A more realistic way of measuring the comprehensiveness of the ODP is this: find a good, OBVIOUSLY unique website, then ask whether it's already listed.
Note that no e-sales website can be OBVIOUSLY unique, and that's where the affiliate/doorway/mirror spammers congregate anyway, of course, so you'd have to ignore them for the purposes of this test. Just pick any other ordinary commercial, artistic, or social website created without regard for getting an ODP listing.
There's still lots of work to do, of course: I estimate (very roughly) that there are a million or two of good sites still to list. And there are no doubt even e-tail sites that, upon careful investigation, will prove unique -- we will continue reviewing those, albeit not with the urgency and enthusiasm that we share for sites that may possess "obvious uniqueness."
But the difference between the approximately 40% of good sites listed (as in Yahoo) and 80% of them listed (as in the ODP) -- is pretty significant. I think it validates the open-content model of content creation as implemented by the ODP (and many other sites.*)
(*)Recently a Project Gutenberg project manager noted that they had virtually exhausted the canon of non-ephemeral English novels...and they have more volunteers who focus on general fiction than they have books to feed them. So such books, when available, are proofread at the rate of hundreds of pages a day. They whip through two rounds of proofing in a matter of hours.
On the other hand, the harder, more esotoric texts (Greek and Roman Literature, History of the Physical Sciences, Philosophy) are lucky to see a page or two proofed a day. These books may take six months to go through proofing. But they do get through eventually.
Do you see something of a pattern here?
Unfortunately, I'd been scribbling down all the good ones but the dog ate it... 