martinibuster - 12:40 pm on Jun 28, 2013 (gmt 0)
Deeper, good, you're filling in some details. Now let's fill in some more.
The search engineers build-in exceptions so as to avoid false positives. A false positive is when an innocent site, one that is not the target they're looking for, loses their rankings. When it happens it's called, collateral damage.
In the case of DMOZ clones it is a fact that some of the higher rankings it is a fact that many high quality sites are listed there. It may be the case that Google uses DMOZ as a control. If a significant amount of sites listed at DMOZ loses their rankings they can fine-tune the algorithm to weed out the factors causing the DMOZ-listed sites to lose rankings.
Of coures, we don't know if Google uses DMOZ. However the process itself, checking for collateral damage, is indeed something that happens. Algorithm tuning happens before it is let out into the open Internet and then again after it is released.
Now let's return to your DMOZ example. What do you think would prevent DMOZ clones from affecting the rankings? Here are some ideas, feel free to add ideas of your own:
On the clone side:1. Most of them have poor PageRank
2. Most clones likely have close to zero relevant inbound links to their individual categories. This means that each category does not have any influence to pass along. If the page itself is not relevant, then it may not be able to pass anything but heavily depreciated PageRank, if any.
3. Using our knowledge that certain sites are excluded from counting toward ranking/de-rankings, that there is a built-in depreciation for certains pagse, it can be assumed that DMOZ clones are the kinds of sites that tend not to count.
On the Penguinized side:
1. Sites that have been Penguinized often have multiple kinds of reasons for their situation. It can be like the guy at the dance party who is holding up the wall. Is it because of his bad breath, his shifty eyes, his hipster mountainman beard or the "show me your boobs" t-shirt?
2. Is it possible that some sites accumulate so many bad signals that all it takes is a signal with the weight of a feather to push them into a Penguinized state? Should a feather-weight signal even count toward a Penguin score?
3. Please contribute more ideas. :)