However, the scheme has additional constriants (besides trust metrics) that are designed to decrease the chances of false positives which must be satisfied before the catalog servers will acknowledge them. This likely explains why it looks to you as though most/all signatures are 0 or 100.
It would appear those pre-selects are very selective, creating a very substantial bias in results seen by end users. However, I did find a handful of cf results that were not 100 or 0.
What I saw fits flawlessly with your explanation Jordan.
For those interested in the numbers (which do greatly support Jordan's explanation):
Out of 3422 e8 hash results logged here overnight:
cf 100: 3371
cf -0: 37
others: 14
cf 90-99: 6
cf 80-89: 4
cf 70-79: 3
cf 50: 1So, 99.59% of all e8 hits returned to the client are cf 100 or 0, but there are a few that aren't..
For comparison e4 has a pretty heavy bias too, but not so severely, presumably due to having less pre-filtering on the server end. 89.8% of my e4 results were cf 100, and only one came back as -0.
Total e4 sample 1177
cf 100 1057
90-99: 8
80-89: 9
70-79: 3
60-69: 16
50-59: 6
40-49: 22
30-39: 3
20-29: 8
10-19: 1
1-9: 1
cf -0: 1
cf < -1-99: 32
cf -100: 11
------------------------------------------------------- This SF.Net email is sponsored by: IntelliVIEW -- Interactive Reporting Tool for open source databases. Create drag-&-drop reports. Save time by over 75%! Publish reports on the web. Export to DOC, XLS, RTF, etc. Download a FREE copy at http://www.intelliview.com/go/osdn_nl _______________________________________________ Razor-users mailing list Razor-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/razor-users
