They use "proxies for temperatures" and discard the proxies that contradict their hypothesies. In all sciences. Which is why there is a reproductibility crisis.

( most studies cannot be replicated without re-introducing the biased methodology of the original paper )

Cherry-picking.