Matt Yglesias is talking about high speed rail here, but it matches pretty closely my thoughts about selling OLPC:
On the idea that ridership estimates are unrealistically optimistic, it seems to me that the sad reality of politics is that it would be irresponsible for advocates of any large-scale infrastructure project to do anything other than present unrealistically optimistic measures. For better or for worse, that’s politics. Similarly, I never really understand the sentiment that Large Infrastructure Project A shouldn’t be done because Large Infrastructure Project B might be better. Sometimes you really do get asked “should we do A or should we do B” in which case, of course, if B is better than A you ought to answer “B.” Similarly, sometimes doing A really does prevent you from doing B — like if A and B would both require the same right of way. But that’s not generally the case, and it’s certainly not the case when you compare a statewide HSR system to a series of different local transit projects. In general, large infrastructure projects should be evaluated on their own merits. If California HSR is worth doing, then it really doesn’t matter if there may be other transit projects that are also worth doing. You do the HSR, and then you start organizing for the other projects. Doing worthwhile infrastructure projects ultimately grows your capacity to do future infrastructure projects.