Transaction Granularity refers to how much “complexity,” message length, number of syllables, characters, words, images, icons are involved in the enactment of a given meme or “memetic gesture.” The idea is that the longer, the “worse” it is, in terms of memetic reward. Short memes win out over long memes, at least in theory. Obviously, there are exceptions, and is it even generally true, and if so, what does this mean?
I want to say that enterprise of liberation, elimination of sexism and racism, tends to run afoul of the principle of transaction granularity. An example is more-than-two-gender liberation. It makes transactions to do with gender more complicated. There are two factors in play. It’s certainly a principle of memetics that the brain loves complexity, but it also loves reliable rewards. So there’s nothing wrong with increasing complexity, the number of memes which may be deployed, so long as the reward certainty does not decrease.
So, what does that mean? If I am a memetic agent (person) in a memetic environment (society/culture) and some movement, or exercise in changing around the memetic system that my peers and I have been used to such that I can’t say the same things I used to, and they aren’t allowed to react in the same ways they were used to, then we’re going to be unhappy with that movement. One way for us to express our unhappiness is to just carry on using our memes and calling things (gender) the same things we’ve always called them, and deploy oppressive memes (sexist, homophobic, etc. comments, actions) as we’ve done hitherto. Another is to deploy immunomemes directly attacking said movement.
 And maybe even classism, dare we hope?
 See polyvariablity and pseudomutation, for example.