As someone who has iterated and launched a lot of stuff in the past year, I love this approach. I think 54 hours is crazy small to release a quality product, and dont know if I could do it, but Bubs is right with the framework. If you have to do it, this seems like a great method.
Regardless, it really isn't about that. It's MVP at its truest sense. Does the world need this product? That can be answered in 54 hours, with enough talent, caffeine and motivation.
To cram everything in, you cheat. Use Django, Rails (or better), Flask, Sinatra, whatever. Use an off-the-shelf authentication schema (Facebook, Twitter, whatever). Get a themed template from Themeforest or wherever.
If you're lucky enough to have a designer, let them put polish on it. If you're lucky enough to have (or be) a developer, let them choose the language and framework.
At the last Baltimore SW, Dave Troy built and launched a couple of addons for (the not yet launched) Shortmail[1] service. Yasmine Mustafa (seen on HN front page[2] recently) was building 123Linkit, which was later acquired, and the winner, Parking Panda[3] was built and demoed (and looked spartan, to say the least), and they were recently featured by Mashable.
Things can happen, and 54 hours is enough. Getting everybody involved to see that isn't necessarily the easiest thing in the world, but it's a great way of whittling down even modest ideas into their smaller kernels.
[1] - http://shortmail.com
Startup Weekend is great, if nothing more, than building something, meeting people, and maybe, just maybe, starting something great (Remember, it's Startup Weekend, where you go to start, not finish). The pressure of having something to demo, of being asked questions about the business model, of having to be judged in front of your peers is exciting.
I highly recommend attending one. Even if you aren't planning on doing a startup or don't have an idea, the connections you make, the things you learn, are awesome.