The central idea is that debt liquidation causes deflation which then increases real debt.
The more debtor's pay, the more they owe. He calculates that the debt was paid down by 20%, but in "real" dollars, increased by 40%.
Similar to the well known paradox of thrift, this assertion hangs on a strong, fixed relationship between asset prices and overall price levels.
We have seen a bit of this cyclical behavior already. Some of it is part of any market cycle, but the parts of most interest are where he notes that the money interest on safe loans falls and the money interest on unsafe loans rises. In other words, interest rate spreads increase.
Also of note is that relatively early on, commodity prices fall.
Another interesting point is that he believed that the Fed had a number of opportunities to reflate and didn't take them. He didn't seem to think that it would have been difficult, rather it simply wasn't tried before things got out of hand.
His recognition that you don't get debt levels down by liquidation under extreme circumstances is astute.
I suppose the part I find least convincing is the idea that high debt levels, per se, cause deflation. However, once a spiral starts, high debt levels make it much worse -- of that there is little doubt.