If anybody is guilty of cherrypicking returns, it's you

I have every right to cherry pick when demonstrating cases when stocks perform far differently from what steady growth models predict.

no one is recommending investing solely in the NASDAQ

So let me get this straight- you don't believe past performance of a single stock will predict future results, or that the past performance of a single index will predict the future. But you do believe that the past performance of a single asset class, stocks, will predict future performance. Sounds like selective faith to me.

the compound annual return for the S&P 500 for the 5-year period ending 3/31/04 is -1.20%, which is hardly catastrophic,

It wasn't -1.2% for the retiree withdrawing 4%-5% per year. He would have lost 30% or so of his wealth during that period.

However, long-run returns averaged over the bull market and bear market will still generally be pretty good. For the 10 years ending 3/31/04, the compound annual return for the S&P 500 was 11.68%

Again you're mixing past and future tense, assuming what's happened in the past must reoccur in the future. What has the Nikkei's 10 year return ending 3/31/04 been? Japan's demographics are leading ours by about 20 years, so this may be a better predictor than the S&P. Or do we get to ignore stock market returns that don't fit your model?

You seem very concerned about risk, but I think you neglect to consider the risk that a portfolio will not have a positive real (i.e. inflation -adjusted) after-tax return -- i.e. the risk that you will lose purchasing power over the long-run

Inflation is certainly a risk- and a portfolio with 50% stocks 10% REITs, and 40% bonds (some inflation protected) as well as home equity gives you quite a bit of protection against inflation. But you're not keeping up with inflation when the stock market drops by 50% either.

Stocks give you the greatest probability of staying ahead of inflation, especially after taxes, over the long run.

No, TIPS and I-Bonds do.

Nick