I don't like their "Better too hot than too cool" conclusion. What about error rate, I thought that was affected by temperature. And, from their data and graphs, I'd say that a medium or normal temperature would be the best to keep it at, around 35 +- 5 C.
|