I think the data aggregation query included in the article can produce wildly skewed results because of the bug in the data collection query. I think it could lead one to conclude an index is used very lightly/rarely when it's used very heavily/frequently. I think it could lead one to conclude an index is used very heavily/frequently when it's used very lightly/rarely. Some folks have dismissed the issue and said the data collection query is acceptable. I have to disagree because the query can produce very misleading results.