FriendFeed Helping Twitter?

A few weeks back, I wanted to pick up C#.  I have been hacking away in Python for a couple years, but now that I am in the Developer Platform Marketing division at Microsoft, I really felt it was my obligation to pick up some C# skillz.

The first thing I went and wrote was a small app that utilizes the FriendFeed API.  It polls the service for their public timeline.  It then iterates through each entry in that list and gets the last 20 posts (I think) for each user for the entries in the public timeline.  Within the code, I added the following:

dictRegExStrings.Add(“Twitter”, “.com/(.+)”);
dictRegExStrings.Add(“Digg”, “.com/users/(.+)”);
dictRegExStrings.Add(“delicious”, “.com/(.+)”);
dictRegExStrings.Add(“Last.fm”, “.fm/user/(.+)/”);
dictRegExStrings.Add(“FriendFeed”, “.com/(.+)”);

What that allowed me to do was to take the service name and then use my list of RegExes to pull the user name for those services (in case I wanted to go ping their APIs or do other such things, like recreate a social graph…).  I have about 10 of them that I wanted to track specifcally, and I list the ones above to show how I was going about doing my matches against the FriendFeed object.  I iterate through the public feed fetches, as seen in this code snippet:

FriendFeedClient ffClient = new FriendFeedClient();

ServiceNames = new List<string>();
for (int i = 0; i < length; i++)
{
    pubFeed = ffClient.FetchPublicFeed();

I do my tabulating, and move on to the next public feed fetch.  I was happy that I got the code running, and how easy the FriendFeed guys have made it to use their service in this way.  In order to be a nice net citizen, I put a sleep in my for loop.  In any event, here’s the interesting bit: when I first wrote the code I started dumping data to a text file.  The last one I dumped before today was:

11/27/08 1:00AM
Twitter:        10433   23.84 %
Delicious:      681     1.56 %
Last.FM:        465     1.06 %
Digg:           3933    8.99 %
FriendFeed:     3157    7.21 %
Facebook:       703     1.61 %
YouTube:        580     1.33 %
Disqus:         33      0.08 %
Tumblr:         1599    3.65 %
StumbleUpon:    433     0.99 %
Flickr:         762     1.74 %
Blogs:          13273   30.32 %
Other:          7718    17.63 %

Before today, it was merely a novelty for me.  I actually ran the code again on Monday, and had seen that the Twitter traffic had trended up to 40%.  However, this week, it was announced that FriedFeed had implemented a feature allowing their users to import their Twitter friends.  It was postulated by some that this would have the impact of an increase in traffic/usage for FriendFeed.  Others went so far to say that the announcement by Twitter to up their rate limits on APIs was a shot at FriendFeed.

Since I had the data just sitting there on my computer, I decided to do another run just to see what was going on.  I ran it a couple of times.  Once with five iterations, once with 10, and something like 300 iterations.  Here are those results:

01/22/09 12:00PM
Twitter:        93121   52.45 %
Delicious:      9431    5.31 %
Last.FM:        1302    0.73 %
Digg:           9200    5.18 %
FriendFeed:     13743   7.74 %
Facebook:       4165    2.35 %
YouTube:        1337    0.75 %
Disqus:         335     0.19 %
Tumblr:         2645    1.49 %
StumbleUpon:    1827    1.03 %
Flickr:         1858    1.05 %
Blogs:          19425   10.94 %
Other:          19169   10.80 %

Look at the percentages for FriendFeed and Twitter.  Quite a dramatic difference.  More importantly is that the posts identified as FriendFeed have stayed flat, but Twitter is zooming.  Whether this translates out to people viewing Twitter through FriendFeed has yet to be seen, but the Twitter usage is way up.

Since I’ve got the code, I will continue to track this for a while.  In fact, I think I am going to create and Azure app that has a worker role that just pings FriendFeed from time to time and track this data over time, and publish the date for everyone to see.  That’ll be my next code project.