This seems great to me. The awfulness as a top deck is a real thing but it's great until hands are empty. On turns 5-7 is is going to be good against most opponents. If it could play lands I'd definitely consider it because then it's just great from t3 until hellbent. It's worse than Mind Twist by a substantial margin but Twist is a top 5 black card imo.
Gold sections are always hard to break into and I don't know if I'll be able to make room for it.
Definitely think this is better than Rakdos's Return. In select scenarios this can even be better than Fractured Identity in that your opponent will never reap any benefits from their own card, but you will. The problem with Extract Brain is there is a limited window for this card and your opponent has to have an ideal target in hand during that window. I think of this as Bribery with the pro of not being limited to creatures, but with the much larger cons of stealing from the hand as opposed to a whole library and being gold. This is narrowly outside for me, but I will keep my eyes out and look for reports from those who give it a shot.
The top 2-3 cards for any gold sections are pretty much 100% set and we can all agree on. When it comes to the 4th to 6th gold card (depending on the size of your cube) its always a coin flip and I find it depends heavily on your archetypes, curves and even play style.
The second thing I learned about cube is cards that have decent floor but good to high ceilings make great cube cards - they're consistent in a format without that much consistency and offer great stories/ build arounds.
The problem I find with steal your opponent's cards is you're deck isn't built around them and often you cannot use them as efficiently as your opponent. The reason why cards like Phantasmal Image / Bribery etc. are good is they are insane tempo swings.
This feels too much like another Feast or Famine type of effect - which is something I've started to avoid - i.e. Opposition Agent
Private Mod Note
Rollback Post to RevisionRollBack
I'm actively maintaining a comprehensive article to help explain to new cube players how some complex vintage level cards work in a cube environment.
You probably need to do this for x = 3+ to consistently get something decent. Opponent showing lands, being hellbent or dead removal spells are all brick scenarios.
With rakdos return, at least a quick x = 3 will always be good. With extract brain, an opponent has a full grip, x = 3 can STILL brick.
You'll probabily need X = The number of cards in their hand (-1) to get reliable upside.
Maybe if you like big swingy plays that lead to epic stories it'll be worth it to test out , but I'm not a fan of this card for fun AND power level reasons.