Their release approach was probably the key issue, besides the technical issues (primarily performance as I recall), that killed that project. I had access within a couple months of its release and had some number of invitations (single digit, 4 or 5?). I sent them all out, but the invitations were a lie. Invitations added your invitee to a pool of people who would one day, maybe, get an invitation if they won the lottery.
The result was that I was using a collaboration tool with no collaborators. By the time they announced they were killing it (about 15 months after launch, perhaps a year after I sent out my invitations), no one I invited had received access to the system. This crippled adoption for everyone who was interested in it, and reduced Google's interest in trying to salvage the system by dumping money into the engineering side (and Google already liked to kill products even by that point in their history).