Making spells stronger for each level would also be an acceptable option. I really just wanted 5e to be more modular, a basic frame that you can add new mechanics on top of in order to add more flexibility. There are too many things that just kind of got baked into the core game or presented as optional rules when they really arent all that optional.
The issue here is that it is a lot harder for most people to create the things you want to leave out on their own and the largest share of rpg groups will never use anything outside of the initial release.
Basically, if you want to reach a broad audience, you NEED a lot of those options in your core books, even if you’d rather not use them yourself. Your potential audience will have moved on if they have to wait a year for a supplement to allow them to create the character they want. 4E launched without even all the core races and classes from 3e and, even though they all got released within 9 months, 4E never heard the end of it for not having them from the word go or that you had to buy a supplement to get them (similarly, 5e lost a lot of 4E fans by dropping the Warlord class).
You want irony? I only play humans in every campaign I’ve ever participated in. I generally feel that probably 99% of other fantasy races are just charicatures of certain types of humans and I’d rather have the full range to work with.
Yet the system I’ve written includes probably one of the largest assortments of playable species options I’ve seen outside of Rifts. Why? Because my audience is more than just me and I wanted to be sure people could create a version of anything from all the prior editions of D&D in the system using just the core rules.
If the GM doesn’t want a world with playable sapient golems, it’s easy for any GM using my system to just say “no golem PCs.” It would be a lot harder for a GM, particularly a new one, to have to create a golem race from scratch, particularly if the only examples to compare it with are humans and the near human dwarves, elves and halflings. They’ll probably just decide it’s too much work and just drop the idea of having sapient golems in their world.
The same goes for classes; one of the primary reasons for my breaking the D&D class concepts into two parts (combat class and non-combat background) was because the resulting mix-and-match allowed a LOT more D&D class concepts to be expressed with a smaller page count.
The nature-themed options of the barbarian background can be applied to a daring berserker fighter w. the ravager path for a D&D barbarian, to a swift wary fighter with the striker path for a D&D ranger or to a Mystic for a D&D druid. The religious background that provides non-combat divine spells can make a paladin (strong fighter), a cleric (theurge) or even the non-combat “cloistered cleric” variant from earlier editions (with the mastermind class).
Between all the class and background combos you can get close to even the most obscure classes from past editions of D&D.
And again it’s easier to remove certain options; ex. No gadgeteer class; than to add them. A GM could absolutely restrict class/background options down to, say, “the only options allowed are military strong tactical fighters with the striker path (OSR Fighter), outlaw swift daring fighters with the brigand path (OSR thief), religious militant theurges with the benedictor path (OSR cleric) and arcanist lore wizards with the interdictor path (OSR magic-user).”
I’ve even got an optional rule for pre-set features in place of having player choices at every level because it’s easier to restrict than create your own.
You can say you prefer a modular system, but unless the system has modules ready to plug in, it’s just an empty toolbox. Likewise, billing something as a complete toolbox, but then requiring them to buy additional tools to actually make it functional is going to see customers turn to the guy offering an actual complete set even if some of your tools are better built.