"Evil" Or Realistic Bears?
The DeLong-Krugman argument rests on historical experience that shows that growth tends to be rapid after deep slumps, as there are a lot of unemployed workers that can be employed quickly and so produce high growth. What is noteworthy in this context is that if you look at Mankiw's original post, he doesn't even deny that, and says explicitly "when recovery comes, it will probably be a robust one.". Yet DeLong and Krugman pretends in their posts that Mankiw denies this.
But Mankiw's point is that we don't know when the slump will end and so we can't say that growth from now until a certain year, say 2011 or 2012 to be high. Even though that is the main argument discussed in his post, DeLong and Krugman fails to mention or discuss that argument. Mankiw is BTW right about this. To illustrate this point, assume that someone in late 1930 had concluded that because GDP had fallen 8.6% that year, there must be robust growth, say 8% per year in the coming 3 years (1931 to 1933). That person would have been dead wrong, as GDP fell another 6.4% in 1931, 13% in 1932 and 1.3% in 1933. To illustrate the point further, let's assume that he had projected 8% per year growth in the coming 6 years instead. That would include 3 years (1934-36) where output expanded. How would that proto-Krugman-DeLong forecaster have performed? Not so good.
While there was indeed a very robust recovery during the 3 years of growth, with output in 1936 being 36.3% higher than in 1933, that followed 3 years of contraction of 19.6% (And that doesn't even include the 1930 contraction). GDP in 1936 was therefore only 9.5% higher than in 1930 (and it was unchanged compared to 1929), which is an annual average growth of only about 1.5%.
Thus, even assuming that the recovery will be robust once it starts this time as well, that doesn't mean that we can say that average growth will be high from this point in time to a specified future year because we don't know how long this slump will persist (and right now, the slump seems to be accelerating in pace) and also, we do not know if there could be future slumps. Japan during the 1990s had some brief periods of high growth, but they were fairly quickly cut short by new slumps (And during the 1930s, there was a new slump in 1937-38).
The reason why there tends to be persistent damage to long-term economic growth is that so much of all investments go lost as they are revealed to be malinvestments, meaning that available capital declines, something which reduces not just short-term but also long-term output. Furthermore, the fact that politicians like Herbert Hoover and Barack Obama tends to respond to slumps with political actions that damage long-term growth, such as higher marginal tax rates and higher spending, compounds this effect.