Hacker News Reader

i'm not sure if i understand the problem. couldn't you just take the inverted result of every other flip?

No. With a large bias you'd have hundreds of one result for one of the other.