Kabouter Games
Explorer
Welcome to old school. 3d6 in order, pick a class, slap on some gear, and get adventuring! Survive as long as you can. Originally, this game was designed with setup for the players being very simple and quick. As long as the DM was prepared, you could get out the D&D game with a table full of fresh new players and be adventuring in 30 minutes or less. That ease of starting to play was huge plus in attracting new players. Spending hours noodling over what is essentially, tweaks for your little car, shoe, or top hat would be baffling to many early players.
That's what's behind my reaction to "optimizing," as I pointed out up-thread. The grognard editions, being truly class-based games instead of skills-and-powers games, made chargen very simple and fast.
That's why I have a significant problem with optimization in 5E: It's back to a class-based system. You can just roll your stats, choose a class, pick some gear, and start playing. This is great for new players, folks coming from Olde Schole versions, or those who just want to get rolling with a minimum of fuss.
You can also use the optional rules of feats and multi-classing. This is great for those who want to tweak things to get exactly the character they want, or who simply want more crunch in their game.
The problems come when you try to mix the two in one party.
In my experience, the character with feats and multiple classes will consistently be more powerful, levels being equal, than a "by-the-book," class-based character. The Shadow Monk Assassin I've built does more damage than the class-based Paladin even when he switches on his Smite. He does more than the class-based Rogue even though she gets more Sneak Attack dice. With the Alert feat he always goes first unless my dice are trying to kill me.
I am not a skilled optimizer, and yet I can do this. I can only identify two possible causes: The classes as written are bollixed up beyond belief, or optimization gives an advantage compared to by-the-book PCs. As by-the-book PCs work very well together, with no one class being heads and shoulders above the rest (I think the current crop of designers have done very well indeed at balancing the classes as written), I can only deduce the latter is the cause of the inequality.
For me, this points to a mixed table being thin ice. One must tread very carefully to avoid disaster. It can be done, as others have noted, but the likelihood of cracking through is high.