He also could've included some graphs. I think the upshot of what he's saying is that if you draw the curve for expected returns -- probability on the vertical, return on the horizontal -- the increasing length of the leftward tail means that the integral of that segment, representing your total possible loss, goes up even though the curve appears to get steeper sides.
Good stuff. In practical terms, this would suggest that even a younger worker ought to include some low-risk investments, like T-bonds in their retirement strategy, though exactly how much is unclear. (After playing with FinancialEngines' calculators a bit, I've decided to put 10% of my future 401k contributions into long-term bonds. I also moved a bit more of my contributions from domestic equities to international ones; until I have spare cash to invest in foreign currencies, my 401k's international equities fund is the only hedge I have against a fall in the dollar.) The subtext, of course, is that currently Social Security (which is invested in T-bonds!) is what ameliorates our long-term risk, allowing younger workers to invest more in the high-risk/return stock market.
ETA: And, here's Doug Orr explaining in layman's terms why the "crisis" talk is silly.
ETA: And and, here's Eugene Steuerle, Reagan's chief advisor on tax policy in the later years of his administration, saying, "Social Security benefits are already equivalent in value to about a $400,000 401(k) plan for the average-income couple retiring today. Add in Medicare, and the total package of benefits for a couple is projected within less than 25 years to exceed $1 million," and that the whole concept of personal accounts is incompatible with the mission of Social Security, because SS is intended to push aid to those who are neediest, while privatization would mean those able to contribute the most receive the most benefit.
Obviously I part ways with Steuerle on the notion that SS's benefits are unaffordable and need immediate reform -- the numbers show that if the economy grows more than 2% per year (and you can't find a 20 year period, not even if you include the Great Depression, when it didn't), SS will be able to meet its obligations under current formulae for pretty much forever -- but it's useful to have somebody on the right pointing out to anyone who thinks SS is a "bad deal" that a) it's not supposed to be a "good deal" for most of us, and b) even so, it's actually not a terrible deal for anyone but the very-well-off, the top quintile or so.