Couldn’t resist. Sorry.
1. It’s debatable whether publishing on demand has gone mainstream, but word processing, webogs and inkjet printers certainly have.
2. Plastics are complex but so is colour printing. If you can do even a small subset of what professional 3D printing needs with just 3-4 compounds then 3D printing can be the inkjets of manufacture.
3. Laser printers used to be expensive. Now mono ones are less than 100 pounds and colour ones are less than 200 pounds. Innovation and economies of scale can work wonders.
4. Plastics are large and intricate but then so were electronic components. RISC and miniaturization can work in both cases. You can fit a jigsaw in a smaller box than a poster, and it can be more instructive to put together.
5. Designing in 3D is really hard but this is what “commons based peer production” is for. Make 3D design copyleft and it may still be hard but it will be decomposible and improvable. Make copyleft distros of fabbable components and objects made from them and watch them grow.
3D printing won’t go mainstream any more than home photocopiers did. But we will get the Kinko’s, Lulu.com, YouTube and HP Deskjets of 3D printing, and these will change things as much as the Internet and free software have.