The only actual design point that's really been close to settled is that of subclass level divergence. Tony Vargas has been asserting 1st level, while I asserted 3rd level.
The 1st vs 3rd choice influences the nature of the class vs subclass relationship, in terms of power derived from each aspect, the sorts of abilities that are presented and available, and the approach to the conceptual design of the character.
I've been doing a lot of thinking about the implications of the types of splits, both thematically and functionally, as well as reviewing whether I was even describing things properly. I've looked at how the design would play out in each version, trying to see which one would let things work more smoothly in those defining aspects where it matters (since there are a lot of aspects that don't really change with the choice), and I find that I didn't really set things out properly for the decision that I made.
First, I feel I used inappropriate terms for describing the types when I last discussed it. I described the 1st level split as "specialization", while the 3rd level split was described as "uniqueness". The terms used were sort of off the cuff, and thus didn't truly map to what was being described. Both 1st and 3rd are types of specializations. Plus, I blended together ideas from classes that get subclasses at 1st and 2nd level, forgetting that they are separate approaches.
We have classes that choose their subclasses at 1st, 2nd, and 3rd level.
1st: Sorcerer, Warlock, Cleric — The character identity cannot exist independent of the subclass. The Warlock's patron, or the Sorcerer's origin, or the Cleric's domain must be defined in order for the character to work at all; there's no "specializing" involved. The class is just a container to hold the subclass; it just provides the underlying mechanics for the subclass to use. These subclasses are "types" of the class.
2nd: Wizard, Druid — The character's identity exists without the subclass (as a broad concept), primarily defined by the unrestricted spell selection options, but the class provides no mechanical support for further identity resolution. The subclass provides specialization directly related to the features that the class has available at 1st level. This is not about character concept or mechanics grouping. Each subclass is just choosing to be better at some aspect of the base class.
3rd: Fighter, Monk, Paladin, Rogue, etc — The character can exist entirely within the class, and not need the subclass. Instead, the subclass provides a way to choose a direction for the character to go once you have a better idea of the general character implementation, but is not dependent on specializing on anything the class provides. Rather, it introduces entirely new abilities to match the direction the character is going. Is the Rogue more flashy or manipulative or interested in stealing stuff? Does the Fighter approach combat from a more tactical mindset, or does he want to incorporate magic into his fighting? What oath does the Paladin swear, once he's proved himself? The subclass is a layer on top of the character's core elements.
So we have subclasses that are fundamental "types" of the base class (gained at 1st level); those that are "specializations" of the base class and what it can do (gained at 2nd level); and those that are "evolutions" of the base class, that branch of into entirely new directions (gained at 3rd level).
Given Tony Vargas's comments, and allowing that he got drawn into my improper terminology, I believe he is pushing the 2nd level split, where the base Warlord class is defined by the variety of gambits available, and the subclasses focus on being better at certain types of them.
Great Split. I'll try to adopt your terms for the rest of the conversation as you did a pretty good job at highlighting some important differences. It's something I tried to touch on earlier but that you have explained and examined much better.
I do want to add one additional piece of insight. The reason their are no casters that get subclasses at level 3 is more mechanical in nature. It's because their basic class spellcasting ability powers up so much at level 3 that there just wouldn't be much room left at that level to put anything else. So there is also a mechanical reason to place a subclass at a certain level and that mechanical power difference may be why you see level 3 subclasses as evolving the class (they grant much stronger abilities at level 3 than subclasses at level 2 generally grant).
Level 1 is perhaps a bit different as your whole class is defined from level 1 on and so their is no way for the base class to evolve. It's as you described, essentially a class.
Anyways, what I've been going through all this to ask is:
Why can't the Warlord adopt more of an evolution stance and still have a primary ability that is strongly defining the class itself but still has the subclass evolving him into a different direction? If so wouldn't the most reasonable place to put the subclass be at level 2 as opposed to level 3?
I developed 4 broad concepts that I felt would be related to the Warlord concept, developed such that each subclass could handle a few different actual character types. To a certain extent it feels like a 1st level split — an Icon is not a Commander is not a Strategist is not a Defender. Each have very different problem-solving methods, and, for example, it's difficult to fit the princess concept in as something that could grow out of the Warlord class as a whole. Basically, the princess version of the Icon subclass is very hard to conceive of as not being a 1st level "type" subclass, whereas the shonen hero is easy to see as an "evolution" subclass. But then if you go to the Commander subclass, it's very easy to view it as a specialization type 2nd level subclass.
This is why I think Tony and others are focusing on the Commander-style subclass, with all the subclasses being specializations. It's much easier to take one thing that allows for some specialization, and consider that as something that will provide enough subclasses to be viable, than to look at different evolutions that approach the problems a Warlord deals with in radically different ways.
There is a reason martial classes generally evolve instead of specialize. It's because they have much less flexible class mechanics so they are incapable of generating new concepts without evolving. Specialization can only happen because there is a flexible mechanic attached to the primary class that allows for multiple concepts to already be somewhat mechanically covered.
You mentioned the paladin earlier as a class that evolves instead of specializes. I would actually view him more as a specialist. He's still a paladin and can fulfill a variety of paladin flavor with his spells. Even without a subclass he could play in such a way and pick spells and abilities that would mechanically fulfill the demands of just about any of the subclasses listed. Instead he just gets explicitly better at doing the things the subclass is supposed to do.
Why do we think a strong and flexible primary class mechanic is needed? Because there are soo many different variations and expectations around what a warlord should be capable of doing and not capable of doing. Because the different styles of Warlord's often blend together moreso than are totally distinct. A tactical warlord may sometimes inspire his allys. A inspiring warlord may sometimes bring together some good tactical plans. There's just not a clear line IMO between where one warlord ends and another begins and it's almost impossible to have total evolution and have tactical plans present in the core warlord. If it's total evolution then the tactical warlord is the only one that gets tactical stuff.
However the specialization approach is also extremely limited, and bland, if the core class does not evoke a wide variety of concepts on its own (as the Wizard clearly does, and the Druid does to a lesser extent). Providing evolutions allows for very different character types, which makes it much more useful for long-term design. Using the "types" subclass method, on the other hand, allows you to provide for a variety of narrowly-scoped ideas using the same mechanical underpinnings. Their similarity and differences are due to circumstances (accident of birth, choice of god, who they managed to find to give them power, etc), rather than fundamental to the class itself.
It's odd that while you find specialization based subclasses bland and uninteresting I find them empowering and perfectly suitable. I often find the evolving ones redundant and while the are more mechanically interesting they are more conceptually limiting. For example, I can't play a fighter that uses same magic and some tactical maneuvers. Those two things can never come together now under the same fighter character because we have them siloed off in subclasses.
I understand why they needed to do it that way. But I can't say I find that to be good design, at least not when they aren't making "hybrid" subclasses to bring life to the conceptual spaces between the current subclasses.
At least with the broad class and specialization route, all those options are automatically open to me and I just choose to be a little better at some area that my class already covers.
One interesting tidbit. The assassain subclass of rogue is a good example of a specialization subclass. The arcane trickster is a good example of an evolving one. Both types of subclasses are present in the rogue class. I think this doesn't have to be an either/or approach. Some subclasses can add evolution while some add specialization.
So where does the Warlord fall? Or rather, where should it fall? I don't know. I can give justifications for any of the three types. The Warlord can be a bucket of mechanics for a variety of different ideas people have and want to implement (ie: tactical vs princess vs lazylord) that differ based on circumstances rather than intrinsics. The Warlord can go all-in on the gambits, and just provide paths to be better at certain types over others. Or it can provide a strong underlying class that can evolve in radically different ways.
I think the biggest factor is whether the primarily evolution method is going to be able to generate good warlords that are more hybrid focused than primarily pushing a single path. Ultimately the desire is for a warlord that doesn't get to do anything tactical just because he didn't pick the tactical subclass and for a warlord that doesn't get to do anything inspiring just because he didn't pick the inspiring subclass. There is going to be a major design challenge in the evolution route to make sure that doesn't happen. Most of us want a warlord that can do some inspiring, some tactics etc.
My personal opinion is that the specialization route is the worst option. I reviewed a ton of the 4E exploits that the Warlord had, when putting together my own design, and at least 80% of them are worthless when translating to 5E. 5E just fundamentally doesn't work the same as 4E, and you can't pile on tons of micro-abilities and pretend that provides a useful choice mechanic on par with the spell system. And without sufficient choices to draw from, the specialization mechanic just doesn't have enough to work with to be viable, long-term.
1. Most 5e warlord abilities should scale instead of staying micro abilities.
2. I don't think it's been demonstrated that there will be a lack of sufficient choices for specialization to work long term.
3. My personal opinion is that it's the best option. I believe there is a reason no sufficient warlord has been created yet and that reason is because everyone has tried to do it so far under the evolution route. It always ends up leaving warlord concepts that warlord fans want to explore on the outside looking in. You almost certainly lose the lazylord/princess. You almost certainly lose hybrid tactical / inspring warlords. Heck there's even the question of how much baked in combat prowess you place in the primary class vs the subclasses and that's another hybrid tradeoff that makes warlord fans cringe.
On the other hand, I don't know whether "type" or "evolution" is better for handling the general concepts that are being applied to the Warlord. Perhaps some of the concepts just fundamentally don't belong in Warlord, as they only existed in 4E due to the mechanics matching up. A princess isn't a Warlord, and shoehorning it into the class just shows a poor understanding of the design process. It's not a conceptual match, it's a mechanical match, in designing a character that can still be functional despite the concept, rather than because of it.
Can a warlord be bad at fighting himself. In general I think it's possible. I used to think the princess concept would be impossible to translate to 5e. I now see multiple paths as long as you have the single broad flexible warlord mechanic and something non-attacking they can substitute their attack for. I don't think its the traditional warlord concept 4e was going for, but I also don't think its just a mechanical phenomenon. A lazylord/princess in conception would probably be more like a scholar that has studied war and battle and tactics. He might be sick or frail or just untrained or unexperienced with weaponry and thus might lack some of their insights. He's probably more tactical focused than inspiring. But that shouldn't rule out him being able to inspire the party with a story about some historical battle or person and then reminding them of it on combat etc. But that would be my conception of a princess/lazylord. Is that a concept that really doesn't belong in a warlord class? If it doesn't belong there then where does it belong?
Of my general concepts, I'm seeing:
Icon/Shonen hero: evolution (leading people)
Icon/Princess: type
Icon/Vanguard: evolution (leading the charge)
Commander: evolution (commanding people), specialization
Strategist: evolution (manipulating people)
Defender: evolution (using terrain to advantage)
Defender/Ambusher: evolution (using terrain to advantage)
On the other hand, I could see the Princess evolving into the Shonen hero. In that case, the Princess is just something that needs to work from the baseline of the class, and have an evolution path available to her.
So, after a more careful look at things, I think evolution works best for my view of the Warlord. It is still using the 3rd level subclass split, but now I can see where I was making mistakes before, and have a better path to work with.
I'm not sure what a shonen is.