Something's has been bugging me for a little bit since my last couple of posts...
I think we're (me included) sometimes making an incorrect argument when people describe dry spells of one form or another when they did not see all the recipe rotations over the total time period they are describing, or at least sometimes including or implying an incorrect argument in a list of several.
For example, suppose Pet Food recipes should have a 15% chance of showing up per rotation, and I were to describe going for four weeks with only 8 Pet Foods total, but mention that I miss one rotation a day because I sleep. I don't think it's a correct rebuttal to say that you should have been picking up those overnight rotations, that by missing 28 out of 112 potential recipes, 25%, your observation might not be as unusual as you think. Implying that the balancing, missing occurrences that would better produce your expectations might have been in the missed rotations, that you could have had better outcomes if you had looked at all the rotations..
But, all the rotations should be independent. There is no expectation that the fourth rotation a day would be better than the other three. Your observations weren't 'incomplete' because you didn't see every rotation. Your expectations should be the same on a percentage basis in either case.
Mathematically, you just need to make sure that you are basing your expectations on the correct number of observations. In my above example, it would be incorrect to assume that the four week observation should have seen about 28*15% or about 4 per week, or 16-17 in 4 weeks, and thus 8 total was abysmal luck, because that would be incorrectly computing expectations out of 112 including 28 opportunities not taken. It would be correct to assume that the imperfect 4 week observation was the equivalent of an 84 rotation 3 week complete observation, with the expectation of 12-13 Pet Foods, and thus 8 was bad but not quite as abysmal. 84 observations is 84 observations; it shouldn't matter which 84 they were, if they were consecutive or dispersed. It's important to have a correct count of the observations to have a correct computations of the expectations, not to spot the holes where unobserved successes might have slipped through.
Having said that, it is still a good suggestion to try and not miss any rotations, whether by sleeping, or by not picking up a finished recipe before the next rotation. Not because you might be missing the good ones, or because you shouldn't leave that area of uncertainty, but simply because it increases your sample size. 28 chances a week is better than 21 chances a week. You can't win if you don't play. The more times you reach into the duck pond, the more chances you have to grab the Pet Food duckies.