Steven Landsburg has some good points to make about why trade-induced changes in income distribution should not automatically call forth compensation for the losers. We accept such losses (and offer no compensation) in too many other circumstances--for example, when technological progress leaves some people behind.
But he misses the important point that we do not blithely accept all such situations either. The kind of introspection Landsburg asks us to undertake also suggests that we judge the appropriateness of a change in income distribution by asking whether the mechanism that causes it is consistent with basic moral precepts and prevailing norms. In other words, we use a procedural fairness test. If we discovered tomorrow that Bill Gates had obtained his billions by lying and cheating the companies he decimated (instead of by ingenuity and hard work), we would think far less--or even less--of Microsoft.
Similarly, what is at issue in globalization debates is the procedural fairness of some types of outsourcing. Trade is controversial because it involves exchanges of the type we routinely block at home (e.g., exchanges that involve unfair labor or environmentally harmful practices). This often makes trade look different from other instances of redistribution. I have written about this before.
Another key point not to lose sight of is this. The question of how we should respond to a trade-induced change in income distribution is not one on which economists can offer any expertise. This is a question about ethics, values, and norms, none of which is part of an economist's training. Landsburg's take on this is as good as mine--which is as good as that of any person on the street. But once we accept that trade creates losers, at least we can begin to confront these questions explicitly.