At high levels of play people are so good at not dying that the average champ-killer champ will be about 7-12 kills... and that's the assassins. Posting monster kill counts only proves that you're noob enough that such a thing is even possible. The difference between knowing how to play and being genuinely good is when an enemy gets a few kills in a row and you change your planned item set to compensate. Huge feeds among any of the 10 players happen in maybe 1 in 50 games, possibly less.
And I haven't read the last 6 pages, but I'm sure plenty of people have pointed out that KDR is a terrible stat to judge overall usefulness. Damage Dealt is probably the best one if you must look at a single stat, since it seems to correlate well to general consensus of observed skill in the majority of champions, but certain tanks, supports, and bursters don't usually represent well there either.
The problem is that people confuse score with effectiveness because they're used to games like basketball where the higher score wins. In LoL, the first team to get a nexus kill point wins, and everything else is indirect and murky. Many of the generally most indicative stats are available, but this is a far cry from a real measure of effectiveness.
