Unified Salad-Likeness Coefficient

[h3h3h0h0]

Since there are 2 main coefficients which determine how salad-like something is (ingredient and arrangement), I propose a unified coefficient that combines the 2 in some way. A logical method would be to multiply e to the power of each coefficient.

First, this means that salads with high scores would have a rating increase scaling with how high the score is. This is because it becomes increasingly difficult to accept incrementally un-saladlike foods as salads (i.e. for each decrease in salad-likeness there is an increasing decrease in acceptability). On the other hand, the increase of acceptance goes up for the foods that are right on the edge of being a traditional salad (like a fruit salad, which, even though it does not have usual salad ingredients, is almost universally considered a salad).

Second, unlike straight multiplication, it does not prevent a food from being a salad by simply having one of the coefficients being 0. This makes sure any unmixed salads are included properly. Multiplying by 1 means it does not completely nullify the salad-ness, but it instead moves solely with the other coefficient.

so C_salad=e^a_c e^i_c

For example, a piece of salmon nigiri (which was mentioned in the original proposal) has 2 ingredients (salmon, rice), and thus has i_c of log₂(2)*2=2, and has a_c of 0. The combined value is then calculated to be 7.38. Note that this is still more salad-like than water, and thus should have a higher value (the combined coefficient would be the same, 0, if only multiplication was used).

Water has i_c of 0, and a_c of 1 (since you can rotate a usual symmetrical cup along the vertical axis). The combined value would therefore be 2.71. It isn't 0, so it fits the "everything is a salad" proposal, but it is significantly lower than the nigiri.

One possible further improvement would be to scale it based on the distance from the current food with the highest known C_salad food. However, this may change as more salad-like foods are discovered, and thus is not very concrete.

[gmalmquist]

I like this idea; please feel free to submit a pull request with this (probably as another entry in the appendix)