My main problem is with the premise that as long as it meets some minimum level of happiness, then it's a life worth living. That means that you keep getting a

My main problem is with the premise that as long as it meets some minimum level of happiness, then it's a life worth living. That means that you keep getting a bigger (wider) happiness area by adding those almost miserable people. But what if the threshold wasn't whether they'd rather commit suicide? What if there was no threshold? Suppose that you measured the difference from the max happiness anyone could possibly achieve with the actual happiness of the individual. Subtract the "missed out happiness" from the happiness level, and do this for each individual that has ever existed through the course of the population's history. If the average happiness is below half of whatever the max happiness is, then you add a negative value each time you add a person. This decreases the population/history net happiness. When things are good, then adding one person will have a positive effect on the population/history net happiness. Just like in real life animal populations, when things get too dense for a space and its resources, then net happiness decreases, so the optimal value lies in some middle sweet spot. Since we're calculating the value for the entire history of the population, killing unhappy people won't increase the population's net happiness value. Only adding more people will increase or decrease it.