The cherry is depulped, and then the washing step simply does not happen. The mucilage stays on the parchment and goes to the bed with it. In Cenicafé's framing this is an incomplete fermentation, stopped before the mucilage would have degraded enough to rinse away.
It sits between the other two by construction, and the house treats it as a method with a wide spread of outcomes rather than a guaranteed middle path.
What the Colombian trials found
This is where primary research and marketing part company. Cenicafé ran nine treatments in a randomised complete block with nine replications at a controlled nineteen degrees, and found no statistically significant difference in cup score across any of them.
The correctly processed washed control averaged 81.6 with no defects. The honey-intent treatments carried a twenty-two to seventy-seven per cent probability of producing cups with sensory defects, and a sixty-seven to seventy-eight per cent probability of not beating that control.
Read plainly: on those trials a honey was more likely to underperform a well-made washed coffee than to beat it. That is not an argument against honeys, which at their best are very good. It is an argument that the label on the bag predicts far less than the execution behind it.