How is the similarity between different monsters (shown on the Monster editor) determined?
The monster being edited and the standard monsters are converted to vectors in an 18-dimensional vector space. The standard monsters are then sorted by their distance from the monster being edited. Starting in SE 2.00, the default is to use a Hamming or Manhattan distance metric. In previous versions (or if list-monsters-by-hamming-distance is set to no), then a Euclidean distance metric was used. Monster distance is categorized into “very similar”, “similar”, “somewhat similar”, or “dissimilar”; the distances for each range are determined by the list-monsters-distanceX group of keys. A distance of 1 unit corresponds roughly to one edit step: for example, going from a Toughness of 2 to a Toughness of 3. The editor stops listing monsters once it has listed at least five monsters and the monsters are no longer similar. If you would like a more detailed ranking, you set the settings key list-monster-similarity-distance to yes and the computed distance will be included in the list. You may
