I think the key thing with physical products is that there needs to be a reason for it to be a physical product. A coloring book, a pop-up book, a book of art photographs, or a book of fold up buildings is still viable.
A book where the main draw is the text isn't.
Games Workshop gets away with it because the physical quality of the presentation is key to the product. I don't think any games company has understood and implemented the integrated visual identity of their product so fully and it pays off for them. And even then they struggle though, I think their key assumption that they can increase their prices ad infinitum without consequences has probabably cost them money.
There's loads of RPG companies that still produce books as physical product, for various reasons. The main issue, as I understand it, is expecting to sell enough to get a decent price-point. It's not like you have to be Wizards or Paizo to produce physical books, either: Troll Lord Games, to pick one, does and their product is an old-school semi-retroclone and the associated adventure modules. For that matter, Adamant and other Savage Worlds publishers (like Triple Ace) have been getting physical product made. You don't even have to do it yourself; people like Cubicle7 act as printing partners (or if you're not going to sell enough to interest them, as to-order printers). Some indie games have been breakout successes in print (Dresden Files, to pick an example; first run was, I think, 5000 of each and they made their money back almost on the pre-orders, let alone when the entire run sold out*).
I'm not saying it'd be right for ICE, as I don't know what sort of numbers they can reasonable expect to sell, but it's not as if physical product is only for items where the primary value is non-text, or for companies like GW, WotC, Paizo, etc.
*5000 is a lot, particularly for large quality-layout and colour illustration products like DFRPG. My understanding is that a print run of ~750-1000 is pretty much necessary to get value from a printing order or else you just go PoD.
We will be producing print versions of products wherever it is feasible, and we have access to sufficient sales data to know what is possible, and I personally have sufficient contacts in the industry to discuss the possibilities of printing partnerships with the right people.
For old products where we only have scans, pdf is all there is ever likely to be. It is technically possible that some of the scanned pdfs could go through the OBS/LightningSource PoD process but the results will
not look good at all. It is extremely unlikely that any of those products would sell enough to make it worthwhile doing new layout and new art.
For existing products where we have at least partial layout files, we will be trying to make them available as print-on-demand. As these products have been on sale in some cases for many years, print runs again don't make sense at this point.
Currently the strategy with new products is also to go print-on-demand as that allows us an opportunity to provide printed products without significantly adversely impacting the opportunities of future print runs and gives the new cohort of authors a chance to actually write a stream of new products.
Best wishes,
Nicholas