I am asking myself this : If all common search engines see: http://directory.website.com different then http://website.com (I mean they see 2 separated websites on them) then ... if I am doing reciprocal linking with somebody *his website www.marcomedia.com* and I place a link to macromedia.com at http://directory.website.com and they link to me at http://website.com , then ... how will search engine see our links?! as reciprocal ?! or 1 way links?! Search engines thinks that sub-domains are separated websites...
My vote is that they have probaly covered this and sub-domains and main domains would be treated the same. It isn't a reach to think the algo checks reg. It certainly checks for class C IP, so not much of a reach. As usual, no one really knows. I would try a test if you have sites on different class C IPs. Do a straight test (made up keywords to boot) for a basline of SERPs/PR. Then do another via sub domain... interesting.
I have been doing that for my reciprocal links for a while now. I dont know if the theory actually works. But if the theory is true then i have gained, if it is not true, than i have lost nothing.
There is a similar case I have been wondering. How does a search engine evaluate the quality of each link? Reciprocal or non-reciprocal may be evaluated if a search engine can see links in both ways. That's why I think 3 way linking is a smart way to do. But I really like to know what exactly is quality link? How search engine determines relevant or not??