[platz]

much like the trolley problem and driverless cars, you can easily get lost in the weeds in a legitimate, complex theoretical question, when for a concrete problem a simple answer can easily be obtained. In fact, the theoretical question obscures or acts as cover for the concrete case. For driverless cars, you just apply the brakes. For google, do you outlaw "The double Irish with a Dutch sandwich"

[sokoloff]

It's not that a simple answer can't be obtained, it's that many different simple answers can be obtained...

[platz]

sure, if you are trying to solve the fully parametric abstract equation over all possible inputs, which is precisely what I was saying to try to avoid.

In absence of a grand unified theory, tackle obvious concrete instances when they crop up.

If the different "simple" answers (which I am not sure exactly what that means) solve the problem, then I really do not care if there is some equivalence principle there; just pick one as long as it produces the concrete outcome that you want.

[sokoloff]

The problem can be roughly stated as: For a company registered in country I, selling product/service in country F, with inputs licensed from and/or created in countries N, U, B, G, A, and C, and costs incurred in a potentially different set of countries, what's the amount of tax due to each country represented in the graph?

There are many simple answers to that question; who gets to decide which simple answer is the one enacted? Countries F, I, and U are likely to disagree on which simple answer is "best", which is why we rely on laws and courts to frame and adjudicate the issue.

[sisu2019]

Outlaw what exactly? Subsidiaries? Licence Agreements? Profit/Loss calculations? The European Union? "The double Irish with a Dutch sandwich" is an emergent property of a bunch of laws "working as intended".

So what you really ask is to get rid of the rule of law so you can target the bad guy du jour directly. Just remember that you may be "the bad guy" tomorrow.