Fetch is slow when using index and where condition on @rid

[Herve Eichwald]

I have an article cluster with 400 000 records

select from Article where @rid <> #13:250000 limit 1     takes 20 ms

while select from indexvaluesasc:Article.publishDate where @rid <> #13:250000 limit 1   takes 400ms

Why?  It should just have to examine the first article and see that @rid are different, so why is it so slow?