Been chewing on this one for months and it still gets me: a 100% tax can change absolutely nothing. Not a loophole, not evasion, not offshore anything. Just arithmetic.
Start here: if wealth grows by multiplying, it condenses. One holder ends up with everything. That isn't a flaw in some particular model, it's what multiplicative processes do. Which means every wealth distribution that isn't condensed is being opposed by something. Fine .. but by what? I stopped asking "which institution" and started asking about coordinates: base, rate, periodicity, threshold. And then one more that turned out to matter more than the other four put together .. realisation, the share of a period's gain your base can actually see.
Set realisation to zero and a 100% levy on flow leaves the wealth vector exactly unchanged, agent by agent. The rate never gets a turn. You can't tax what you can't see, and the rate is just a multiplier on nothing. The collapse from there is violently front-loaded too: the first five percent of the realisation axis covers 32% of the reachable range.
So the base caps the region and the rate only moves you around inside it. Everything reproduces from open code and 18 tests pin it down. Paper's here if you want it, it's 20 pages.
Start here: if wealth grows by multiplying, it condenses. One holder ends up with everything. That isn't a flaw in some particular model, it's what multiplicative processes do. Which means every wealth distribution that isn't condensed is being opposed by something. Fine .. but by what? I stopped asking "which institution" and started asking about coordinates: base, rate, periodicity, threshold. And then one more that turned out to matter more than the other four put together .. realisation, the share of a period's gain your base can actually see.
Set realisation to zero and a 100% levy on flow leaves the wealth vector exactly unchanged, agent by agent. The rate never gets a turn. You can't tax what you can't see, and the rate is just a multiplier on nothing. The collapse from there is violently front-loaded too: the first five percent of the realisation axis covers 32% of the reachable range.
So the base caps the region and the rate only moves you around inside it. Everything reproduces from open code and 18 tests pin it down. Paper's here if you want it, it's 20 pages.