Thanks for your service to my limited understanding.
It seems to me, however, that the work of Axe, et al, demonstrates that a minimum of 2 and usually three specifically correlated mutations are necessary before anything like a new adaptive “trait” can arise. So unlike your card trick which relies on one card being “kept,” a more realistic requirement would be that the correct two or three cards would have to come up TOGETHER before they could be “kept.” That would change the timing substantially.
Another article has recently been posted which throws a different wrench into your card trick depiction of evolution. The following article deals with genetic convergence in avian species.
sciencemag.org/content/346/6215/1256846.full.pdf
According to the above paper and corroborated in several other recent papers on the topic, there is a consistently occurring set of just over 50 genes that are required for the brains of birds and humans to be capable of vocal learning. These genetic constituents are not in the genomes of birds that do not exhibit vocal learning, nor in non-human primates, so they must have evolved independently of each other.
The evolution of vocal learning in at least three specific bird species with the convergent traits are separated from any common ancestor by a 68 to 310 million year gap. Essentially, that means the exact 50 + gene combination had to arise in the same orderly way (that your card example illustrates) at least three separate times in birds and once more in human primates randomly from radically different starting points.
This is, by the way, not an uncommon phenomenon. Convergent traits have occurred numerous times in the past requiring that similar genetic combinations arise independently by random mutation across several species.
aeon.co/magazine/science/how-horizontal-gene-transfer-changes-evolutionary-theory/
A third issue with your card trick is that “keeping” a card is predetermined beforehand. It is unclear how a trait such as vocal learning involving 50 + individual genes could have ended at the same “kept” sequence given that vocal learning was not an intended outcome, but arrived at randomly. Yet, why would a sequence of adaptive trait development that “keeps” one trait as adaptive necessarily lead to the next and not some other down the road of survival? It would seem that ending at the same “end” sequence would be inexplicable by mere random determinations since mutations could have gone anywhere along the “road” of adaptive survival.
uch:
