On p.2, Example 1.1 (in this paper), there is a description of a coin tossing protocol with bias 1/4. In the paragraph below the example, they note that for a protocol with $r$ rounds (assume for the sake of clarity it is $poly(n)$) there's a bias of $\frac{1}{r}=\frac{1}{poly(n)}$.
I am quite new to Cryptography, and since the paper they cite in this context is quite old and very different to their example, I'm left with two questions:
How can their example (Example 1.1) be adapted to a $poly(n)$ round coin-toss protocol with bias at most $\frac{1}{poly(n)}$?
How is the final outcome in a multi-round coin toss determined? (i.e., we tossed more than one coin each, so what is the final result?)
