Throughout whole history, successful armies, such as those mustered by Alexander the Great, the Roman Empire, the Mongolian Empire, the Huns, Oda Nobunaga, Tokugawa Ieyasu and others, have been composed of *professional fighters*, even the heroes of those armies were professional fighters, and most, and of course you will find some exceptions, did nothing at all except fight. Had no special skills whatsoever outside of battle, and this is why you have many historical accounts of former knights/warriors/samurai turning to common banditry & mercenary work: it's because they didn't have ANY useful skills outside of battle. There are even some very famous Ronin mentioned in historical japanese accounts.
Okay, it's an outdated comment, but I wanted to pick that one up nevertheless.
First, extremely few people have knowledge at "nothing beside x". In fact, the AD&D2e PHB made a point by listing the different skills an average player in his teens might have due to schooling and the skills an NPC of comparable age would have because of his upbringing.
The ronin/bandit/highwayman examples you brought up? Somewhat valid, but you have to keep in mind that perhaps (speaking in ingame terms) these guys weren't that high level, had an int of 10 and dumped all skill points in skills that are relevant for physical feats and little else (spot, listening, intimidate, tumble etc.), instead of diplomacy, crafting and whatever. Or they had better skills in woodmanship, but instead of turning to a live as a hunter, they thought that travellers are a more profitable game. Even with a far wider array of class skills and 4 points/level, you can still create a one trick-pony that doesn't function well outside of purely physical situations.
That said, I think that the extremely limited skill sections for most classes are an artificial limit on their usefulness and their versatility. Okay, I can understand that the rogue is the only class that's really apt at rogue skills for balance reasons, but beside that? Nah.
That's what I liked about Howard's Conan stories (apart from the manliness-factor) - even though the guy is a high level fighter (or barbarian) with some levels of roguelike classes thrown in, he nevertheless has a lot more skills up his sleeve than one could recreate by using the classes and therefore can be of use in pretty much every scenario from robbing a wizard's tower to scouting the Pictish Wilderness to leading an army into battle. And on top of that kick pretty much everyone's ass... and he still wouldn't be a lot better than t3-4ish.
(which points at another problem aka "why are casting classes that good?" - but then again, in the Conan stories, Wizards can't just come up with anything and everything but are actually very limited when it comes to application of instant powers and are, like Thoth-Amon or Xaltotun, dependant on artifacts when they want to harness high-level or epic level magic)
For example, i don't think the fighter, the ranger, the paladin and the barbarian are sufficiently different to warrant completely different classes. Why don't consolidate all of them into the fighter and even open up some more paths for him to follow? Like an intelligent, strategist kind of fighter, or a charismatic warlord that leads hordes to do his bidding. I think 3.5 left too much up for multiclassing and prestige classes, and i think a more open approach to the classes might help alleviate the problem, not only for the fighter, but also for some other redundancies that exist in 3.5.
Here I wholeheartedly agree.
I have to admit it: I liked the basic concept of the warrior class with its customizability. Adding more feats into the game to represent different ways of fighting could have covered anything variety. Why, for example, is a samurai class needed if one might as well say "okay, bastard sword, short sword, weapon spec, all dual wield skills, take up the kiai feat, awesome, you're done"?
And the same applies to baseline classes. Back in 2000(?) when I bought the PHB I wanted to recreate a Dark Sun campaign. However, just by looking at the classes I thought "cool, I don't need a gladiator because the fighter already covers everything needed." The same could be done with the barbarian (take up hp feats and introduce bonus movement and raging feats with some follow-ups, voilą, le barb). The ranger? Well, an archetypical archer/woodsman or dual wielder/woodsman, the druidic spells don't fit the concept that much anyway and could be represented by taking up some levels of druid or nature priest. The paladin is the only class where a separate class makes sense because of the code of conduct and the abilities being very specific (healing, immunities etc.). But even here one might as well have introduced a separate tree of feats to recreate a warrior paladin (btw, the same can be said about the bards in regards to the rogue). Instead of creating another (prestige-)class for each and every niche there is, they might as well have sticked to the concept of customizability via skills and feats, but of course this wouldn't have been as lucrative...