This is actually a really interesting concept, and maybe its just becasue I'm tired from hours of EE prototyping, but I think we have on our hands a really, really interesting way to generate cube lists.
You can rapidly create a 360 enviornment by choosing 6 cards: a W, U, B, R, G and a colorless card, then multiply by 53. Add the classic 2xFetch, 2xShocks + 2Wastelands and BAM you're at 360.
How good is this envioronment? Probably shit, because it lacks depth and counterplay. You're basically gaurenteed to see the same decks over and over because there are only so many ways to pair these few cards together, and everyone sees the same sets of cards. So we need to introduce counterpicks.
Each of your 6 original cards probably has some sort of counterpick to it. A card that it simple loses to. As an example: Liliana.dec is hard-countered by Loxodon Smiter. SavannaLions.dec can't beat Izzet Staticaster. So you find 6 "counterpicks" and now your Cube consists awkwardly of 27x of "original" cards, 26x of "counters", and lands. Depeding on how strong your counters are, they may push the original picks out of the meta entirely. Ideally, they're strong against their inteneded targets but weak to what their targets are strong against. If you couldn't tell, we're building a rock-paper-sisscors environment.
What would be interesting is, say we take CubeTutor's "average" 720 Cube (minus lands?), and we index every card into a network that maps that card's relationship to every other card in the cube. We tell a the program to select 6 of them at random. Then we tell it to find the six best counters (or maybe the 6 most "even matched" cards) to those picks, then to counterpick 12 cards based on the 12 cards, etc. until we get to having 128 enough unique cards ((360 - 40 lands) /2). This results in a Cube that's got some depth of play to it, and is probably running 2x or 3x or even 4x ofs, depending on how many unique cards we want.
After generating a playable rough draft, we as the human designers can tweak the list according to our preferences to make it draft smoothly and correct obvious flaws.
I don't know if making such a card network would be feasable, given how unique Magic cards are, but if that could be done, we'd have something really cool on our hands.