With a recent suggestion from Blizzard for ongoing ideas, I'd thought I'd post my view on how to fix the third of these flaws.
The Missing End-game
It's been a long acknowledged paradigm that players will consume content faster than developers can generate it. However, this is only really true when we talk about static content, and it's a realisation that is commonly addressed in other ARPG's now.
To provide end-game PvE content that can be generated faster than players consume it, you need to design randomised end-game content.
One of the design flaws in the current iteration of the Diablo series was to forgo the sort of dungeon randomisation that earlier entries in the series were known for and that are common in modern ARPG's. It's likely too late to correct this flaw, however, the following system could still be put in place to ensure that the generation of end-game PvE content always equals its rate of consumption.
Obtaining the random end-game scrolls
A new drop could be added to Inferno, a "dungeon portal scroll". The scroll, as the name implies would allow a one-time access to a single dungeon level, similar to the current system in place for the Hellfire rings.
A newly dropped scroll would have only one attribute, that being the name of the dungeon floor to which it grants access, for example "Keeps Level 2". This would be your standard "Keeps Level 2" layout, with the exception that there would be in lieu of an entrance and exit, a portal in point and portal out point. Thus the scroll grants access to one specific level of one dungeon only.
The scroll would not be bound in any way and could be traded on the Auction House.
Boosting the random end-game scrolls
Having obtained a scroll, the scroll could be enchanted (making use of one of the artisans), with a chance to add a random property either making the dungeon more or less valuable. For example, below is a short list of potential suffixes that could be used (only the positive ones are mentioned, but you could see easily how negative suffixes could be generated).
* 20% increased gold drop
* 20% increased magic find
* 20% decreased mob speed
* 20% decreased mob damage
* treasure goblin added
* 20% increased legendary drops
* 20% decreased melee damage
* 20% decreased ranged damage
Each enchant of the scroll would require a certain amount of money or ingredients, and the scroll could be endlessly enchanted (with an increase in cost to each enchant - this providing a useful gold sink for the game). Similar to current loot itemisation, suffixes could be doubled up (for example 40% increased gold).
Like the initially dropped scrolls, these scrolls could be at any time traded (to be sure, certain AH upgrades would be required to handle these transactions). Indeed there could be developed a vibrant market for end-game scroll trading, and if the range of suffixes is made suitably expansive enough, you could see scrolls with certain suffixes becoming preferred for certain classes even (consider the value Monks or Barbs would place in reduced melee damage dungeons for example).
Using the end-game scrolls
Scrolls could be activated at a main town in the appropriate act, and could be used in public games as well as private ones.
Thus teams of players could work together to construct through team effort a suitably tricked out dungeon scroll that they all partake of together.
The scrolls could be activated and used on any MP level.
While this system would work better with genuinely randomised dungeon maps, the proposal above makes use of the current static dungeons, only modifying certain global properties (such as drop rates or damage levels). While not trying to underestimate the volume of work described in this proposal, using current static dungeons would at least greatly reduce that development time.
Without a system like this, the game will always suffer from an asymmetry between the rate that players consume content, and the rate at which that content can be generated.
I take no credit at all for this idea, because systems like this are becoming much more common in ARPG's these days, indeed are used by ARPG's competing directly in the market even as I write this post.