It's like having a bag 1 filled with equal quantities of capsicums, cherries, and apples, then selecting the top reddest 25% of bag 1 and moving them to bag 2.
You will have much less variability of fruit type in bag 2 because you used color selection to fill it (apples and capsicums come in three colors at will be selected 1/3rd the rate of cherries).
That's the whole point of sexual vs natural selection.
Yes, I completely agree with you! You will have much less variability.
However, that's not what the paper argues. The paper argues that you would have more variability.
I'd appreciate you taking a look at the original paper (specifically, the bottom of page 2 [edit: and Figure 1]) and understanding my original argument.
You have increased variability between generations by reducing the variability in the second generation.
I agree with you that the variability has been reduced in the second generation.
It seems that we both disagree with the original paper, then, as it states
"If sex A is relatively selective and will mate only with the top most desirable quarter of sex B, then all of the next generation will be offspring of the more variable subpopulation B1"