It seems to me that once a person builds an illegal habit, it demonstrates they: (a) probably have a higher risk tolerance than most people and/or an inaccurate understanding of the empirical risk level and/or (b) they have, perhaps only temporarily, found a self-interestedly-rational aspect of law/society to exploit.
Once one's behavior crosses into that realm, "dude just got greedy" seems to miss the point. It was greed that got them there, right? How many optimally greedy criminals are there? I don't claim to know, but I'm suggesting that risk-adjusted _optimality_ and the criminal mind are not highly correlated.
Anyhow, it doesn't seem the 'level' of greed was the problem; it was the level of spending; the article suggests that Kara's lifestyle upgrade attracted considerable attention.
More generally, the risk-adjusted calculus between "steal a lot quickly" vs "steal a lot slowly" is interesting. First, being a repairman of said machines, he would be a suspect if any discrepancy was found. Second, dragging a larger-than-usual bags of coins to the bank for years seemed rather brazen. But perhaps the consistency of his behavior made it seem 'normal' to the local bankers? Putting both together, sure, larger amounts would attract more attention, but the relationship between intake and risk is probably not linear. As a guess, one might think halfing his intake might only reduce his risk by 10%. If so, a self-interested, amoral calculus might suggest that his intake was reasonable.