Guild Wars 2 is scarcely a proof of conceptual failure. Its own developers have admitted to issues in execution and backing designs irrelevant to trinity or non-trinity systems that made their result far from ideal. And even after "patching in" tanks and healers it is far from being a "trinity" MMO. The tanks themselves lack any significant general manipulation utility and are generally just higher armor classes with an occasional during-animation DR and those "healers" are specific specs of specific specializations or else little more than hybrid tack-ons. It simply added bits of threat control, mitigation, and on-ally health restoration to the existent mix. It still depends primarily on evasion and target swapping to trade off threat where able and, sadly, death for threat-resets when not.
That said, if I were to look with a critical eye over various MMOs that didn't allot entire classes to, essentially, meat-tanking and direct allied health restoration, or did so to a far lesser extent than WoW, XIV, Rift, etc., e.g. GW2, Tera, Vindictus, B&S, BDO, DN, MHO, NNO the one thing that really sticks out isn't any necessity for passively-set or specialized "roles" in order to develop meaningful content. It's the need for capabilities for manipulation and consequent interaction (enmity, CC, and kiting, in some of their simplest terms), whether they be given to all classes or not. If there's nothing for your play to build around that of others, combat complexity falls drastically. For the more mechanistically-challenged, this can be substituted in part by filling hotbars with the stuff of 16-step openers and the like, even tacking on some intentional button-bloat to really fill them out. But less you allow for an actual effect on, from, and between group interactions (which the trinity can restrict at least as much as allow), the less you get out of each button, so to speak, and the less creativity and variance is made available.
MMOs are by no means a completed genre. A stale, convention-driven one, sure, but far is any from reaching anything ideal or pinnacle-like, even in their own narrow design philosophies. Why restrict all ideas, then, to what's already been produced in MMOs specifically? "Fighting games, MOBAs, shooters, [and] dungeon crawlers" all fall under, in one game or another, successful multiplayer titles. Even single player RPGs, for instance, can yet shed light on what works and does not work in swappable specialization in end-game activities, enjoyable environments, story-building, team interactions (NPC -> players), creating windows of urgency, and so forth.