There are just the N ways PL research has found for implementing generics. Either they implement them, and tackle the complication one of those N ways entails, or they don't.
Besides, has anyone seen much experiments, with actual coding or at least design, for finding this "non-complicated" way anyway? Not just some post once in a blue moon re-iterating the basic well known tradeoffs.