Since brainstorming the original design, I've come to a couple conclusions.

Adding cards during play is probably too significant in a balance sense. You'd need to change several core mechanics in order to add them without causing issues. Likewise, it either requires us to slow down and go through a process (not ideal) or make these cards super-simple, which is not the goal of TPI.

Swapping out cards could be done as part of a storytelling conceit, but then you've got to question why you're playing such a game in the first place.

Formatting the cards is another issue. Poker cards are too small for most people to do by hand and have it work well. Fortunately, they're not a pain to do printing on (at least in terms of legible results). Originally my plan was to have a 52-card deck and be able to play with poker cards, but as I've watched the design evolve I don't think it'll work out smoothly.

My plan right now is to include the source files for the card templates so players can still add/remove cards from printings as they wish, and have an opportunity to print their own. Because none of the card creation will be during play, this gives people an opportunity to do this via computer, then print.

The biggest pain with 3x5 is that they don't print well. You get five cards to a page if you're going for efficiency, and since we're shaping up to have somewhere around a 60-70 card deck (this is an early estimate, and may not reflect the finished product), that would be a lot of printing. You can get 11 bridge-sized poker cards to a sheet of paper or 9 full-sized poker cards. I'm leaning a little toward the full-size poker cards, especially because they make everything just a little less crowded.

I'm probably going to wind up making templates for both down the road, but not immediately on release.