The other answers to this question are very strong, but are also somewhat frame challenges. I'll attempt to directly address the two parts at the end of the question.
Why Are People Challenging the Frame?
Tier rankings for classes have never been especially popular in the 4e optimization community (rankings for finer-grained things like "level 13 ranger powers" or "wizard paragon paths" yes, classes no). I would attribute this reluctance to the lack of clear distinguishing markers for tiers that existed in 3.X. It's easy to point to the specific class features that make a 3.5 wizard tier 1 versus a 3.5 sorceror being tier 2. There's no corresponding bright dividing lines in 4e, because so much more of a class's potential rests in combinations of powers, feats, paragon paths, and epic destinies. How many tiers there should be for 4e classes and which classes are in which tier is based much more on personal opinion and which criteria an individual optimizer wants to use than would be the case in previous editions.
If there's a weak link in the roles system, it's the controller role.
- Leaders: healing and buffing
- Defenders: draw foes' attention and force them to make hard choices
- Strikers: deal damage
- Controllers: AoE for weak opponents (especially minions), battlefield control, and debuffing
What makes controllers stand out is not the value of what they're intended to accomplish, it's that their classes don't go the extra mile to help them accomplish it. Every defender has a way to mark, every leader has a way to heal, every striker has a class feature that boosts their damage. Controller classes, on the other hand, don't have clear class features to help them do their job; instead, they rely on their powers, which can be poached or competed with by other, non-controller classes. It's not that the controller job isn't important, it's that you don't necessarily need a controller role class to do that job.
In general, optimal role distribution looks something like this:
- 1 leader, possibly 2 if they're both buffing/enabling focused (particularly "switch" builds designed to hand out basic attacks to allies) and you have at least one striker built around powerful melee basic attacks
- 1-2 defenders; if 2 defenders, at least 1 should be more damage-oriented
- 0-1 controllers, depending on whether someone wants to play one and how well the other classes can cover the role
- everyone else as a striker
Alpha-striking is very viable in 4e; you bring just enough defending/controlling to keep foes that can't be burned down immediately from shredding your squishy characters and just enough healing to get through the day, and the rest is as much damage as possible as front-loaded as possible. That said, a well-planned 2-defender 2-leader 1-controller party can be just as effective as a 1-leader 1-defender 3-striker party; it won't get through battles as quickly, but it will generally be safer.
A Brief Discussion of Optimization
Where the tiers that are typically used for 3.5 largely assume similar levels of optimization between the classes, attempts to rank 4e classes have often taken optimization potential into account. In this context the discussion is usually about a class's "floor", i.e. its effectiveness using a fairly standard build with no emphasis on optimization, and its "ceiling", i.e. its effectiveness when heavily optimized to squeeze out as much capability as possible. Classes that have more options available usually have higher ceilings (though this is not always the case).
Rankings of individual classes also don't necessarily take party composition into account. For example, several leaders get a lot of their strength from the ability to hand out free attacks to other party members; having a striker with powerful melee basic attacks in the party significantly boosts these leaders effectiveness, while a party with nobody capable of making a decent MBA will reduce their effectiveness. Other classes may not be as strong individually but lend themselves well to certain party-level optimizations (such as a Radiant Mafia party).
This is all based on my rather sketchy memories of when I played a lot of 4e. Take it with a grain of salt, especially the relative ratings of the Essentials classes. If you'd like to argue with me about the ratings, drop me a comment and we can set up a chat room.
Classes will be presented with the following additional information: (floor->ceiling, any viable secondary role focus)
All leaders heal at least decently; a good leader boosts the party's damage output.
- S Tier: Warlord (mid->high)
- A Tier: Artificer (low->high), Bard (mid->high controller)
- B Tier: Ardent (low->mid), Cleric (mid->mid), Runepriest (mid->mid), Shaman (low->mid controller), Warpriest (mid->mid controller)
I'm not familiar enough with Skald or Sentinel to rate them.
All defenders force foes to choose between attacking them and attacking the squishies. Good defenders make both of those options as unattractive as possible, or just outright make one of the options unavailable.
- S Tier: Fighter (mid->high striker)
- A Tier: Paladin (mid->high), Swordmage (low->high), Warden (mid->high)
- B Tier: Battlemind (low->mid striker), Knight (mid->mid)
- C Tier: Cavalier (mid->mid)
All controllers have AOE. Good controllers also have hard status debuffs (blind, daze, stun, dominate) and battlefield control that prevents foes from reaching anyone in the party.
- A Tier: Invoker (mid->mid), Wizard (mid->mid)
- B Tier: Druid (low->mid striker), Mage (mid->mid), Psion (mid->mid), Sha'ir (mid->mid), Witch (mid->mid)
- C Tier: Binder(low->mid striker), Bladesinger (low->mid), Hunter (low->mid striker)
- D Tier: Seeker (low->low)
Mages, sha'ir, and witches can take arcanist (PHB) wizard powers, so they're basically just wizards with different class features. Witch and some of the mage options can be as effective as wizards, they’re ranked lower here due to being less versatile.
All strikers deal extra damage. Good strikers can bloody or even kill a solo in 1 round.
- S Tier: Ranger (mid->high)
- A Tier: Avenger (mid->high), Barbarian (mid->high), Monk (mid->mid controller/defender), Rogue (mid->mid), Warlock (low->high controller)
- B Tier: Executioner (mid->mid), Slayer (mid->mid), Sorcerer (mid->mid controller), Thief (mid->mid)
- C Tier: Hexblade (low->mid), Scout (low->mid)
- D Tier: Assassin (low->low), Blackguard (low->mid), Vampire (low->mid)
I'm not familiar enough with Elementalist to rate it.
Last But Not Least: Hybrids
In addition to dipping into a second class via multiclassing, you can combine two classes by making a hybrid character. Don't do that. A hybrid X|Y character is 40% X and 40% Y. You may notice that this only adds up to 80%; that's because in the vast majority of cases a hybrid character is at best only 80% as effective as a non-hybrid character. It is possible to make a very effective hybrid character, but it requires an extremely deep level of optimization knowledge. To use the optimization floor/ceiling terminology from earlier, hybrids have a very high ceiling, but their floor is way down in the sub-basement, much lower than even the D-tier standard classes. If you really want to try a hybrid build, I would recommend going and finding one of the really good ones that optimizers made back in the day, like Darth Vader (paladin|warlock) or Killswitch (warlord|artificer), rather than attempting to create your own.
An Addendum on Stealing from Other Classes
One of the best ways to shore up a weak class is to use multi-classing and other similar options to steal the best powers and paragon paths from other (stronger) classes. You can absolutely make a heavily-optimized tier D character that's stronger than a minimally-optimized tier A character. When both characters optimized, it starts to get back to the floors & ceilings discussed earlier in this answer. With the resources that a tier D character invests to compete with a standard tier A character, that tier A character can pull ahead of them again. The tier A character will run into diminishing returns, though; nothing a ranger can do will give them as much of a boost as a vampire gets from stealing a ranger's best powers. Optimization by poaching from other classes can narrow gaps between tiers, but bottom-tier classes will never quite catch up with top-tier classes.