For the jitter, I did a fancy version of this idea

x: (d) => d.rating + Math.random() - 0.5

except I wrote it as a

jitter

helper so that I could use a seeded random number generator so that it always produces the same result, like so:

function jitter(value, radius) {
  return {
    transform(data) {
      const source = d3.randomLcg(42); // seeded random for determinism
      const random = d3.randomUniform.source(source)(-radius, radius);
      return Plot.valueof(data, value, Float64Array).map((x) => x + random());
    }
  };
}

We could definitely add a jitter (jitterX, jitterY maybe?) transform to make this a little easier. I’ll file a feature request for that.

Faceting seemed to work pretty well, so I’m curious if there are still reasons why it wouldn’t work for you here? The most awkward part in my experience was that I had to calculate the overall average rating for the ruleY mark that connects the average per-doctor to the overall average. Also, I first tried to use the link and arrow mark for that, but we don’t allow them to be one-dimensional currently. I filed a feature request for that too!

Also, thank you so much for sharing your work in progress like this! Whenever you share it spawns like a half-dozen feature request ideas for Plot and helps us ground what we build next! 🙏