Even games, these days, are often decided on because people can play with their friends or at least share and trade games amongst their social circle.
Actually, it just struck me how much less solitary computing is than it was even five or ten years ago. So much of what we do now is interacting with and sharing with others. Funny how that snuck up on me.
I think your criteria of "explosively successful" might require very strong network effects - it certainly requires some way to reach lots of people, and network effects are one way of doing so (the de.li.ci.ous guy mentioned 3 needed factors, and this was one - he also put a new-to-me twist on it. I'll try to find the post). If you relax it to modest, steady growth, it might not require strong network effects (and possibly might be more stable - because resistant to explosively growing competitors...).
He says one way is to make the product acceptable to more people - ensure it has something that each of them wants ("a bit wider"). Another way is to make it so compellingly wonderful for one particular person that they will "bully" their colleagues into getting it ("much deeper").
I'm not sure how applicable this is to software. Often, it's just one party that needs to be satisfied. But... it can also be an organizational purchase, where the users have to be happy, the tech support guys have to be happy, accounting has to be happy, management has to be happy, etc. Each party has sufficient power to veto a purchase (this isn't always the case!). Seth is saying you can sweeten the deal slightly, to try to please the hold-outs a little more (just one of: easier to use, easier to support, easier to pay for, easier to justify ROI). Or, the other approach, is to make one attribute so astoundingly amazingly good, that it's so far more than "acceptable" to one party that you galvanize them into action, to champion your product, calling in favours, making sacrifices, doing deals to get the other players to agree (like a senate, right).
He didn't say this, but I think he's implying it's easier to identify the hold-outs and just please them enough for the purchase to be unenthusiastically acceptable to them (Oh well, I guess we can get it), instead of exciting one party so much that they move heaven and earth for you.
I reiterate that the starting point is the very special case where several people have veto power, and the product is not acceptable to one of them.