There were a few reason why 4E failed. Two of the key ones from my perspective as a player:
- Insufficient play testing, so excessive errata. It got to the point where I could not make a character using the books.
- Removal of free access to the online platform to create a character with the class abilities with errata.
Correspondingly, from my perspective as a player, a couple of the reasons why 5E D&D has done well is that it has had better play testing, less errata, and the
https://5e.tools/ is an outstanding resource when building a character. Now I would also love a good free character builder that has access to the latest errata, and that is coupled with something like 5e.tools, but that's not a (lack of subscription) revenue model that Hasbro will want to support. Good virtual table top support also makes a big difference with 5E - that has been a minor pain point with A5E, and a more significant problem with a RPG like Vampire the Masquerade V5 that just doesn't have a large enough player base for Fantasy Grounds to get a smooth and easier to use implementation of V5.
Mechanically, 4E has some great ideas, like daily, per encounter, and at will powers. It also had much better support for high level play, and the way they handled multi-classing looked decent, though I never really tried it sufficiently to see how well it worked in practice.
So for me, 6E will need a good free access searchable with keywords (both inclusion and exclusion), virtual table top support, and be a solid system without excessive errata. Switching to have some per encounter powers would also mitigate the short rest and nova issues that we have today with 5E D&D. I also would prefer to abandon the long rest Vancian spell memorization/preparation mechanics, and have more classes like Warlock, with most common powers refreshing on a short rest - high level powers or novas being daily is fine.