The enclosure opens at the back, too, so a long piece can pass through. The wooden chassis is such that it's incredibly rigid. I think it's one of the reasons the aluminum cuts worked out well in the vids I posted. None of the moving parts are made of wood, so it would be much less prone to structural failure - I would have NO worries about building that part out of MDF. Just seal it thoroughly and it's all good.
The belts are used to drive Acme thread on the X and Y axes,
* and I think the Z axis is direct drive to Acme. It's small enough that I don't think that upgrading to ballscrews would cost that much more. I've discovered that http://www.UGRACNC.com
is a good place to get that stuff.
I've made my decision: this is the design I'm going with for machine #1.
It's something I think I could build for under $2,000. Future rigs will be determined when I'm ready to do them (hopefully no later than 2014), but the Joe CNC 4' x 4' and the CNC RP platforms would be my only two serious considerations. I may just build as per the design without modifications. I would like to expand its cutting envelope a bit (24" x 28"), but I'll need to study everything fully several times to know where I have to add to things so I don't "under buy" materials.
* = Edit. Me being stupid. It is real belt drive, like you mentioned. The only Acme thread is at the Z-axis.