Firstly, thanks for all the input on the drop duration. Sounds like 2 weeks is a good consensus number; we'll go with that.
Now onto a much bigger (and, frankly, at this point somewhat comical) issue.
Technical constraints and ABS DSI have been working with the engineers at SP right now on the finalized designs, and it turns out that the doubleshot molding process is actually very ill-suited to our particular legend set. I'll give you the TLDR on the technical constraints below, but the short version of the story is:
if we want good looking legends, we might actually need to switch back to PBT! 
I am trying to speak to one of their engineers directly right now, and I am going to do everything I can to see if ABS DS will work, but given the technical document I just got from SP, it's not looking good.
Given that PBT actually won the vote by a small margin, I presume this is going to be fine. But I wanted to run it by everyone just the same.
We've gone back and forth on the issue so much that I've kind of forgotten the details and considerations, and want to know if anyone sees any problem with PBT that I'm not remembering. :S The big problem with PBT from a design perspective was the awful orange color, however, so if we have to use PBT I'm going to see if they can make us up a custom "spot color" yellow from a Pantone swatch, even if it costs a bit more. I was also originally concerned about legends being more blurry with dyesub, but I have since seen (in person) some SP dyesub work and it appears to be excellent and crisp, even if not quite as much as DS.
Price: for this particular set, the cost is more or less identical, so no worries there for people participating in this GB. In the long term however, it means that subsequent rounds will also cost about the same, whereas with ABS DS would have cost less (by virtue of being able to re-use the molds).
TLDR technical stuff on why ABS DS is a problem for our set.
So, the way that doubleshot molds are made is through the use of a robotic rotary cutter with a round tip. Rather than defining vector
outlines, you can thus only define a
centerline along which this round cutter tip will travel when making the mold. Ends of lines using this technique are thus invariably rounded, and it's very hard to make hard corners and angles, unless they are the sides of a straight line. So, for example, on an octothorpe ("#") you would get crisp and hard right angles at the four points where the four lines meet, but the endpoints of each line would necessarily be rounded. You can also not vary the width of a single line within a single cut.
Have a look at the following doubleshot legends, given the above, and you'll be able to envision clearly how the mold-cutting process that made them works, and why they look like they do.
Notice how every endpoint is rounded, and how generally rounded things are that would otherwise be right angles in a conventional design. I have always wondered why doubleshot keyset legends have had this rounded look, and I always assumed it was an intentional choice of the set designers. Well, I guess now I know. =\
Of course, the problem with all of this is that the canonical "LCARS" font (namely, Bitstream's SWISS 911 Ultra Compressed) has lots of hard endpoints, which are kind of important to the characteristic and futuristic look of the font. There are also other issues with doubleshot as it applies to our set (the numpad as originally envisioned wouldn't be possible, since they can't bring doubleshot legends to the edge of key faces).
By constrast, dye sublimation is an actual printing process, so there are considerably fewer constraints. Here is a SP dyesub key:
This seems much more likely to yield faithful results. Regardless of how one might feel about doubleshot vs dyesub or PBT vs ABS, I think the most important thing for this set is not just a set that looks inherently good, but one that expresses a particular design idea and theme that is faithful to the original that we are trying to evoke. So I have tried at every point in the process to favor whatever manufacturing processes and materials that would give us the best result along that specific axis, and that's what I'm recommending now.
As I said, this is an in-process discussion with SP, but I wanted to solicit feedback and thoughts to bring to my discussions with their engineers and Massdrop as we figure this all out.
Thanks!
