Also remember that the average person breaks even, so affordability isn't really an issue.

You probably want the initial value to be high enough that you can actually displace some existing means-tested aid programs, even though you may not initially be able to do all of them.

Slight adjustments in the number of beneficiaries doesn't really get you much benefit in terms of eliminating the inefficiencies of existing programs (in done ways out actually exaggerated them since there is a fixed component of the overhead costs for each program.) The real efficiencies Congress when you are able to eliminate programs altogether.

One benefit of that kind of ramp up is that it naturally phases out existing programs, making them easier to kill when their time comes (and much clearer when their time has come).

It also means (to the risk point above) that if we discover some ruinous dynamic caused by the UBI, we can more easily scrap it and fall back on existing systems.

You are assuming that every one of those people stops working and lives exclusively on the dole.

That's almost certainly not going to be the case.

Now sure... wealthy people will pay more in taxes than they receive in basic income. But it would still be a tax credit.

Yes, but that doesn't mean it needs to be enough for a "living wage." It can still be a universal basic income while being low enough that recipients generally still need to work.

(Of course you'd want to supplement it with welfare programs for those who can't work, which under more generous schemes might be eliminated entirely.)

http://www.wolframalpha.com/input/?i=us+gdp

Edited to add: I wondered how much of this was population growth, and certainly there has been some but real GDP per capita has grown about $5k over that period.