Back when I worked developing large software systems, every now and then we ran into a bug that management decided was too much trouble to fix -- "It's not a bug. It's a feature!" This is the approach that Kevin Drum seems to be taking when it comes to volatility in cap-and-trade programs.
The short version of the volatility problem is that with a trading system, permit prices vary not only in response to how many permits are issued, but also in response to general economic conditions. As a result permit prices bounce up and down a lot. Kevin, like a number of cap-and-trade supporters argues that this volatility is a good thing, because permit prices drop during bad times when people don't have money to invest, and they rise during times when they do. In short they argue that counter-cyclicality makes volatility positive rather than negative. But, just as in the software industry, I'm afraid it is still a bug, not a feature.
To the extent that emissions pricing accomplishes anything it drives investment in emission reducing infrastructure. But when emission prices drop too low, firms project long-term prices to be low as well. Managers get a lot more points for increasing or preserving market share than they do for managing environmental risks. Top bosses don't want to hear that emissions costs are going to rise, and the company needs to invest in reductions to comply with a cap-and-trade system. They want to hear that they can concentrate on their core business and buy low-cost permits from all the other firms reducing emissions. There is always a sound business case to be made for the other guy to reduce his pollution.
