As mentioned earlier, I don't think it's possible for an even-length koan to be white. This does fit with the earlier hints about how adding a single letter to a white koan will always make a black one; but adding a single letter to a black koan could make a white one. However, Dullstar did also mention there's a minimum length - so far, the shortest we have found is five letters long. I would assume the hint doesn't intend to merely rule out the empty koan, but would at least suggest that no single-letter koan can be white. We don't have any three-letter koans that are white yet, but I wouldn't be prepared to rule them out at this stage.
At a glance (there might be something I missed) it also appears that if a koan is white, the same koan in reverse is also white. This leads me to believe it has something to do with symmetry in some way.
I'd also believe it's something to do with the vowel / consonant structure. In particular, look at "BACEIBBDE" and "BACEIDFGO". These both follow the same consonant / vowel structure as PINEAPPLE, just with different letters - the former takes into account repetition (ie: "P" is always replaced with "B"), while the latter does not (ie: every consonant is replaced with another, unique consonant; and every vowel with another, unique vowel).
However, unless it's nothing to do with vowels and consonants directly, and simply a matter of specific letters with the vowels just happening to fall into the pattern correctly (or perhaps that it doesn't have to match 100%), there are counterexamples to it being this simple - "AACAH" comes to mind. (I chose to use this as the example and not "AAACH", in case it's relative (ie: VV can match CC, but eg. CCV couldn't match CVC).)
Let's try something.
PINEBPPLE
PINEEPPLE
PINEXPPLE
PINEOPPLE
PINEYPPLE
