I agree that it puts the devs in a tight spot, and the real solution is a publisher that doesn't insist on gameplay being shown years out. But given that's not always the reality of the situation, of the options you laid out, my choice would absolutely be " intentionally scale back on your aims, to then reveal a better looking game and ensure you don't overpromise". From business, to relationships, to almost any other situation in life, most people would agree that "under promise, and over deliver" is a good quality to have. Even if the initial reveal doesn't blow people's minds, and leaves them thinking "yeah, that might be alright", you have lots of time, and lots of public appearances left to improve and then have people leave thinking "man, this is looking better and better!"
Decisions like this are made solely to make a big splash, and get people hyped up, with no regard for how the next two years are gonna go. It's blowing your load early, and leaving it up to the devs to clean up the mess and pray they can live up to those expectations.