louisgray.com - Latest Comments in louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Re: louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Howard greenstein — Tue, 18 Nov 2008 10:01:06 -0000

Louis:
The way it calculated my rank was totally different based on if I was typing my name into the top box, and if I logged in. So, I suspect your rank above would change.
But yes, there's something funny here.

Re: louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Jesse Farmer — Mon, 17 Nov 2008 13:14:12 -0000

Louis,

Why argue? The data is there -- no need to speculate. You're probably right, though.

You're surprised that the list isn't what you expected. To me that's good -- hey, something about Twitter that doesn't involve Robert Scoble! To you it's bad because it doesn't correspond to your intuition and is therefore somehow "wrong."

My point was two-fold: one, neither of us know all the variables that go into TR; two, even if we did, we don't know precisely how they're combined into that final number.

For example, it might not just be the volume of the @replies, but also who is doing the replying that matters. Think of how PageRank works. Each outbound link is a vote whose worth is weighted by the rank of the linking page. Replace "outbound link" with "@reply" and you have a pretty direct analogy: a few votes from highly-ranked people might be worth more than a lot of votes from not-so-highly ranked people.

Besides, if "common sense" is the metric why not just come up with The Louis Gray Twitter 100 and call it a day?

Just promise to put me on it. :)

Re: louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Louis Gray — Mon, 17 Nov 2008 12:23:02 -0000

Jesse, I'm not knocking Twitterank because the usual suspects aren't
at the top. Many of them are. What doesn't make sense to me is how an
unused account ranks higher than some of the most active, and how
accounts that in theory rank lower in every metric are given a higher
score. In your example of the # of replies to a person, I would argue
that @loiclemeur and @scobleizer get a lot more @replies than do I,
let alone @google, the mysterious unused account.

The leaderboards don't have to match, but they should emulate common
sense.

Re: louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Jesse Farmer — Mon, 17 Nov 2008 11:54:05 -0000

You're knocking Twitterank because it doesn't put the usual suspects at the top?

There's one key variable you're missing from your equation, viz., the number of people replies sent to a given person.

Re: louisgray.com: Twitterank's Leaderboard: Odd, Mysterious and Broken

Matt — Mon, 17 Nov 2008 04:17:43 -0000

I couldn't agree more. A few in the top 50 are known spammers. A list without Kevin Rose, Pete from Mashable and Robert Scoble is a joke whether you like them or not. Maybe twitterank has a built in hidden agenda?