I suggested in this All Things Considered interview  that not all infrastructure yields a positive benefit. As an example I mentioned the 12 or 14 lane stretch of highway in Huntsville Alabama that did not seem terribly useful but that was probably the result of a powerful Senator rewarding a friend. Here is a picture  of that stretch of highway from Google maps. In the stretch I've linked to, the road appears to be 12 lanes wide, but part of that is due to an exit ramp–for most of the stretch to the left, it's a mere TEN lanes wide. Still plenty wide given how little traffic there is in Huntsville along that stretch. But I exaggerated the distance where it was really wide–I said 3-5 miles, I think, when it's only a fraction of a mile. Confirmation bias, probably. But the point remains that not all infrastructure spending is particularly beneficial. Yes, roads and bridges are important and valuable. Yes, maintenance of said roads and bridges is crucial for safety and effectiveness. But that doesn't mean that the next $825 billion of roads and bridges will yield great benefit.
Anybody out there from Huntsville who knows the history of this stretch of highway? Please let me know.
The other response to the ATC piece is the argument that everyone knows we've been neglecting infrastructure for years and we're in desperate need of infrastructure. Remember that bridge in Minnesota? But that is perhaps a different kind of confirmation bias. As David Leonhardt pointed out  in the New York Times, infrastructure spending as a proportion of GDP is at a higher level today than anytime since 1981. 
Despite my confirmation bias, I wonder if this picture (and the data underlying it) is really meaningful. How is infrastructure defined? I wouldn't be surprised if the definition of infrastructure hasn't widened in recent years to make pork more palatable. It's one thing to get funding for the National Bowling Museum, a tourist attraction in St. Louis. But when you classify it as "infrastructure" you're clearly helping the nation.