Well, if you pack each gene with introns, and every base has an equal chance of being an indel, then you increase the chance that a frameshift mutation will be canceled out by another one, downstream.
I still don't follow that. Yes, if you make the gene longer, than there's more likely to be two indels in it, but there's also more likely to be one, or three which are not all of the same kind, or four, or ... etc. The ideal number is zero.
Also, how would it help if you got a compensating indel in an intron? This intron is taken out before the translation. If it was translated and then the amino acid sequence corresponding to the intron was taken out after the translation,
then an indel in the intron would compensate for an opposite indel in an exon, but it isn't ...
... I don't really see what you're getting at, or if I do, it isn't right. One or the other.
Edited by Dr Adequate, : No reason given.