I ran the Nmix program to see if it worked, but after 2 hours it's still going, meaning that I probably did something wrong. But I emailed Jay, who supposedly knows how to use the program and hopefully he'll get back to me sometime.
While that was running in the background I decided to learn a bit more about kurtosis. Something with positive kurtosis is pointy with thick tails and something with negative kurtosis is flat with thin tails. This is because kurtosis doesn't affect the variance. Since more points are concentrated in the middle with positive kurtosis, the variance would decrease unless more points were added to the tails. This stuff about the tails is important and I didn't really pay attention to it before.
But I just got an email from Jay, which I'll analyze over the weekend and try out next week.
Hmm. The short Nmix doesn't ever end either. Something is clearly wrong. Oh well, that's what I'll do next week when the computers actually have power.