One more vote for Palo being worth trying. We've had brunch and dinner once each on both the Magic and the Wonder, and have really enjoyed all of the meals. Here are a couple of "BUT"s though:
I would not plan Palo if you are in a time crunch. Especially dinner. I think our dinners have taken about 2.5 hours, and we did not feel we were going too slowly. It's kind of designed to be a slower-moving dinner. Brunch could go faster if you want, though you still should try to allow more time if possible.
One "surprise" we had at one dinner: for us, Palo was the "romantic, get away from kids for a while" experience, but for others, it's a "get together in a large group and have a good meal together" experience. We had to ask to be moved during our first dinner because the large party (which looked to have already gone through a few bottles of wine) was overly loud. Don't feel bad about asking to move if you are in a situation that is the opposite of what you're looking for - they moved us, and it helped tremendously. Just realize that although you might think you're getting the quiet dinner, there's a chance you end up with something louder...
There was a slightly negative comment about food quality above. I'll say that that has not been our experience: I have loved the meals we've had at Palo, and with the exception of a couple of buffet brunch items, I've liked everything I've had there a lot (and my souflees were great - not underdone). I'd put it above anything you're going to get at a chain restaurant, for instance. But, it's true that it's not at the level of some of the better independent restaurants in cities - I can easily name a couple of nice restaurant meals in the last couple of years that were a clear notch above Palo (and a couple that weren't!). These were the kind of meals where we'd pay ~$75 per person, which was part of how we decided how much we felt we should tip at Palo. I would not let that dissuade you, though - we felt the Palo meals were some of the highlights of our trip, and easily worth doing.