Even outside this pathological case (where depth scales inversely with complexity), the general case is that depth scales slower than complexity. Doubling the complexity (by some measure) may only increase the depth by 10%. This is one of my big gripes with 3e D&D--tons of complexity and options theoretically, but most of it is utterly pointless. It's just noise to be dug through by people who want to actually play the game. Or, worse, fodder for "look at my system mastery" comparisons (ie letting the experienced players say "git gud nub" when a new person falls into one of the traps the system digs for people). Plus, increased complexity often leads to unintentional dominant choices. Where by combining X, Y, and Z, you get something that the designers never planned for and which breaks the game unless the DM accounts for it (which often leads to arms-races and party struggles as one person's character warps the game around them due to mechanical prowess). That kind of thing results in even less actual depth, because with dominant strategies like that, your options are
1) play to those dominant strategies both as a player and as a DM (resulting in a much more limited palette of choices)
2) everyone agrees to avoid those dominant strategies (which is fine, but often takes system mastery because you can accidentally break things)
3) have a mix and suffer unless everyone's fine with playing BMX Biker and Angel Summoner.
A lot of classic monsters have to be completely rewritten or just left out--giving everyone and their mother spells so they can keep up with the T1 casters is one example of this adaptation.