In other words: Be excellent and one thing, but make sure that one thing is worth being excellent at.
Trying to be all things to all people is a great recipe for failure, but it's not the only recipe for failure. There are lots and lots of ways to fail.
Amazon worked; bookstores are a very viable market. Ebay worked; flea markets are common in every town. PayPal worked; everyone has a checking account.
There are also plenty of pet stores. Pets.com could have easily been a success if they had built their company properly.
What you're saying seems to validate the article's point: execute a proven "commodity" idea well, and you can succeed. Bet the company on an experimental vision, and you risk failure.
Building a pet community and then up selling stuff is an ok idea, but it's also a different idea than just creating a pet store online. Limiting yourself to just selling generic pet stuff without said community is a poor idea.
PS: If you go to pets.com now you see the second version, but they initial push was all about selling stuff and not building a community. The issue is one of trust, if you sell yourself as a store it's much harder to build a community than it is to have a community and then sell stuff (the good) stuff
If you love pets and have great connections in that community, it's a crap-ton easier to build a "pet stuff" store than a "everything under the sun" store.