Even in abstract games, I tend to favor very short rules, acknowledging the fact that some fixing may be needed to create a proper rule set. I prefer games that focus on one sort of interaction instead of several ones and exploit it elegant ways. Using this criterion, Hex should be the most simple and deep game there is. The case of Go comes near but... e.g., the KO rule is a kind of second-order rule. It is a needed fix for a nasty consequence of the capture rule and it would be hard to come up with a better replacement but, nonetheless, KO is not the essence of Go. KO also implies extra information besides the one in the board (like castling in chess players must have memory of previous moves or positions) which, personally, I find unaesthetic.
We may try to look to all the rules of a game and partition them into the following (arbitrary) classification. First-order rules are the essence of the game (and, possibly, are the criteria to define if a variant belongs to the family of the standard or 'essential' game). Second-order rules are those in order to make the game playable and its dynamics as enjoyable as possible. Rules like these, if well chosen, are a good pathway to create a fine abstract game (but, at this moment, starts a new business: make all those rules interact properly which is much more difficult than joining some unrelated good ideas in the same pot). There are also third-order rules: extra rules that provide little to nothing for the game essence except to contribute artificially to the game dynamics.
I would say, for FIDE chess, castling and e.p. are 3rd order rules. The piece movement of the FIDE army is 2nd order (remember Betza's different armies for other instances of chess soldiers) and also the one piece per turn move sequence (progressive variants still use the knowledge and instinct from chess experience). The concept of royal piece and capture by replacement would be first-order.