Several weeks ago, a Conscientious Reader commented that I have a long history of giving a lot of design consideration to spellcasters of every type, at the apparent expense of fighter and rogue characters. My Swordplay series of posts aside, I'm comfortable calling this an accurate observation of a blind spot in my general tendencies. I'd like to talk a bit about why this is, while also trying to mend my ways.
The fighter class prior to 4e filled a design space that the 5e designers have been talking about a lot; ENWorld has collected tidbits of commentary on this topic here. (Do go ahead and read that, as much of my commentary is in that context.) In 2e and prior editions, the fighter was by far and away the simplest class in the game, mechanically speaking. Creating a fighter PC takes fewer decisions than any other class, and the player makes very few decisions during combat beyond which enemy to attack this round and (if using a grid and minis) where to stand. If there's a usable stunt system, maybe the fighter comes up with a stunt. Over the course of several levels, fighters gain toughness, better chance to hit, maybe weapon specialization (you'd have to be nuts not to buy this in 2e - it's a primary class benefit) and extra attacks per round. If there's a skill system in place, that might improve, but fighters are really always going to be bad at noncombat things, simply because the class was explicitly designed to have more fun in combat than anyone else. Admittedly, 2e's kit system added considerable complexity - but not all that many of the Fighter kits had broad appeal for my players in those days. Oh, and they published a whole book intended to add more variety to combat for fighters.
There's nothing explicitly wrong with a low-complexity class, even in a game where other classes are more complex. D&D as a whole is complex and forbidding enough that a class that is easy for new players but still rewarding for experienced players is excellent. But then, the whole motivation behind this post is addressing the latter clause, so let's come back to that.
3.x fighters saw an increase in complexity, matching and in fact caused by the increase in complexity from 2e to 3e. Unless the player has a spare very high ability score to throw into Intelligence, single-class fighters aren't going to see all that much of the skill system, but no class gets anywhere near as many feats as single-class fighters. I regard the tiny number of skill points and narrow range of class skills as a significant problem with keeping the players of fighter PCs engaged once the combat stops. I realize that this commentary isn't exactly a revelation to anyone. Still, I feel that while "dumb thug" is one species of fighter, it shouldn't be encouraged, much less the default, because it's hard to keep dumb thugs engaged in what's going on in the game as a whole.
I have extensive anecdotes (almost extensive enough that I would think of them as "data") from LARPing and tabletop gaming alike that point to fighter archetypes being especially vulnerable to getting shut out of engaging intellectually with encounters because they either don't have applicable skills or don't have applicable knowledge to cue them in to how they could apply their skills. It doesn't have to be this way; particularly in LARPs, a "dumb thug" played over enough years does eventually get hooked into enough other plot threads (with some nudges from Plot, perhaps) that they can engage in more conversations. To put that another way, dumb thugs tend to get boring after a year or two, so the characters wise up.
The obvious part of why this happens is that the players start out by saying, "I don't want to have to think too much." I understand this perspective, and I've been there myself. None of my commentary here is intended as criticism of players who start with that baseline for their characters; tabula rasa and the Everyman are perfectly good ways to play. They don't hold up well for long-term play, though - all plotlines of even halfway decent writing yield dividends directly based on how much a player invests in them. (Any cases where this is not true are rare enough to be irrelevant over a long enough timeline.)
I'll continue in my digression about LARPs here, because I think LARPs have a lot to offer tabletop gaming when it comes to solutions. I'll start by talking out of my ass: In NERO, there's a societal system in place to provide a ladder for fighters (and everyone else) to climb - with its American ideals of meritocracy (and, I suspect, influence from the SCA), characters can start out as commoners, become squires, receive knighthood, and ascend the echelons of nobility based in no small part on their prowess in combat. In Shattered Isles, there were very few pure fighters - really not more than three or four long-term pure fighters - and the vast majority had at least a bond to one of the elements to provide another place to dump some XP. That bond then got them caught up to some degree in interacting with elementals, which proved to be an essential on-ramp to engaging with other areas of the setting by providing Personal Mechanical Plot and a paradigm for grasping World Mechanical Plot. (For an explanation of these idiosyncratic terms, see here.)
King's Gate introduced martial schools as another build path for fighters, specializing them into a particular style. Eclipse and Dust to Dust have continued this trend, renaming them combat disciplines and warrior orders, respectively. These make it drastically easier to come up with ways to get even dumb-thug characters engaged in the world: there are NPCs they need to talk to in order to advance. Those NPCs belong to a broader society of like-minded individuals, any of whom can have goals, needs, or problems. So again, an on-ramp to Personal Mechanical Plot, as well as some amount of Character Plot. If the committee planned for it, martial schools may also tie into the Main Plot. If it's done with any grace at all, this is very rewarding to the player(s) of those schools because it makes them feel like they matter.
Bringing this all back around to tabletop games, part of the problem is that there are no existing mechanics in any edition of D&D to make fighters reliant on anyone else to improve, aside from some highly optional training rules. (Heads up: this is all coming down to an argument for more extensive training rules, at least as a modular rules option.) In 2e and earlier editions, there is really nothing for the fighter to learn from another character beyond weapon specialization, so there's no mechanically-motivated choice between learning what one instructor has to teach over another. Obviously, the DM can cover this with making one instructor's motives and requirements more palatable than the other options, and this is fine if it's enough to pique the player's interest.
In 3.x games, one could certainly improvise into existence a set of rules around each individual feat in a feat tree needing teaching, since these do represent the choice of one fighting style over another. (In the interest of cutting some lengthy discussion, assume I comment extensively on the role of prestige classes here.) The thing to avoid - and this applies to absolutely everything that requires a PC to have an encounter and possibly some downtime with an NPC - is using this as a roadblock for the character's statistical growth. In a tabletop game, in order to have the encounter that allows the character to advance, the player needs the consent of the DM (to stage the encounter) and often the tacit consent of the rest of the party (to get the group to go to the encounter's location). LARPs have it easier here, in that characters don't expect to stay in the physical proximity of their teammates all the time as they often do in tabletop games, so the rest of the team can shoot the breeze in the tavern while the training character has a one-on-one encounter with the teacher. (Given the investment of energy from a staff member, though, group teaching time is Highly Advisable in LARPs.)
3.x fighters have one other really interesting application in the rules. Fighter is one of the strongest classes in the game for multiclass dabbling. The fighter class improves physical combat abilities and gives a very handy infusion of feats, without tying a character down to a code of beliefs, restricting weapon or armor usage, or anything along those lines. With the fighter class, a player can improve her character's fighting ability without seriously diluting or altering the theme of her other class. Given the very tight economy of feats in 3.x, the fighter's theme-neutrality is hugely useful, and it bothers me in Arcana Evolved that there aren't any theme-neutral classes for class-dipping when you just desperately need more feats or skill points.
I've gotten away from my point here a bit, but I think D&D settings could really benefit from warrior orders, Mithraic cults, and secretive kung fu masters that teach or improve the character's fighting ability, which also means leaving mechanical room for that to mean something. Of course, if every character in the party needs to resolve multiple encounters or extended quests to gain the full benefits of a level, you're going to be stuck in an endless loop of training-montage scenes. If that's what you want, go for it - but most groups would like to spend a little time either screwing around in the local dungeon complex or pursuing the main plot (these may be the same thing), so I'd recommend requiring intensive training only a few times in each character's career, with additional information-dumping encounters or connections between the instructor and other plotlines along the way.
This post is already very long, but I haven't gotten into 4e fighters, which are remarkably different from their predecessors. I would go so far as to say that the fighter class underwent more change from 3.x to 4e than any other class. 3.x and prior classes are founded on a basic attack that the character can always perform, with some variations and options introduced to that attack either through a stunt system or feats, such as attacks to disarm, trip, bull rush, or grapple. They defend party members mostly through physically standing in the way, and through the DM arbitrarily choosing to have monsters target high-defense fighters over more vulnerable characters.
In non-Essentials 4e, fighters are all defenders with the ability to apply the Marked condition, which has varying effects but always in the general theme of punishing the enemy for attacking characters other than the defender. These fighters also have at-will, encounter, and daily attacks just like other classes, but there is no in-world explanation other than "fatigue" or "tactical openings" to explain why the per-encounter and per-day limits exist. If this lacuna bothers you, 4e isn't for you; if it doesn't, you're set. The great thing about per-encounter and per-day limits is that it means the class's design involves a lot more in the way of opportunity costs, and that's pretty useful design space. (Oh, and 4e doesn't have the same dynamic of class-dipping that 3.x does, so that's not really a thing.) I've mentioned, quite some time back, that I'd be interested in a 4e game that did focus on fighters needing training from NPC kung fu masters to advance.
I'd love to see warrior orders, honorable rivalry, and training in secret arts be an engaging part of the game for those "dumb thugs," because those are great plotlines for visceral stakes that even new and nervous players can readily grasp. Giving them mechanical underpinnings, though, often means either increasing the overall power of the fighter class, or forcing them to work for abilities they "expect" to receive by default. It's a tough row to hoe - and that's why I don't write about fighters all that often.
For 5e, the following has been revealed: "Weapon specialization benefits - some are like at-will attacks." I hope that this means more or less what it seems to say, so that there's more design room for diverging paths of specialization in a weapon; I'd be fine with one type of specialization ability that comes from the School of Hard Knocks, while other schools teach different (but approximately mechanically equal) specialization benefits.