Thank you! That's what I get for falling behind on the theoretical computer science stack exchange.
Does anyone know who wrote that answer?
Check https://cstheory.stackexchange.com/a/38832/5630 Alexander Razborov was the first to prove it wrong by providing a counterexample. He got a Nevanlinna Prize in 1990 for a result in this field, it's not a surprise at all he was the one.