nobody knows, which is kind of the point.

pretty much every agrees that older outputs are more likely to be cold storage or lost coins or whatever.

but nobody knows *how much more likely*

theres considerable debate if it's even knowable at all.

so for now, its just comparing a guess about what someone thinks the real age distribution of spends should be

against the age distribution that appears.

decoy selection is hard and this has always been a problem. which is why Monero will move to full chain membership proofs and dump ring signatures altogether.