I'd agree with that. Alexa traffic rankings are ballpark estimates, and I think the rolling 3-month average is useful if a site is in the top 50,000 or so. If nothing else, a ranking of 500,000 or 4,000,000 will tell you that a site isn't a player in its niche.
My own site ranges (in thousands) from the high teens to the high 30s, depending on the time of year. (Right now it's at 22,289, which seems about right for its current traffic level.)
We need to remember that most site owners aren't Webmaster World members, aren't SEOs, and have better things to do than sit around clicking on different computers with the Alexa toolbar installed or devising software to skew the Alexa rankings.
I think the biggest problem with Alexa rankings is that the user needs to avoid comparing apples with oranges. Alexa ranks domains, which means that (for example) a little-trafficked site called widgetbaking.about.com would show an Alexa traffic ranking of 58 (About.com's ranking), while a far more successful site called www.widgetbaking.com might show an Alexa ranking of 20,000 even though it has 10 times the traffic. This doesn't mean Alexa is defective; it just means that, like any tool, Alexa is most useful in the hands of people who understand how it works.