Optimizing searches on short_desc

[Joel Peshkin]

I noticed that most searches for strings in short_desc do not take 
advantage of the fulltext index. 

3 questions:

1) myk - Do you have any data showing what fraction of all BMO searches 
are "contains all words/strings" on the summrary?

2) bbaetz - would this be better using MATCH (short_desc) AGAINST 
('+word +word' IN BOOLEAN MODE) instead of the current index-confounding 
query?

3) should we change it?