Friday, September 08, 2006
Digg Algorithm For Scoring Stories
Here is a condensed version of what Kevin Rose & Jay Adelson confirmed
as being some of the core elements used in determining the ranking (could substitute relevance here) for a story on digg. Again, these are my personal notes from our meeting with them and hopefully this will make it fairly clear what parts the "machine" looks for on new stories and user
diggs. Also, if you know anything about seo, then all of this will sound a bit familiar. Digg is now where Google was circa 1999.Digg Scoring Elements
1.) Not a set number of diggs in order for a story to be promoted.
2.) Brand new users are like automatic penalties in the search engines (sandbox effect).
- I asked Kevin if they were going to call this part of their algorithm diggbrother
, that got a good laugh out of everyone that was on the call.
3.) Buries are often overlooked but are key to their ranking
3a.) Buries are not boolean
in value. Digg looks at...
- Number of people that buried it before it was promoted
- What is the audience like (history, tendancies, etc...) that diggs / buries the story
4.) The time of day the story was submitted is included in the ranking algorithm.
Because people watch the digg spy, labs, etc...
5.) Digg crawls the stories that are submitted to look for additional content such as images,external links, etc...
6.) The person submitting a story is not factored in when ranking a submission.
7.) URLs are all treated equal - no preference to the mainstream.
if url has been banned. They were emphatic about this point and said they know they aren't perfect and that some good urls got banned on accident but they're more than willing to manually correct those.
- if reported by enough users
- fraudulent url, spam site, etc...
8.) Reputation plays a role. A positive reputation means...
- If you agree with the same person over and over again it's a red flag (think automatic penalties in Google & Yahoo)
- Agreeing with a friend is not a behaviour they want to penalize but if the same 20 people digg the same stories together as if they are one mind
they must be treated as one mind. Definitely a clever one-liner Jay has had to use many times in the past for sure.
- Karma DOES impact promotion (as they have said in the past).
9.) The lifeline of a story is monitored as well.
- who diggs the story and the sequence of those diggs (is it the same people digging the stories?)
10.) Age of users is a conundrum
- Readers Digest Version: It doesn't take a year to gain a good reputation
11.) They will not show who buried stories in rss feeds because they don't want to start flamewars by showing people that are burying stories.
12.) They receive 4,000 - 5,000 stories submitted a day and it's impossible to handle that kind of volume without having sophisticated algorithms running on the backend. Ultimately, they stood by their statement that Digg is all about the wisdom
of the masses.
Jay "A submission is equal weight"
By Jason Dowdell at 08:47 AM
(8) Thoughts on Digg Algorithm For Scoring Stories
Thanks for posting your notes.
Greater transparency is better for everyone; what would be best is a transcript -- but maybe you're working on it anyway.
In some ways the partial transcript generates more questions than answers, but it seems their intentions are in the right place -- for what its worth.
I guess time will tell if blunting elite diggers "diggs" will work.
And what the _next_ "gaming" effect is ... intentional or otherwise.
Thanks for your coverage of this topic.
Tony @ dji
Comments by Tony : Friday, September 08, 2006 at 10:53 AM
The notes here are just mine and John Gartner was asking most of the questions, being the veteran journalist he is, so I'll see if we can round his notes and Ev's notes up as well.
As far as the transcript goes... Don't think we'll be able to do that since we don't have one. Their pr firm may have something but I honestly doubt it. I'll let you know if we get some new stuff though.
At this point I'm so sick of hearing their name I want to throw up.
Comments by Jason Dowdell : Friday, September 08, 2006 at 11:06 AM
I'll add my notes here, they are summarized but the points can be taken from what I've wrote.
Yesterday was Jay's birthday and no one said Happy B-day on Digg. :)
Q: How does Editorial Work?
-philosophy is automated in terms of promotion
-"no editorial process"
-"admitted weaknesses that are being worked on"
Q:What makes kevin digg a story?
-"It is usually because I look at what my friends have dugg"
-I use DiggUpdate to monitor digg.com
-Digg gets around 4,000 submitted stories a day.
-They also use obvious things like IP tracking to help curb gaming and fraud.
These are all the notes i have that Jason didn't put in his post.
Comments by Evan : Friday, September 08, 2006 at 11:18 AM
Should be interesting if people who submit to Digg regularly will change their strategy or behavior, in any way, based on what was disclosed.
Comments by webmetricsguru : Friday, September 08, 2006 at 02:11 PM
"So sick of their name I want to throw up?"
Those are strong words ...
Whom do you refer? The PR firm? Digg? or Kevin Rose and Jay Adelson?
Comments by Tony : Friday, September 08, 2006 at 04:10 PM
SOrry -- another thing:
"No editorial process?"
I'm not sure what they mean.
I think what they mean is "we don't control which stories go to the front page -- but we're happy to delete users and posts we think are spam".
And there's the crux.
When stuff like that happens, it _is_ an editorial process.
t @ dji
Comments by Tony : Friday, September 08, 2006 at 04:12 PM
To be frank, I look at Digg as a website with 10 active, biased editors.
Slashdot is BY FAR a better social website.
Comments by Cristian Mezei : Sunday, September 10, 2006 at 07:30 AM
Comments by Cristian Mezei : Wednesday, September 27, 2006 at 05:32 AM
Post a Comment