Jeffrey, yDNA is a bit of a different animal here because the whole of the chromosome is passed along to male children, and the overwhelming majority of the Y chromosome is free to mutate without detrimental repercussions. Even then, the mutation rate is fairly low for SNPs/SNVs and the aggressive variance is usually seen in copy counts among the fastest-moving STRs.
Although older males can have a greater tendency to have offspring with genetic anomalies, often structural in nature (as opposed to Mendelian dominant/recessive genes), what definitely impacts children of aging males isn't so much about variants being included in the gamete as it is about where crossing over during meiosis is most likely to occur.
There are well over 50,000 identified "hotspots" for double-strand breaks during meiosis (many of which, BTW, not accounted for in our decade-old GRCh37 reference genome and, many believe, not adequately measured in GRCh38, either). These are specific chromosomal locations where recombination is most likely to occur. We have a better handle on where crossovers are not likely to occur (e.g., at the centromere and in pericentromeric areas, heterochromatic regions).
However, unlike females where the ova undergo crossing over and the first stage of meiosis while the woman herself is still a fetus, males are a produce-on-demand operation. A process called deamination occurs as a result of DNA methylation over time. This, among some other things, has the result of shifting many of the crossover hotspots.
The end result is that a child still gets 50% of the father's DNA, but our calculation models for centiMorgans doesn't take into account these shifting hotspots. Since a centiMorgan is nothing more than a probabilistic model of where crossing over is most likely to occur when a gamete is produced (which is why the male and female genomes differ by about 70% when it comes to cMs), with males having children as the fathers approach senior citizenry, our predictions about shared segments breaks down.
Males will undergo crossing over somewhere around 26 times per meiosis compared to a woman's approximate 45 times. So if our parameters of where we expect the crossovers to most likely occur gets a wrench thrown into the works, the resultant calculations of centiMorgans as computed for future generations can be off. If we could test Samuel, his daughter, and her half 1st cousins, the cM calculations might look a bit wonky, but the percentages should remain consistent with what's expected.
That said, with genealogy we always sex-average the results (which kind of makes centiMorgan comparisons with distant cousins a bit of an exercise in futility), but just like time does in fact average out the centiMorgan calculation and dilutes the impact of limited pedigree collapse, it also dilutes effects of deamination in male gametes. My guess is that from testing Samuel's living descendants today we'd have no good way to be able to tell the difference in the DNA passed along to his children, whether he fathered them as a young or old man. Mind you, though, I have zero research papers I can cite about that opinion, so it's worth every penny you paid for it.
