The MtGO game was constrained by remaining backwards compatible with the table-top game, overlaid with that damn fool idea about exchanging the virtual and real cards[1]. That it worked at all is basically a software engineering miracle.
For me the real challenge, when going from tabletop games to computer games, is that in a table game it is trivial to have a rule-modifying rule; in a computer this is a source of substantial complexity.
Think about it: most of the player genius in CCGs is deck construction. That is, studying the rules, studying the cards, and mentally simulating the interaction of multiple rule changes.
Most computer games don't have rule-changing rules, or very few of them due to the cost of special-case coding. They generally resemble CRUD databases with some element of chance applied.
To me the central, and interesting, challenge for a computerised CCG is to capture that rule-modifying rule capability. This might be an area where the dynamism of Lisp is perfectly suited.
[1] It reminds me in retrospect of that camera Kodak produced which took film pictures, but had a screen on the back to show you what you'd taken. That is, a self-defeating and half-hearted attempt to deal with a disruptive change to a fat, profitable industry.