I was not talking about this in the context of building an optimizer. I was talking about this as a simple way to as humans gauge whether we had made ethical progress or not. I still think your specific concern about my specific ideal was not warranted:

Since killing everyone qualifies as "death" I don't see how it could possibly qualify as in-bounds as a method for reaching this particular ideal. Phrased differently, for instance as "Suffering and death are bad, let's eliminate them." the ideal could certainly lead to that. But I phrased it as "Causing less suffering and death is good."

I used the wording "cause less" which means the people enacting the ideal would not be able to kill people in order to prevent people from dying. You could argue that if they kill someone that might have had four children, that four deaths were saved - however, I'd argue that the four future deaths were not originally caused by the particular idealist in question, so killing the potential parent of those potential four children would not be a way for that particular person to cause less deaths. They would instead be increasing the number of deaths that they personally caused by one, while reducing the number of deaths that they personally caused by absolutely nothing.

It does not use the word "eliminate" which is important because "eliminate" and "lessen" would result in two totally different strategies. Total elimination practically requires the death of all, as the only way for it to be perfect is for there to be nobody to experience suffering or death. "Lessen" gives you more leeway, by allowing the sort of "as good as possible" type implementation that leaves living things surviving in order to experience the lessened suffering and death.

Can you think of a way for the idealist to kill everyone in order to personally cause less death, without personally causing more death, or a reason why lessening suffering would force the idealist to go to the extreme of total elimination?