DiggStatus: Digging Deep Into Digg Users' Statistics -- Page 2

December 31, 2006

[ Visual Aid: Chart Below! ] Do Top Users Control the Front Page?
I decided to put the Top Submitters in perspective with the rest of the Digg community. Many people complain about the Digg algorithm allowing the "cool kids" to have priority over average users. Though I cannot disprove this theory, I can show reasonable evidence that indicates that this behavior may not be the doing of the Digg algorithm. There have been occasions when Top Users' submissions seem to get on the front page with fewer Diggs (~30?) than average users' submissions. However, on December 15th, my first promoted submission (DiggStatus) hit the front page with about 50 Diggs.

I set out not to defend the Digg algorithm with this experiment, but to show evidence of this bias. I wouldn't want to make claims about Digg's algorithm being flawed without having proof. After all, with all those front page stories having familiar submitters, how could the playing field possibly be fair? I analyzed the user statistics gathered by DiggStatus and noticed trends that seem to indicate that the community might be impacting the success rate of the Top Submitters, rather than an additional weight applied by the Digg algorithm.

Connectors
One key to a successful viral marketing campaign is utilizing the type of individual that has the ability to spread your message to many other people. The advent of the internet has revolutionized viral marketing. For an internet viral marketing campaign to be successful, there is one type of individual that can make or break your campaign: the blogger. The blogger acts as a connector between you, an individual or company with a message, and hundreds, thousands, or even hundreds of thousands of people.

This principle is key to understanding how the community is naturally increasing the success rate of the Top Submitters. The Top Submitters, on average, have the most people befriending them. This means their articles, on average, are presented to more people than a typical user, thanks to the handy feature Digg has where you can see what your friends are submitting, digging, and commenting on.

Combine the extra exposure with the way Digg's promotion algorithm (supposedly) takes into consideration the rate at which a submission is being dugg, and you are bound to have a higher success rate.

Analysis
The graph below illustrates the impact the total submissions has on a user's success rate and the number of users befriending him/her. First, I took my set of statistics and sorted all the users by total number of submissions, in descending order. Next, I broke them into 100-user chunks and averaged their promotion ratio and befriending user count.

The apparent trend is that the more stories a user submits, the lower their promotion ratio is. This makes sense, because if someone submits many links, more of them are likely to be low quality or uninteresting. People with fewer submissions tend to have higher promotion ratios, on average, because they might only submit stories that are particularly fascinating. This trend of descending quality breaks near the top, as the befriending user count begins to rise. The people who submit more stories accumulate more friends, and thus get more exposure for their stories. It is a typical 'rich get richer' situation in which the more stories Top Submitters submit that get promoted, the more friends they get, which in turn increases their promotion ratio.

With a high befriending user count, the Top Submitters unwillingly become Connectors with enough influence to effectively get a message (submission) out to the Digg community faster than any other user.

 

Chart
Legend below. CLICK on a bar to see a detailed view for that set. You will see the users that were averaged together to get that figure, and their individual stats.

	- Average number of people befriending
	- Average ratio for submission promotion

start -	end	befriended / promotion ratio	total submitted
1 -	100	218.05 15.8%	985.3
101 -	200	55.76 10.6%	303.72
201 -	300	29.48 8.6%	198.44
301 -	400	27.39 8.9%	149.57
401 -	500	23.23 8.7%	120.66
501 -	600	29.83 8.4%	101.88
601 -	700	18.79 8.7%	86.45
701 -	800	14.59 8.3%	75.44
801 -	900	12.16 9.2%	67.76
901 -	1000	10.72 7.5%	60.92
1001 -	1100	7.04 8.3%	55
1101 -	1200	9.89 8.8%	50.66
1201 -	1300	7.35 9.2%	46.85
1301 -	1400	8.09 11%	43.07
1401 -	1500	32.54 10%	39.7
1501 -	1600	9.93 10.4%	36.84
1601 -	1700	5.88 9.5%	34.13
1701 -	1800	9.17 9.6%	31.57
1801 -	1900	6.65 10.4%	29.37
1901 -	2000	5.24 11.7%	27.72
2001 -	2100	4.84 10.9%	26.1
2101 -	2200	6.49 11.1%	24.54
2201 -	2300	4.87 11.1%	23.29
2301 -	2400	5.17 11.6%	21.89
2401 -	2500	5.66 12.5%	20.65
2501 -	2600	9.72 12.3%	19.55
2601 -	2700	4.87 12.1%	18.52
2701 -	2800	3.85 13.6%	17.8
2801 -	2900	9.07 14.5%	16.82
2901 -	3000	3.16 12.4%	16
3001 -	3100	5.04 12.9%	15.02
3101 -	3200	5.19 12.5%	14.28
3201 -	3300	5.2 12.9%	13.75
3301 -	3400	3.18 12.4%	13
3401 -	3500	3.78 14.7%	12.31
3501 -	3600	5.12 14%	12
3601 -	3700	4.5 16%	11.19
3701 -	3800	3.5 16.5%	10.87
3801 -	3900	4.15 14.7%	10
3901 -	4000	4.41 15.7%	9.94
4001 -	4100	2.48 15.3%	9
4101 -	4200	3.48 14.7%	9
4201 -	4300	3.28 17.1%	8.17
4301 -	4400	2.92 13.3%	8
4401 -	4500	2.68 18.9%	7.46
4501 -	4600	4.07 15.5%	7
4601 -	4700	2.61 15.8%	7
4701 -	4800	3.61 22.1%	6.16
4801 -	4900	2.33 14.8%	6
4901 -	5000	2.25 18.3%	6

*Note: The user 'kevinrose' was omitted because his obscenely high befriending user count was single-handedly screwing up the trend curve. I left the user 'jayadelson' in, even though he is the reason the group 1401-1500 sticks out.

 

Page 1: Introduction
:: Page 2: Submission/Promotion Graph
Page 3: Lame Users

 

 

Copyright © 2007 - Brian Shaler | Site Map
This site is powered by Brian's keyboard.