Obviously, the need for virality also differs based on what you are building. For social networks, having your friends there is a big part of what makes it a "good product". For a search engine, that's less important.
What bothers me is the focus on virality over actual value creation. The best examples are some of the "zombie" type facebook apps that are pretty much just pure virality.
And I think any product that completely revolutionizes a domain doesn't need to "go viral" to succeed - ultimately the excellence or game changing experience will cause these apps to succeed, Amazon, and the iPod are two great examples. Google was a success because they created a game changing search product - it was (and still is) vastly better then anything else out there.
I think however that while everyone tries to create a game changing product, the vast majority of products are really incremental improvements and in these cases going viral helps. Twitter being a great case in point - really just an iteration somewhere between IM/Email/Blogging - but went viral enough to be important, over some better platforms such as Jaiku.
Great point. He really skipped over Gmail's marketing strategy, which wasn't just being good. Gmail didn't get a lot of attention soley because of something as abstract as "it was good". Gmail combined artificial scarcity (invite only) and an outlandish offering (1 GB space) to spark demand. That is the core of the hype machine that drove the word of mouth. Being good was only a multiplier on that, not the base.
There are many superior products out there that have lost to the competition (especially in the social network space) because they never gained enough early traction.