Each of these things has three real-valued "genes", phi, beta, and chi. Phi is sort of a generic "fitness" attribute. It's used to select which of the things of gender I is going to reproduce in the current step. Chi influences the mate selection process described above. If it's positive, there is a preference for high Beta in the mate. If it's negative, there is a preference for low Beta. If it's 0, neutral. Beta, therefore, obviously influences the rate of a gender-II critter's chance of being selected as a mate, and also it as the effect of decreasing the effective value of Phi above. Beta is the "showing off" or "costly signalling" parameter.
Ideally, if this model were to effectively confirm the theory
nightspore is talking about (which presumably is well-supported by whatever differential model they're using in their paper) there would be nice upward selective pressure on Chi since organisms with high Beta must have damn high Phis if they're still around after pissing so much of it away into Beta. But I can't seem to get any consistent pressure on Chi no matter how I fiddle parameters. Nonetheless, like in the graph shown, every now and then I see that for runs that have domains of high and low Chi, there seems to be a faster increase in Phi while Chi is positive. I wonder why that doesn't actually translate into higher Chis? Maybe it's the lack of any spatial modelling. For then if one population chanced on the positive feedback cycle of high Chi/high Beta/high Phi it would spread and take over the ones that hadn't... 