When you use certain tools, like marginal analysis, you always need to be asking yourself "what does this tool do?".
Marginal analysis answers the question "how are prices set when everybody wants to maximize their objective?". It is a standard undergraduate exercise to demonstrate that when you have agents that want to maximize their objective prices are set equal to marginal contributions to the achievement of that objective.
So that's what marginal analysis is for.
In this case, Mr. WheelerDealer's labor supply influenced these marginal contributions, so as Bob demonstrates, a labor supply shock to Mr. WheelerDealer would raise his wage rate and lower production worker wage rates. That's the sort of question that marginal analysis is designed to answer.
But marginal analysis is not designed to answer questions like "what is total output?". Marginal analysis describes price setting behavior of individuals with particular objectives. It's a partial equilibrium thing.
Robert Solow and others smuggled that into macroeconomics, and for certain insights that's fine (like capital accumulation and growth theory). Why? Well because capital accumulation is influenced by the return on capital and growth is influenced by the capital ratio (which is influenced by accumulated capital, which is influenced by the return on capital). So if you want to talk about all these problems, Solow was right to use marginal behavior in these traditional production functions. Add a capital accumulation function, and you're ready to go.
But these aren't the problems we're dealing with in this discussion.
In this discussion you need something like an aggregate demand and aggregate supply curve. Once you've got that it's very easy to see that a supply shock reduces total income. The other option is to add factor markets in the Bob Murphy rendition. That's what I talked through before.
Now that raises the question, of course: what happens with the taxes on the rich? I just talked through the factor markets. But presumably what we're talking about is a tax wedge that reduces Mr. WheelerDealer's labor supply. Those taxes go to government spending.
Crap, now that's more to model!
So yes, the story is more complicated than just "total income is reduced". Because government is demanding stuff too. And in some cases, as in state and local governments, government demand is very closely tied to tax revenue.
So the story could go on even further, and this is the part of the story where I'd probably start switching back to agreeing with Krugman and not Bob.
But the basic point (where I still agree with Bob) is this: if you're going to finance government demand (which you need to do), why in the world would you do it by cutting into the supply of a non-idle factor? Why wouldn't you make use of a relatively idle factor instead (for example, relatively idle loanable funds)?
As a long-term discussion we can talk about taxing Mr. WheelerDealer more simply as good policy. I'm very willing to have that discussion. But we shouldn't pretend it doesn't have labor supply or total income effects. It does. And those have relative price effects.