I just don't believe you need pre-plotting to get the good stuff you describe. That's never been my experience.
Well, I think part of the problem may be what different people consider to be railroading. Personally, I don't necessarily think its railroading for the DM to keep his plot lines running per se so long as you are not forcing the party down a given path.
Take the evil baron vs. Sigil example above. The PCs have reason to believe that the evil baron is up to no good. However, they find a cool little trinket/ritual/whatever that lets them head off to Sigil so off they go. Now, the DM is left with three choices regarding said evil baron. Option 1) Make it so that the players really don't want to continue their planar adventures (i.e. "Ehhh, yeah there are a lot of people here, but there's a surprising amount of order here, nothing really for you to do, unlike back in Evil Baronville.") Option 2) abandon the baron plot line and come up with a new one that involves Sigil and the planes. Option 3) come up with a way to tie the planar adventures into the events going on in Evil Baronville.
Perkins is advocating Option 3. Option 1 is far and away the most railroady. Its basically saying "Sure, you're free to make choices but the game will suck if you make the wrong ones." Option 2 definitely puts things in the hands of the players (a good thing) but also runs the risk of really having a schizophrentic campaign where one session may or may not flow into the previous and following sessions. Each session stands on its own but is not necessarily related (though they certainly can be).
Option 3 on the other hand, finds a way to combine the two and get the best of both worlds. Maybe the evil baron is being influenced by a powerful devil, or a cleric of Tiamat, or whatever. Maybe the PCs start to uncover a plot by Vecna-ites to kill the Lady of Pain and learn that one of the keys to the plan is securing an artifact located somewhere in Evil Baronville, etc. The idea is to have a basic framework for where the campaign is going, but to leave the "How to get there" in the hands of the players.
Personally, my experience has been that my players, if left to their own decisions, will make none. I love the idea of cooperative world building. Unfortunately, most of my players are merely interested in getting together for a few hours every couple of weeks and only having to put in a few minutes of work before each session. There's nothing wrong with that. The only problem is, if I don't start tying their decisions into the overarching plot, we'll end up simply doing a bunch of random encounters.
Now again though, I think the idea is to let your plot evolve with the player's decisions. The decision to go to Sigil needs to become important rather than "You fools, why are you going to Sigil?" Doing this, you can make the players feel as though they are driving the plot while you are subtly drawing them into yours (even if that plot is a bit different than it initially was). For me, this will create a much more compelling game than if I were to simply improvise based on the decisions the players make each session. Some DMs are able to provide great games in that context, but a lot are not. Without some direction, most plots will unravel fairly quickly imho.