Please note that I stated the experiment was terminated by the forking, so all of the "normal" network activity was happening during the experiment. In fact I state in the article that the forking terminated the experiment. So the numbers are still good.
Additionally, (@jmarndt would have to verify) when he finally staked the network weight was "normal" at 1.xxx on a non fork.
Network difficulty is always fluctuating. I've seen it close to 7.x and under normal conditions before the recent forks as low as 0.4x though that low was an extreme example of "normal".