The author similarly leaves out what I think are the most important pitfalls. Namely, that Facebook can simply decide to end FBConnect at anytime or make it so much less usable (like they did to applications) that it essentially kills you. The more users you get, the more valuable that data is, but with FBConnect you only have that data so long as you're on Facebook's good graces. Since you don't have any identifier (email, username, etc.) if Facebook terminates FBConnect, you may still have the rows in the database, but now they're not mapped to anyone since you no longer have access to Facebook.
Likewise, the more data you give Facebook, the more of a business advantage you give them should they ever decide that your time is over. They can easily track what sites are hot, how users use the applications based on what gets posted to walls, etc. Then they can run you out of business with something better integrated - like they did with some apps.
Without email addresses, you can't contact users. I hate sites that constantly spam me, but some email is valuable.
Finally, you're saying that your site isn't good enough for users to spend the time to open an account there. If you've made something good, your pitch shouldn't be, "well, it's really only worth using because you don't have to put in a new password." If you aren't getting an email from that person, then a registration is no-hassle (like here at HN). I put in a username and repeat a password twice (nearly as easy as logging in).
FBConnect really just looks like an attempt by Facebook to lock out competition both in the social network arena as well as the application arena. They can easily target your users for their own replacement of your site, they're getting your data on what's popular, they're in a position to shut you down should you impact their business plans.