Clay isn't much of a fan of 3D and that's legendary. On the other hand, I can't quarrel with his skepticism regards SL. SL seems to be to real-time 3D on the Web what Netscape was to HTML browsers and may be headed for the same fate as the fad wears.
IBM's entry into the market is more serious. They see the same markets that some of the 3D veterans have known were there. The value of the interface is precisely that it engenders the sensation of presence, what the old simulation experts referred to as the power of onset cues to the human brain. So as a collaboration medium for work at a distance, it is better and possibly cheaper, maybe even more reliable than video conferencing.
Then you have to consider the 'off the web' markets. For example, the music industry can't seem to figure out how to keep CD sales from sagging although their immediate demise is exaggerated. The Wal-Mart economy can be bootstrapped by packaging music wrapped in 3D which any gamer can tell you is the cheapest 3D surround sound engine there is. This takes advantage of the phenomenon that once a luxury becomes a feature, it becomes a necessity. The technical aspects of this are easy and the production costs are the dilemma (3D content isn't cheap to build no matter who says it is). Products such as VCommunicator from VCom3D indicate what the military or Berlitz can do with the technology given good authoring tools. If you’ve ever sent any of your employees to India or China or brought them here, you know that the faster you can get past the cultural and language barriers, the faster you can get down to business and the better the results.
What SL and IBM indicate is that 3D is here to stay in various forms. It never went away. Clay was dead wrong about that and probably did give bad advice, but skepticism about the imminent second coming of the web in 3D is healthy. As yet another entertainment offering, it will have a season. As technical innovation in that market, consider it Technicolor or Panavision which had their day and passed into history as digital production overtook 35mm film, but people seldom go back to black and white except for art flicks.
Now it is a matter of smart money, branding and IP. Of those, the third is where the real trouble will start. Despite what Clay said about VRML being dead, there is only one real world-accepted standard for real time 3D. It is the ISO standard for X3D, son of VRML.
What remains to be seen is what the smart money will do about the existence of royalty-free unencumbered international standards for a technology they wish to burgle. If history is a judge, some analog of the W3C which burgled SGML from ISO to create XML will be tried. OTOH, Rome having been savaged once has had the time to get ready for the return of the Celts.