Great advice. There are two prototype videos of my game Shadow Physics on youtube (http://www.youtube.com/watch?v=Bb5DjyoDObA), which is the popular one that demonstrated the concept to the wider public and gained enough interest to secure funding and allow me to work on it full time. The other is the original 3 day prototype I hacked together during TIGJam (http://www.youtube.com/watch?v=Jcjxo9svhxw). Not as impressive to anyone but myself, but it proved that the tech and concept were viable and got the entire project going.
These were both made after about a year of me talking about how awesome the game would be, but of course no one quite got it until they could see it in action. The power of a strong prototype is immense.
Does this advice apply to games that aren't so much fun, as addictive (i.e. Farmville)? I'm guessing most people would watch a gameplay video of a social game in boredom, not seeing the draw to it, and would need to play it themselves to "get it."
As always, don't consider this advice to be a golden rule or silver bullet.
Separating fun and addictive can be really tough. I think many people show signs of addiction when playing their favorite games (which are probably the games they say are most fun). Just because you don't think Farmville is fun, doesn't mean it's not fun for someone else. Even if we assume that the tens of millions of Xbox owners all dislike Farmville, that's only a small subset of the hundreds of millions on Facebook.
Also, browser games can sometimes be less complex than a 3d AAA PS3/Xbox game that's built "from scratch". If users need to play it to "get it", you might be better off skipping the video and moving straight to creating a prototype that showcases the addictive features.
I haven't played any of these games, but it makes sense in a couple ways.
For one thing, a prototype can suggest something compelling without necessarily delivering everything. I found this particularly true in the Limbo example. I found myself intrigued, thinking about the idea of a whole game that worked like that. It was strongly suggestive. But the reality is that it's just a short clip, the rest of the world doesn't have to actually exist. The ultimate release is likely to to make a variety of tradeoffs to realize the suggestions, which imagination doesn't have to worry about.
For another, a prototype only has to appeal to a limited group of people. The important part is selling or proving the concept. Customers represent a much wider array of tastes and while many will be drawn to a good concept, artistic elements are important too. If you're the sort that's more attracted to a clever concept than the overall experience, a prototype is more likely appeal to you in particular.
These were both made after about a year of me talking about how awesome the game would be, but of course no one quite got it until they could see it in action. The power of a strong prototype is immense.