I was mainly referring to the observation that while digital manufacturing in general is considered the 3rd industrial revolution it's 3D printing in particular that's driving it. Also, if it didn't come across I really like CNC machining and I think it will have an important part in manufacturing for years to come.
1. Limitations on geometry
The set of shapes that can reasonably be machined certainly contains a lot of useful items. However, the limitations that the machining process places on the design are subtle to learn and sometimes limit the process fundamentally. Of the examples you mention dental implants and jewelry are actually rapidly moving to 3D printing specifically because of geometry limitations. As an example pretty much all hearing aids are nowadays 3D printed. The geometry limitations are mostly driving adoption of 3D printing in high end applications like human implants, aircrafts, complex machine parts etc.
2. Tooling cost
I think there is a slight misunderstanding. Tooling cost does not refer to the cost of the tool but to the cost of changing the configuration of that tool from manufacturing item A to manufacturing item B. For machining centers that cost is very high, think of it as the work you go through when you switch milling bits, workpiece holding, alignment etc. For 3D printers that cost is virtually zero.
As for material choices I have several copies of the croco fob in question printed in steel. They were ordered through a 3D printing service and costed about $30 per piece. Aside from the fact that the croco fob can't be milled there is nowhere I could order three of them for that price.
3. CAD software/Tinkercad
We could certainly add CNC toolpaths but there is a lot more people who want to 3D print than mill. Given how easy it's to 3D print through one of the printing services compared to the hassle of going through a machine shop I can't really blame them. The main benefits of CNC toolpath generation would be home machinists, like myself, who could then use Tinkercad. That group of people is pretty small.
People can draw things and make a machine make them already.
This is the part where we disagree. Or more properly I think your definition of people is a few magnitudes smaller than my definition of it. To use your analogy, machining is like plotting and 3D printing is like raster (matrix) printing. I think calling it an implementation detail is missing the point.
Example this chair I just made (designed by Phil Seaton.) http://www.fabsie.com/blog/2012/03/04/scissor-chair-phil-sea...
2. Tooling cost of switching tools is only zero because you on about using one tool to 3D print. Could speak about various aperatures of tools of 3D printing various types of plastic to chocolate to ceramic. The main reason you change tool sizes is to increase speed, i.e. bigger router bits cut faster and you don't want to spend 3 hours to do something with a 3mm bit that can be done in 12.5 minutes with a 20mm bit. (opportunity for 3D printing to grow.
3. CAD Software is ubiquitous.
I don't think speculation about what will be a revolution can be correct or incorrect :)
Limitations on geometry.
Definitely true, though it seems to be good enough for machine parts, dental implants, jewelry, and so on. (That YouTube video of the CNC-milled chess pieces is pretty cool too.)
For all the geometry advantages of 3D printing, there are also disadvantages. The materials you can use are limited. Pieces coming out of the printer require finishing work, especially at the low end. Mills are more versatile and produce mostly ready-to-use pieces.
Both 3D printers and mills are inefficient for making more than one copy of a part.
2. Tooling cost.
Tinkercad links to the Makerbot Replicator, it's $1800. Roland makes hobbyist CNC mills for $900. I've used a cheap 3D printer before, and it can't produce pieces that are anywhere near the quality you advertise on your site. They look like a very drunk person vaguely sketched the outline of the piece with a hot glue gun :)
3. CAD software. ... This is a meta observation that "I drew in my CAD program" is often the hardest problem to solve.
I don't understand why your software can produce 3D printing instructions but not toolpaths for CNC mills. Once you've made a nice UI layer (which you have) it's just another algorithm on the backend.
Anyway, I don't think the 3D printing is going to be a revolution. People can draw things and make a machine make them already. The rest is just implementation details. (2D printing was a revolution. Cheap inkjets were just a nice implementation.)