i don't think the data i pulled out is actually good enough to infer that level of detail and scrutiny, although i wanted to be able to do that sort of thing... hrmm... i could go on and on about the process, or i could just send you the script hehe...
reader would only spit out 40 at a time, so i tried to jack it up to 100000 and it gave me 1000 or so... so rather than trying to figure that out, i just decided to start playing with what i had. i then had to do a lot of string parsing of the subject (yuk,threadid isn't in the feed) and grouped author IDs by which threads they posted to, in reverse chronological order... here is a sample line from the .DOT source file, showing a grouping of author IDs:
7931 -> 3 -> 9693 -> 8378;
so this thing would go through and map out author ids in reverse like that back through the feed when it found a similar subject line (splitting on 'derground :: ' and ' :: RE ::' and other such nonsense)
so... if anything, this graph connects you stronger, perhaps, if you only immediately replied to someone who they themselves is a strong replier, and only then just the immediate prior post are you thusly connected... so newer members who are repliers of stronger, more frequent authors, may get bumped towards the center more.