Just as a point, when you say failure, you mean sales failure right? Of course, by that metric, virtually every single RPG on the market, other than D&D, was and is a failure.
You're only a failure in terms of sales if you either don't sell enough to turn a profit (which has a lot to do with your costs, as well), or if you don't sell enough to meet a make-or-break sales goal. 4e definitely didn't hit it's minimum sales goal of $50 million/year, which is hardly surprising, no edition of D&D has ever pulled down that kind of income... if one ever does, it'll likely be able to thank inflation for it. ;P Hasbro never shared it's sales volumes, costs, gross revenue, &c down to that level of detail, so we can't know if it was covering costs or not, but it probably had high development costs for a D&D ed, it certainly seemed to have a lot more folks on the payroll, for instance. But that's moot, because it didn't come near the low end of it's goal, making it a failure.
Other games have made less money on lower volumes and been overwhelming successes, because their costs were low and their prospects (goals) non-existent.
5e's development costs must have been pretty low - just about everyone was laid off and it's mostly retreading past editions' material - so even if it had sold every bit as badly as 4e was rumored to, it'd've likely turned an easy profit. Instead, it's doing wonderfully well by any reasonable measure (only the fad years were better!), but if it had promised to beat the income stream generated by WoW, it'd still be a dismal failure in the boardroom.
::shrug::
Which, of course, runs into the whole gnome effect problem. Just becauses 10% of your audience doesn't like something, that means that nearly 50% of your groups have a problem with something kvetching about this or that change.
Actually, the gnome effect was 10% of your audience liking something and kvetching about it being gone. (Probably more disliked gnomes than liked 'em.) If you dislike something in 5e, you opt out of using it (or, more likely, don't opt in, since so much is optional). But if it's absent, adding it is a much bigger issue.
It's a tad ironic, since the gnome that supposedly caused all that wailing and gnashing, was added back to the game in PH2, within 9 months, probably had been part of it since before the kvetching started, considering lead times.
Healing surges are no longer connected to virtually every source of healing in the game. That's a complete game-changer to me.
To be fair, 4e had rather a lot of 'non-surge' healing, just consistently as daily resources - and far fewer 'slots' available to devote to that than 3.5 or 5e have. The proportions were different than they are in 5e - 4e having had more healing available in surges relative to the very small number of 'slots' that could be devoted to non-surge healing (and had to be chosen at chargen/level-up), in contrast, 5e has less than half (as little as a third or less for fighters, for instance) of the relative healing potential of surges in HD and it's much less accessible , while the number of daily slots is many times what it was in 4e and, if a caster knows even one healing spell (or just preps a healing spell at the start of the day, for the Cleric & Druid), all those slots can be expended on healing, spontaneously (or used for other daily spells). It's not really a huge difference in the nature of healing - healing can be 'natural' from surge/HD resources of the character healing, or from daily resources of a 'healer' character in both cases - but in the versatility and availability of those resources. HD are even less available (fewer, slower to recover, an hour to access) & versatile than surges (which had very few, but some alternate uses), and slots are far more versatile (useable for any spell spontaneously) and available (numerous) than 4e dailies.
So, yeah, HD - a form of non-magical, universal PC healing resource - are 'retained' from 4e healing surges. They're in a bowdlerized form appended to a classic label & mechanic (random hp generation), and don't contribute to class & encounter balancing &c the way surges did, but the provenance is clear.
The squishy skulkers (warlocks, rangers & co) must now emerge to take on a few hits. Why? To spread the damage evenly among all party members. This also felt utterly artificial.
HD represent an even less flexible resource - they're only for healing, and only useable at short rests, which some party members will likely want very much to take a couple times a day - so /everyone/ can access their HD, every day and have nothing else to do with 'em, and if someone isn't 'pulling their weight' in that regard, the short rest could be a waste of time for them.
I suppose it's only a matter of time before 5e adds back Comrades' Succor and allows you to shuffle HD around, though.
I haven't seen anybody mention exceptions-based rule design yet. ...5e feels like it's just as deliberately a cascade of smaller rules like 4e was.
5e does contain the specific-beats-general disclaimer of exception-based design. It doesn't seem to cleave to it as religiously as 4e did, but, yes, it's there.
Another one that I don't think has been mentioned yet:
eliminating the mechanical impact of Alignment. In 3e, there were grid-filled 'team alignment' spells and items, each axis had it's own zap-the-opposite set of toys that worked similarly, and, of course, D&D had always had many alignment restriction. 4e simplified alignment from 9 to only the 5 least-unintuitive alignments, and eliminated mechanical affects and restrictions of alignments. 5e reverted to the more complicated 9-alignement system and re-instituted some alignment restrictions (usually in softer form), but it did not return to the 3e 'team alignment' extreme.