I've seen plenty of testimonies that to make a good game and not waste time, one should prototype early, quickly see if the core is fun as is, and not focus on graphics, menus, sound, secondary features, etc and I agree with it.
However, I'm creating a game whose very philosophy seems to be at odds with this. I want to make a real-time, 2D, 1v1 PVP game (or against AI), in which the player explores, sneaks, setup traps, tries to find synergies with their findings (items, spells, powerups, whatever), inevitably faces their opponent sometimes and must use all this to be smarter and kill them.
The first problem is that the whole point (and what would make it fun according to me), is that the world will be full of very diverse randomly generated various crap, often with both useful and hurtful features, that you must make work together while avoiding if possible the anti-synergies that will inevitably come up. I enjoy roguelites for that reason (and the replayability) and want the same feeling.
But then, don't I need to implement a vast amount of features (even very basically) to find out if that concept is in fact as fun to play as I think ?
The second problem is that a good part of the fun should come from outsmarting your opponent. So I quickly need to either setup a way to have 2 players playing at the same time (LAN ? through a server ? seems a complex and difficult topic and not very prototypy). Or create an AI to make up for it, but it has to be smart enough to make it interesting, which again, seems to go against the "implement quickly and test" philosophy.
The 2 players don't have to be identical in capabilities (especially AI), it could also make a fun challenge to find how to put together a way to beat the dumb, vastly stronger AI.
I'm open on the game features themselves, I don't have a goal set in stone, which is why I'm protoyping in the first place, but I'd like to keep as much of the philosophy above as possible.
Note that the secrecy/surprise factor makes it impossible to do split-screen.
How can I prototype such a game ?
(Using Unity2D for now if that changes anything, seemed like an easy way to get something started)