I had a long conversation with Noah Smith on the twitter, which was fun. He asked that I first imagine that Person 1 and Person 2 own an apple and an orange respectively. Then he said imagine they 1) do not trade, or 2) trade them. He asks whether (1) or (2) has the same amount of violence involved. The answer of course is somewhat unclear, but it appears that -- given the parameters of the hypothetical -- doing (1) and (2) both have zero actualized violence and both probably have the same amount of threatened violence. Since they are both equally violent and trading increases both of their utility, that means (2) is better than (1). It was a fun exercise to play along with there being an initial ownership to see where it led. It's not clear actually what it entails prescriptively because taxes are just as violent as exchange in this scenario (as both rely upon threats of violence to actually run) and taxes can also be used to increase utility. It would seem that this demands that we tax and transfer up to the utility-maximizing level because going from a world where we had that kind of tax regime in place to one where we did not would clearly be utility-reducing, both to those at the bottom and in aggregate. Though the hypothetical was fun to indulge, the obvious problem with it up front -- as regular readers are painfully aware of by now -- is that it has assumed in some kind of baseline ownership. That is always where any of these discussions fails, and similarly why all of economics that relies upon utility-measured-against-a-baseline also fails to be compelling as a normative theory. Consider the following series.

1. Nobody owns anything at this point.
2. Sally takes an unowned apple and claims to own it.
3. Sally takes an unowned orange and claims to own it.
4. Rachel takes the apple Sally claims to own.
5. Rachel trades the apple to Sally for the orange.

Did step five improve the utility of both Sally and Rachel? Was it a mutually beneficial transaction? It appears so. Through their revealed preferences, it is clear that Rachel gets more enjoyment out of the orange than Sally does and Sally gets more enjoyment out of the apple than Rachel does. That's why they traded. So is everything alright then? Did we just achieve justice here? Of course, the intuition here is no. Wait a minute, didn't Rachel steal that apple from Sally in the first place? Can we really just analyze the apple-orange trade to determine whether the outcome here is just? It would seem no. But then what all do we analyze here in this method? Well one argument would be that we should wind the clock back to step three before the "theft." At that point, Sally has an apple and an orange. Against that baseline, the result after step five does not see her in an improved position. She went from having an apple and an orange to just having an apple! So see, we have dealt with this problem. This was not justice even though just looking at step five shows an utility-improving move. But if we are going to wind the clock back to step three, we have to wind it back to step one. If you wind the clock back to step one and remove out the "theft", what you have is Sally taking two pieces of fruit and then having both of them while Rachel has nothing. But was that whole process utility-improving for both Sally and Rachel? Absolutely not. In fact, that whole process was utility-destroying for Rachel because Sally grabbed up two pieces of unowned fruit that Rachel could have grabbed up in the alternative. All initial appropriation of unowned things improves the plight of the initial owner, but only at the expense of all other people (because their previously existing ability to access, use, or appropriate that piece of the world is now gone, taken from them by the unilateral actions of the appropriator). So all initial ownership is not a Pareto improvement! When you embark upon discussions of mutually beneficial transactions, this is the double bind you find yourself in. You can look purely at the latest transaction in question with no background knowledge allowed, in which case number five qualifies as a mutually beneficial transaction. Or you can start imparting prior actions and ownership transfers into your frame of analysis, in which case everything crumbles like a house of cards because the initial appropriation of all property is a kind of theft. All subsequent trades are built upon theft, transfers that were not Pareto improvements. The only way you avoid the double bind is to set up a world in which you assume away the initial ownership of things. That can be a worthwhile thing to do for fun hypothetical games, but only if you don't lose sight of what is really going on and why, like all baseline ownership assumptions, it provides no insight into the justness of the overall distribution of goods at any given moment in time.