Ah, that's quite the list. Let's get started.

Regarding unit number, I think 5-6 is just right. Anything more than that, and 1) battles take too long and 2) complexity starts to overbalance enjoyment. That's my assessment, but I also think it's supported by most SRPGs. I don't think anything should be forcing you to train additional units which you don't want; I think how you play the game should be up to you, and if you want to do it with three units, so be it. However, it might be useful to have more, in case certain battles or quests or something require a specific class. In general, I think the fun of a strategy game is (nearly) all in figuring out solutions to a given problem (battle), or in testing multiple solutions, so I only really see a need, as a designer, to present a question, and let the player solve it as he sees fit. Right and wrong are determined by winning or losing. In this particular game, where the unit development is so complex/customizable, I don't see any benefit to allowing units to die permanently; I think it's enough to have your strategy effectively wrecked by a unit's loss. I also don't think it's really practical, and the last thing I want to do is to force someone who is going to win the battle to have to start over from the beginning because one of his guys died at the last second. That being said, I wouldn't be opposed to some kind of penalty for allowing a unit to die, at least for allowing him to stay dead for too long/at the end of the battle.

I favor a larger map size in general, because I've placed an emphasis on movement/location in several classes and therefore I want the player to have enough space to properly set up. That said, I'd obviously allow for variation in size between maps, because it changes the fundamental question of how you're going to get through, even with an identical team. I also see map size/features as being important to counterpicking in multiplayer settings. But to my mind, an average, "neutral" sort of map would still be slightly larger than the average FFT map, and would probably include some kind of special terrain formations like water. Part of being a good player is in being able to build a versatile team which can adapt to any map conditions. I would like for a standard battle to last no longer than 20-30 minutes, at least between skilled players, because any longer than that is pushing the limits of patience over consecutive games. If you look at professional Starcraft matches, they tend to end within 15-20 minutes, because you don't really have to completely decimate your opponent to beat him--you just have to checkmate him.

Regarding equipment, that's something I haven't thought about overmuch, but I think I discussed it a bit in my Considerations, along with status effects and elements. Those are all things which I think need real reworking in the SRPG genre, because neither currently plays a particularly interesting role in who wins a battle. For the case of poison, specifically, I want it to generally weaken a unit, in addition to the traditional, gradual decay of HP. If you're sick, you shouldn't really be able to perform at your best. I've thought about equipment breaking, because I haven't really decided how I would design a standard Knight class, and what I've determined is that I like the idea, but I don't think it should be permanent. I think the single-player should include shops which can repair broken items, and maybe a Mechanic class should be able to repair stuff during battles. The idea there is that sometimes a unit will be dependent on his equipment, as in the case of a Mystic Knight, or the strategy will depend on a certain character's improved moving range; things of that sort. And in those cases, it's often easier to break a single piece of equipment than it is to outright kill the unit.

I have no problem with unique story characters, and I'd probably even give them unique abilities and special stats. I think that sort of thing is fun for a player, adds to the collectible aspect. I'm slightly wary of AI-controlled ally NPCs, but even then, I don't have a particular problem with them. The only thing I might do is to always allow the player to decide whether he wants to include those story characters. I can see some situations where that would be less preferable, but it'll work as a general guideline, anyway.

Not all summons are AI-controlled; it depends on the class. But when they are, I think (hope) it provides an interesting dynamic for the player. After all, he still has the summoner under his direct control, and there are conceivable ways to influence even AI-controlled actions. I have checks set, to some degree, in things like the summoned units having to stay within a certain range of their "parents," limited functionality when separated from the parent, etc. In a strategy game, some luck is fine, can even be fun, but I want to try to prevent it from interfering--ultimately, though, I think that's part of the risk of choosing that class.

Now to focus more on multiplayer considerations.

Suppose you build that physical fighter who is *only* good at physical fighting. Then you run into a mage who can keep you at a distance, or a defensive physical unit who specializes in dealing with close combat. That physical fighter will be good in some situations, and very bad in others. And when you're dealing with creating a competitive team, you have to try to be prepared for as much as possible, which means each unit must be as widely valuable as possible--which means you may want to reconsider that Knight [but you might not, if he can, say, teleport into the enemy lines with the help of a Phantom and deal some quick, heavy damage before being killed/disabled himself].

Sometimes defensive builds will work, and sometimes they won't, or maybe you're just the kind of player who prefers to play offensively. That consideration is in every competitive game, from Smash to Street Fighter to Starcraft. Generally, people tend to consider it more "fun" to play/fight against more offensive opponents, so most games try to penalize, to some degree, superfluous defensive tactics (turtling, camping, stalling, etc.). But really, I think that, in a strategy game, if people start using purely defensive builds, there will be others who use mostly defensive builds which also can occasionally do small harm to the opponent--and if you can put the pressure on a defensive player, you can force him out of his element. You see it in chip damage, in Smash's diminishing shield sizes (and in how shields can be overcome by grabs), in unblockable attacks, in the loss of potential expansion territory and therefore resources, sometimes even in active penalties as with Guilty Gear. It would be tougher to apply that sort of thing to a strategy game, but not inconceivable. Ultimately, though, it might be useful to be able to impose a time limit on matches. Certainly, one of the greatest complaints against Smash Brawl is that it makes approaching/offensive play impractical and unsafe because defense is so easy and has no real penalties, and as a result, many, many more matches are decided now by time limits than ever were in Melee.

Conversely, sometimes winning is all that matters, and you may be able to do that well with a defensive build. Certainly in Smash Melee, Fox has some of the best offensive capabilities in the game, but he may work *even better* as a largely defensive character, depending on your style. Campy Foxes are usually hated, but if they were easy to beat, no one would really care enough to hate them.

I think, in any game, it's of primary importance to know what your opponent can do, and for that reason, I see no reason to hide enemy stats. Everything, right down to his skill tree, should be visible. That's how you strategize, by knowing the game pieces and using them effectively. I can't think of any good reasons to limit that.

I shy away from giving any kind of specific numbers, even just to illustrate an idea. I haven't really thought about it in that level of detail yet. But I will say that I lean toward giving units more HP than in most games, so they stay around a little longer. The thing to really keep in mind for me isn't the dealing of damage, but rather forcing your opponent into a situation wherein he will have to take damage, or at least be put on unfavorable footing. As I mentioned before, you don't have to decimate your opponent to beat him; you just have to make him unable to win. Sometimes that might be as simple as taking control of the high ground. Sometimes you might have a formation which works just right. So I think less in terms of what flavor of damage each class will do, and more in terms of how I can use a given class to harm my opponent in some way. Damage is only really useful for killing, and sometimes, killing may be unnecessary.

And that about does it for this round. The main thing I want to say is that, even without a skill tree system, there is still HUGE room for improvement in the current SRPG formula, and for me, that really begins in class design, in allowing your players to do more than just deal different flavors/colors of the same damage. Ninjas aren't defined for me by having two swords; I think of them more as one of the most mobile classes in the game. And that's the sort of thing which no game has really done. Even if I could just take FFT and make changes, great things could be done just by reconsidering the classes it already has. That's why I talk less about specific mechanics in my series of notes, and focus mostly on class design. It's really the heart of strategy.