I've come a remarkable distance from my views on game creation as a child. Snekmen showed me that, definitively, execution trumps a "great idea." I consider it to be my best game by a wide margin and it started with the least interesting and most vague idea I've ever gone with: "I want to make a simple stealth game but limit the screen to 5 tiles wide and 6 tiles long." This resulted in me having a completely playable version of the game in a little under two days, with four simple levels and only the first enemy type.
Immediately I began shoving this build in people's faces as much as humanly possible. Both friends and strangers, gamers and the dexterity challenged. I learned very quickly what wasn't obvious and what was too ridiculously hard since I only responded with stoicism should any questions come my way. Because it was so simple, through quick iteration I was able design the game in such a way that most people could play, understand and beat the game with no help from me, while the game never explicitly explained its innards.
I definitely failed in this aspect in many regards, the most glaring being that the triangles are the goal. Regardless, my mother was able to complete the game with no help from me asides explaining that the floating pieces of unearthly beauty are indeed what she should strive for.
In stark contrast with Snekmen, almost everything I've worked on prior started out with a strong idea. Mix snake with an RTS where you act as the mediator between two factions, or blow a box around a massive desert where all of the human artifacts are peculiarly large children's toys (I admit that last one is insane, but I still find it hilarious). These panned out horribly, as I was striving to achieve this very specific and flawed image some part of my strange and elusive mind concocted. With Snekmen, the plan was vague, so I was able to simply make a playable game, see what worked and what I liked, and iterate the hell out of it. This method let me achieve about 5000% more work in the same amount of time I spent on this horrific beast
So while I can't say I'll approach every game this way in the future, it's definitely a very powerful way to learn a tremendous amount while doing work very similar to chiseling away at a slab of marble, instead of shaping an already formed statue into another, less appealing shape.