In only 10 days I got results for the experiment that tried to detect Goog penalization. It turned out that 2 domains are penalized with a -20 fall.
The other result refers to the extent of indexation. It seems that penalized sites receive only superficial indexation. For instance, if a site is well indexed, all the word strings will be indexed, and searching for phrases within quotes will find them. If the site is badly indexed due to penalization, only individual words will be indexed, and the strings will not be detected. Interesting…
The other result is that penalization covers all subjects, even those unrelated to the main one. For instance, a domain penalized for duplicate content will be penalized for content that has nothing to do with the abused content.
However, I am not making a difference between penalization types, which probably exists. In the next test I will include one domain that was penalized for duplicates and other penalized for linking to bad neighbourhood. Let’s see if the penalizations are similar.
I am now working on a new experiment (3rd of my controlled series) using more domains, mixing penalized with healthy domains.
On the other hand, I am analyzing the directories I use for submission with reciprocal link exchange. It seems that some of them are considered bad neighbours, maybe because they include black hat sites, or they sell links, or whatever. My analysis includes only existing factors, because I am not free to upload test pages to them.
There are a few factors that warn you against bad directories: bad ranking in Google while searching for their own Home Page Title or Description, as compared with Yahoo or MSN. Also, few indexed incoming links, and other parameters. We are trying to establish the most reliable of those parameters.
For all the Penalization Detection experiments we need to focus on keywords that have 10-200 results. Less, is not enough to detect a fall in rankings. More, are difficult to detect and count.
We plan to offer a Standard Penalization Detection service (exact value), and a Penalization Diagnosis which will try to find an explanation for the issue. In most cases we detect bad linking, code problems or duplicate contents that explained the problem and could be corrected.