Printing isn't all that high a minimum; above about 5000 copies the price per copy doesn't drop nearly much as run size increases. The thing is, aside from the top few, 5000 copies is likely to be the enitre first year and half of the second.
The problem with hardcopy and multi-versioning is multi-fold, tho'.
1: inventory management
2: standards of the book sales industry (see below)
3: judging demand for the various versions before printing
Any time you have a physical product, you have inventory management issues. Much of the cost at POS is in fact inventory management and shipping cost.
The real problem is the standards of the book sales industry... which WotC, Paizo, FFG, and Mongoose all have to deal with. They want relatively standard sizes, to be able to return unsold inventory after some specified time, to not be charged for unsalable inventory that isn't accepted back. (It's what killed GDW. If there'd been Kickstarter, GDW would still exist.) They want on time delivery, too. And, generally, they don't want people being able to confuse one product for another, because that pisses off the customers.
Part of the reason for no print version of magazines now is that "return the covers and destroy the product locally to get a refund" mode that is standard for paperbacks and magazines. The publisher assumes almost all the risk, in exchange for a relatively thin profit margin. If circulation is high enough, the price per copy to the publisher for a magazine can be under $2... but that magazine, at retail, is at least $6, probably $8.
And judging the demand is always tricky.