@SteveRoth the solution to a maximization (or minimization) problem can be characterized as "interior" or "corner". Suppose we are trying to minimize the function y=x, subject to the constraint that x is a variable that can't go below zero. Then the minimum is x=0, and that's a "corner" solution: you find it just at the edge of the function. 1/

@SteveRoth similarly, if we are trying to maximize y=x, we can say that there is no solution, or call the solution ∞. that is, less straightforwardly but not too weirdly, also a corner solution: it is not some value in the interior, but as far as you can go on the outer edge. 2/

@SteveRoth if we are trying to maximize y=-(x^2)+x, though, we find an inverted U shape. initially as x rises, the x term overwhems the -(x^2) term. but at a certain point (at x=1/2) the negative term matches then overwhelms the positive. the solution is interior: neither going to zero (if that's a lower constraint) nor rising indefinitely towards infinity maximizes the function. there's a value in the middle that balances the terms, or (in a perhaps risky intuition) the tradeoffs. 3/

@SteveRoth in real life, most things involve tradeoffs. when you eat ice cream cones, it's not just that there's "decreasing marginal utility". at some point (not too far above one cone!) not only do you "get" less from ice cream, but downsides like near-term nausea and long-term arterial plaques utterly overwhelm whatever pleasure you still get. eating 20 cones is not just little better than eating 1. it's much worse than eating none! 4/

@SteveRoth so there's an interior solution to U(ice_cream_cones) at (for me) maybe ice_cream_cones ≅ 2. 5/

@SteveRoth our models of U(c) for general consumption look nothing like this. they are always nondecreasing. we can never, in the model, be made worse off with more consumption. economists justify this shape by letting c refer to a vector of the potentially infinite goods we might buy, and by presuming "free disposal" (if the extra ice cream would be bad, we can throw it away, so in fact an extra cone can never hurt us even if it doesn't help us). 6/

@SteveRoth "free disposal" is not in fact accurate, as my garage full of junk attests. but more foundationally, that general near infinite vector of goods that suggests there's always something you'd be better off if you could also consume it is pretty questionable. 7/

@SteveRoth i think a more accurate intuition is that present consumption saturates, we are no better off, maybe worse off, for more, however extensive the amazon catalog, under any realistic understanding of our practical constraints. 8/

@SteveRoth critiques of consumerism might be understood as arguing that this saturation point is lower than most of us imagine it to be. 9/

@SteveRoth economics can rescue itself from much of this critique, though, by letting consumption be intertemporal however low our saturation point, if wealth can purchase consumption in the future, then eventually new wealth will buy us (or our great grandkids) something valuable. 10/

@SteveRoth and when you model wealth is insurance, in the ordinary sense against idiosyncratic risks and also in the extraordinary sense of against systematic risks (lifeboats on our libertarian titanic), then it's easy to argue that *wealth* maximization should indeed lack an interior solution, the more the better. 11/

@SteveRoth but it's an easy, mistaken slippage to imagine that should then be a characteristic of "consumption". /fin

@SteveRoth yeah. economics really confuses itself on this point. 1/

@SteveRoth economics gets, in general, that in nearly all interesting cases, an interior solution will prevail. whatever the good that comes with increasing some variable, some tradeoff will eventually arise and overwhelm it. 2/

@SteveRoth but then the very heart of modern economics is "utility maximization", which *by construction* has no interior solution. it is not a thing in the world, or even a model that tries to describe a thing in the world. it is just a gadget we've invented to organize how we explain and predict human behavior. 3/

@SteveRoth but our weirdly constructed little model bleeds into implausible attitudes about real phenomena in the real world. 4/

@SteveRoth we model utility as an increasing function of what we call consumption, so that consumption maximization takes on the unboundedness without tradeoff of utility maximization. 5/

@SteveRoth but that can only work is we let the modeled phenomenon stand-in for the thing in the real world. like "investment" and "saving" and all the rest, we easily confuse ourselves with false friends, drawing inferences from the models about things to which we give the same name but are in fact not well characterized by the model. /fin