Thoughts and discussions about classes in RPGs

Last night I was in the shower, after having spent some of my day thinking about my RPG, Stuff of Legends, I had a thought. To intro with some background, I've decided to design Stuff of Legends without classes or character levels from the get go, letting skills and their levels describe the character.

So I had a thought, and it was following playing both Numenera and D&D 5th edition lately. In either game, there are classes present, subtly in the prior, overtly in the latter. The way the games function, and how the characters act out based on and despite of their class, got me thinking on why classes were retained as a feature, despite moving away from the wargaming roots.

Thougths
My thoughts about classes are mixed for all kinds of reasons, but now I find a warmer spot for them in my heart. The complexity of the 3rd and 4th editions of D&D had turned me off them, but D&D's 5th edition had given me a chance to rethink why exactly I had issues.
I've found out that I do not actually have issues with the concept of classes. The idea that a character had spent their formative years learning a specific trade or, in this case, class, is more than OK - it makes sense. In this regard, I find classes logical, since they provide a lens on the character and their life before being generated.
But I still have issues with classes, despite it. For starters, in some games, a wealth of information and in-game-world choices are present. If a game were to be set in modern times, it would make very little sense to me to rely on classes. If the game is fantastic, or medieval, a different issue rises: multiclassing. I honestly multiclassing has no room in fantasy gaming, unless it is handled like older D&D did it - once you pick up a new class, you can't return to an old one*.
So the train of thought continued and brought me back to my game: why do I not have classes in my game? The answer feels half-hearted: because it restricts the move from concept to on-page character. Restrictions apply due to setting, so you won't find Earthbenders from Avatar in Steampunk England, nor would you find a wuxia hero in a pre-historical hunter-gatherer style game.
But when you look beyond the restrictions of setting, classes seem to restrict in less-ideal ways. Let's use D&D 5th edition for some examples, since it's the most essentially flexible of the bunch. Let's take the Eldritch Knights path for the Fighter. This path talks about wizard-mixed fighters with strictly abjuration or evocation spells - defense and offense. Spells only up to 4th level and a limited amount of them. Assuming we define within the spell levels allowed, you cannot have an Eldritch Knights with Alter Self, See Invisibility and Dimension Door, for a type of spy-counterspy Knight. If you wanted this, you would have to houserule or multiclass. If the GM has chosen to not do the first and not allow the second, the player has a concept that they cannot realize and that chafes for me.
I think that, in the bottom line, I am torn - classes give a very clear outline of who the character sees themselves to be. As far as I know, the vast majority of people in the past several thousand years required a very good reason to change how they were going about their lives. In the context of fantasy RPGs, for a wizard to dip into fighter there would need to be a very major event in their lives, as would the other way around. But in the same time, classes are restrictive in their nature and even the most flexible systems leave things to be desired.

Discussions
The term discussions serves two purposes for me: the first is to open the floor for discussions, while the second is to point at a subject as something I am not sure about. Below I have a few such discussions for which I would like comments, replies and actual discussion to come up. My thoughts are laid out, in part, above and, while they may change, will direct what direction I want to take with the discussions.
Additionally, before the discussions, are some relevant mechanics from my game, as it is the primary subject of the discussions:

• Currently, characters receive some definition by spreading points between 3 pillars: warrior, specialist and arcanist, which describe the 3 primary types of characters: combat, non-combat and esoteric. Besides providing advantages, these pillars have no other mechanical use as of yet.
• The number of skills which describe character roles and capabilities stands at 17, though this number will change. These are grouped under attributes and are as general as I was able to get them, with skills like Combat and Knowledge.
• Nearly everything comes in 4 levels which map to beginner through master.

Discussions are kept in a numbered list for convenience of reference.

1. The issue with which way to take, classes or classless, stands strong for me now. So does multiclass - or picking up new skills out of the blue. I prefer to remain classless, but then a skill system needs boundaries as to what skills may be picked up and I'm not sure how to work this out.
2. If restricting multiclassing, the question of when the restriction lifts is a big one. As this is restriction, rather than banning, when does the restriction lift, how and where does the limit for lifting stands? If for skills, how to even model that?
3. The pillars currently serve a small but relatively significant role, providing an image of the character which could be filed under multiclassing from the get go. This fails, though, to help map out progression, which is where multiclassing fails for me. I'm not sure if to rely on the pillars for a solution at all.

That's pretty much all I have right now, though more might crop up. I'm not entirely sure what kind of comments and replies I'm looking for, but if I had to define: new views or examples from existing games - things to broaden my scope.

---
*I recall this from Baldur's Gate, so I suppose the right older D&D is AD&D.

D&D 5th Edition and Infrequent DMing

I've run my first game of D&D 5th edition just now. I woke up before the sun did and went in with little prep.
Luckily, the results were good.

So impressions from the DM side about the mechanics that were taken into account, given no combat occurred and barely more than 1 spell used:

1. The minimization of applicable bonuses and calculation thereof is wondrous. Because there are no branching and overly complex stats, calling a roll for anything has become easy and pleasant.
2. The game still relies a lot on combat, as far as characterizing the characters. There are RP spells and RP character options, but they are a bit too few.
3. I'm probably skipping a lot of rules by winging it, since the game felt barren of rolls, but it was fun and interesting. I suspect I might need to give the book a more thorough read.
4. There are not enough monsters to really run a campaign with right now.

However, I'm also brought to another point: I feel like I'm a fair enough, but below average, DM/GM. Partly, the fault lies in how infrequent I get to do so, but I hope to remedy that. In another part, I think I need to get used to getting usable prep done for games, since by how much I'm winging it, I sucking out potential awesome from the game.

While we're here, let me tell you how it went. Playing were +Alex Perucchini and +Other Tim. +Anthony Fournier was slated to play but didn't show up last minute.

The mission the party got was to find out why the heck these local, nice kobolds were acting up and raiding farmsteads for food and goods, taking livestock and anything else they can conceivably lay their hands on.
They headed out, Alex a Paladin and Tim a Warlock, to the den of the kobolds. The guards outside were forthcoming and called for those in charge once they saw adventurers and feared for their lives. Out came a pair of Dragon Priests, who agreed to discuss the recent events within the den.
The party was taken to the priests' chambers, where the aforementioned refused to disclose details, and eventually the priests led the party to their draconic patron, a massive green dragon. The priests did not actually come with the party, so they did not hear the dragon debunk their story and request that the adventurers resolve this. The party got back, entangled the priests and got out of them that they are paying off a hobgoblin racket. Back to the dragon and then instructed to get this all resolved and kill the hobgoblins.
The party got back to see a small riot happening around the priests in the main chamber of the den and through character abilities found out the location of where the priests were keeping the money they were collecting. Since it was hidden, the party surmised that there wasn't actually a hobgoblin threat. With the recovered money, they paid off the town and gave half the remainder to the kobolds to live off.
0 combats, lots of roleplaying and one very big dragon.

+James Young - in case you were wondering what had happened.