But I wouldn't be surprised if the multi-process architecture and the new JS engine turn out to make a surprising qualitiative difference in the performance of certain classes of JS apps.
Maybe so, maybe not. But even though I'm not writing anything particularly complex -- like, say, Gmail -- I've already had the thrill of working around situations where my one-lung JS interpreter is holding up the whole browser.
Heck, perhaps Gmail itself will be significantly smoother. Given the market penetration of Gmail, that would be a pretty good marketing hook for Chrome, all by itself.
I'd hazard a guess that Chrome may have a technical advantage over safari / ff / opera for almost 12 months.. (To be specific - over end user stable versions --> not install-you-own-renderer or use SVN to grab the nightlies)
And especially if it's a Google Killer app - How do you beat their distribution?
But now Google are offering a genuine technical advantage to go alongside those all important standards. It may be the "killer" platform that makes a reality of "zero install" web browser applications for business.
Google already has deals with manufacturers like Dell and Lenovo (I think) to install its desktop search software and Google Toolbar on their computers. It won't be too hard for them to include Chrome in this bundle as well.
Chrome itself will be the first attempt at that killer app.