I just saw this article on hackaday and twistypuzzles was the first thing I thought of! http://blog.reprap.org/2013/02/vapor-tr ... parts.html

I don't know how well this would work (if at all) with SLS parts.

It wont.

Nylon is not soluble with acetone or almost any other standard solvent. Chloroform softens it but I suspect not soft enough to smooth it at all. Chloroform is also very hard to get a hold of and I don't think you'd want to screw around with the vapors.

Edit: I'll also mention that I once soaked parts in MEK (the bad-ass big brother of Acetone) for about a week and it made them a tiny bit softer and more flexible but that went away when they dried out. They weren't soft enough for their surface to be smoothed at all. You'd need the surface so soft that you risk smearing or leaving fingerprints if you touch them.

it might help someone with a rep rap or similar, though. no real need for heated platform if you have a hot plate, but this would make the precision be, uhh, off.

Shapeways once offered a similar process as a trial material called "White Glaze". The only example I can find of puzzle parts smoothed using this process is in this post comparing different 3d printing processes. While it probably won't work for SLS, it's interesting to see that this is easy to set up for Reprap owners, but getting detailed enough results for puzzles will surely be a challenge.