Ya... I'm not dumb, I understand your usage of the terms, but academics and statistical modeling take a backseat occasionally during a discussion like this. Problem is no one is plotting the data, much less collecting it to quantify later for analysis; but I bet if you throw avg daily round times on a Pareto chart that this week and last show a major breakpoint in round length compared to a month ago. I feel like the reported "luck" is way off from what it actually is. We would get one 12 hour round per week before, now we almost get one per day.
While the pool speed at slush has been increasing, it has been decreasing as a percentage of the overall network steadily for quite a few months. Compared to July, it is half the % of the network it used to be. If a pool shrinks in its share of the network, the average round length will go up. Looking at the *time* a round takes is a meaningless number compared to the Shares rounds take in comparison to difficulty.
Additionally, smaller share of the network means you will have more variance. While your expected reward is the same, it will be made up of higher highs and lower lows when looked at over reasonable time frames (such as 30-days).