> The recent (-ly undone) change went against decades of how things were, was crippling for medium size cashflow-positive startups, effectively increased taxes, etc. But it was really just a straightforward application of the general principles that apply to most everything else.

The error was in reconciling them by getting rid of it for software R&D instead of allowing other business expenses to be deducted when they're paid for as well.

For large stable incumbents that have the same expenses every year, the difference doesn't matter except in the first years after you make the change, because it doesn't matter if you deduct all of this year's expense this year or 5% of each of the last 20 years' expenses this year, they add up to the same deduction every year.

Where it matters is for new challengers, because they don't have arbitrarily many years worth of legacy expenses to deduct, so their deduction in their first year will be less than their incumbent competitor's.

It also creates a disincentive (or competitive disadvantage) to increase long-term investments. If some existing company had been making a $5M investment every year but is now facing new foreign competition and needs to increase it to $10M in order to stay competitive, they're in the same position as the upstart. Moreover, then they may not be able to do it, because they were going to have to run lean and divert the $5M profit they usually make to increasing their capital investments, but then the government is expecting tax on most of that $5M which means they can't spend it this year it even though it's ultimately a deduction.

Notice what this does specifically in the case of real estate: If rents start going up the normal incentive is to build new housing, but now you have to put out all the money to build a new building in year 0 and not get to deduct it for decades. Is that the incentive we want? Probably not.