Safari on the desktop is very, very bad about doing this. The browser benchmark tests really should include this metric: the average time to ready after navigating back. Showing a cached image but registering a click on a different image just seems like such an obvious thing to avoid.