You usually want to incentivize R&D, messing with that simple theory is the standard.
Anyway, even that 3 years common knowledge number is biased into long-lived code. Do you spend work time putting out fires? Do you write code that is immediately discovered to be wrong so you have to take more time and fix it? Those 3 years are only for software that already passed through all of that.
(Anyway, I don't have any interest on that fight, I'm just watching from a safe distance.)
> Does "maintenance" on an amortized asset have to be amortized as well?
Usually not. Also, from other comments here, it looks like bug-fixing doesn't count as R&D (anyway, in a development focused company, you'll probably spend more deciding what is or isn't bug-fixing than you'd get from the difference).
Still, there are many more things that make software disappear just after being written. It's actually similar to most of R&D, so it makes little sense to count it the same way as capital investment.
Anyway, even that 3 years common knowledge number is biased into long-lived code. Do you spend work time putting out fires? Do you write code that is immediately discovered to be wrong so you have to take more time and fix it? Those 3 years are only for software that already passed through all of that.
(Anyway, I don't have any interest on that fight, I'm just watching from a safe distance.)