a lot of your genome isn't actually doing anything.
IDist claim 1) "You can't add information, only delete it"
So the point being, presumably, that you can (say) alter a gene, but the genome can't acquire new functionality without destroying the old - presumably the idea is you can change a base pair or delete base pairs, but you can't add base pairs.
This is false. It's demonstrably false (just right off the top of my head I know counterexamples, and I'm no biologist). But I'm not going to give a link for this - other people (like, say ERV) are better placed than me to do that (and people have done so). You can add DNA that does something without necessarily harming the organism. But I'm not going to talk about reality - I'm just going to apply a little logic to the IDists own arguments.
So anyway, one of the things that's easy to point out is that gene duplication happens. So you can have two copies of a gene where there was one. This has happened many times and it's a bit hard to dispute. So there's a shifting of the goalposts:
IDist claim 2) "Okay, gene duplication exists, but having two copies of a gene doesn't add information - you have two copies of the same thing"
Well, actually, that's false as well. In some circumstances, that actually is new information. For example, it's now possible to have 0, 1 or 2 copies of the gene expressing at a time, instead of 0 or 1. The genome can do things it couldnt before, without losing the ability to do what it could.
But let's ignore that. Let's jump right into the denialist tent and agree for a moment that you will consider if duplication were the only way the genome could get longer - any other changes will change an existing function or remove it, and either way you don't add information.
So under dumb IDist logic I'm allowed these transformations, with a tiny piece of pretend genome from a simple organism:
1) [Gene A ]....[ Gene B ] => [Gene A']....[ Gene B ]
(a base pair, or maybe a few base pairs get lost or changed)
2) [Gene A ]....[ Gene B ] => [Gene A ]..
(a whole gene gets lost)
3) [Gene A ]....[ Gene B ] => [Gene A ].[Gene A ]....[ Gene B ]
and the claim seems to be none of those three can add information. The additional claim is that (1) and (2) are almost always deleterious. Well, that may be usually true. But sometimes a somewhat different version of a gene will still work okay at what it used to do and can nevertheless do a bit of something else. An IDist will argue that that doesn't get you anywhere.
(Well, some IDists might agree that you can have some of the other change that are observed, like inversion, but then they'll claim that they're always going to be bad and don't add information. Also wrong, but lets leave it aside for now. We don't need it.)
Now watch. Imagine for simplicity we're just looking at something with asexual reproduction, to make things nice and simple. Remember we're totally ignoring a whole pile of nifty stuff that actually happens. Doesn't matter. The IDist's fake genetics will do fine.
Anyway, here's the argument with our pretend genome:
[Gene A]....[ Gene B ]
Apply IDist transformation 3:
[Gene A ].[Gene A ]....[ Gene B ]
(imagine copy 2 of gene A is always switched off, or only switched on
briefly at some point in development and doesn't do a whole lot. Whether it's advantageous or disadvantageous to have a second copy, imagine for a moment it's not doing enough to be selected for or against. Its pretty much just sitting there, generation after generation. "No real information has been added!" comes the cry.)
Now apply legal transformation 1:
[Gene A ].[Gene A']....[ Gene B ]
That's all. Note this A' is a different locus. A' is not an allele for A. It's a whole entire copy of A, which has then altered.
We now have *all* the information of the old genome, plus an entirely new gene. We have NEW information, and the genes or epigenetics that control its expression can now either express it more or less. New information. End of story.
Well, end of that story.
Now, lets say that most of the time, a change like this is middling bad. No problem, since if the organism isn't sick or dead, it's because the gene is off (or at least nearly always off). If it's Really bad, copies that are "off" more will be selected for, because the organism that happens in is around longer to reproduce. So we have a gene it doesn't use hanging around. Sooner or later it will mutate again, to gene A'', say. That might be bad, too. Some offspring might express it for longer and die, or maybe just reproduce less. Others will express it less and stick around.
But once in a while, a sequence like that will produce a gene that does something that helps a bit in some circumstance. It might be gene A' that does something just a little bit nifty. It might be A''. It might be A'''. One of the descendants expresses that gene a bit longer, and maybe it can metabolize a slightly different food source, or it can remove a particular toxin from its cell or any of a thousand other things.
That change hangs around. A whole lot of other such changes accumulate. Some of these "sort of" helpful changes interact - they can do things together that they didn't do apart. Now, if they could be made to work together better, that would be nifty, but the IDist will say, if you change that, you don't add information - in fact you could harm the sort-of-nifty stuff that you were getting. Here's the thing - if that's true, we can do exactly what we did above - duplication followed by a change, and lose nothing, while having the new interaction free to adapt to work better.
Evolution doesn't really work exactly like all of this. This is fake IDist genetics, remember. Evolution has much more to work with. But even fake IDist genetics can add NEW information.