At this point, we’re releasing updates to RowFeeder at least once a week.  Last night we launched a new reports section of RowFeeder and put out a blog post on where social media thinks Lebron James is headed.

• We added Adam Schoenfeld as a third co-founder to take over the “business” part of the business.
• We we’re nominated for and lost 3 Seattle 2.0 awards.
• We’ve completely revamped RowFeeder and added Facebook support.

We’ve been bad on our blog content production over here, but if you check out the RowFeeder blog, you can see we’ve been quite busy. It’ll be interesting to see how we balance our efforts over the next few weeks.  Hopefully we’ll have more time to share what’s been going on AS it happens.

From tip off until mid-night we captured over 45K tweets at an average pace of 172/minute! The game turned into a great battle down the stretch with the Lakers pulling out a 95 to 92 victory. Kevin Durant missed a good look at 3 with 7 seconds remaining. The dramatic finish caused a big spike in activity. Nearly 15% of the total volume occurred within 5 minutes of the final buzzer.

On day 1 of the conference, @Orian (who is working on a sweet new Twitter client) asked me what clients were being used. Since I was archiving Chirp tweets with RowFeeder, this was pretty easy to pull out and, surprisingly, the web was the most popular source for tweets referencing @chirp or #chirp. After Orian pointed out that somebody had gathered similar stats during the hack-day, I wanted to post my results from those tweets over the 33 hours of the Chirp conference.

Similar to the hack day stats, the web, TweetDeck and Tweetie were the top clients.

# of tweets for each Twitter client during Chirp

Broken down by number of users for each client app, the distribution looks similar.

# of Users for each Twitter client during Chirp

Finally, I was curious to what degree (if any) Twitter client usage changed during the conference given the flaky wifi. There weren’t too many surprises there, though.

# of tweets per hour (log scale) at Chirp

For the data hounds out there, here are the data sources for some of the above:
Spreadsheet of Tweets containing “chirp”
Spreadsheet of tweets/app, users/app, etc

Those Google spreadsheets are public and you should be able to add additional worksheets, so feel free.

Are there any other stats you might like to see from either client usage or the previous social graph analysis I did?

Well then, let’s have some fun with the social graph, shall we?

I’ve been looking into various graph algorithms recently in order to determine clusters of communities in social networks as well simply evaluate levels of connectedness among people. Community and relationships are core to us at Untitled and this is merely the start of more analysis we’re going to be doing in social closeness. As it relates to Chirp, I took a quick look at who was attending and saw that I barely even recognized a large majority of the people. Curious, I decided to take a look at who the most “connected” Chirp attendees were based on a few different perspectives:

• Who has the closest number of outbound relationships (friends)?
• Who has the closest number of inbound relationships (followers)?
• Based on mutual connections, what is somebody’s measure of social closeness?

An easy but ineffective approach to this would simply be to look at follower count, but I decided to analyze how connected each Chirp attendee was to everybody else, and how many hops out in their social graph this required. As opposed to the straight intersection approach, this allows us to evaluate how Chirp attendees are connected not only by their friends, but also by the people they follow and so on until the trail dries up. This approach also gives us an idea of “closeness” that we can use to determine central people in the community. I used a variation of this algorithm, which in the link’s case, is used on Wikipedia data.

The first approach, outbound relationships — the base number of people that somebody has friended. While not much of an indication of influence, it should tell us who has their eyes on the majority of the tweet streams of Chirp attendees. Not surprisingly, the first person on the list (@amous) follows 28,000 people. and @boctor is next in line, which is definitely interesting since he only follows 2,600 people. And then, of course, there’s @Pistachio, CEO/Founder of oneforty.com, who is definitely well ingrained in the developer community.

For the second approach, I took a look at inbound relationships – i.e. who are the most “followed” throughout the community. With the algorithm I used, first-level relationships definitely weigh heavily, but the other relationships we find as we explore the social graph are also taken into account. Not surpisingly, @ev and @jack are at the top of this list. But interestingly enough, between a few days ago when I first ran this analysis and today, @rsarver moved up from seventh position to fourth. This is definitely a direct result of his communication with the developer community over the past couple days.

Finally, let’s take a look at mutual relationships. Mutual relationships are important as they are generally an indication of trust or value between two parties. As opposed to the outbound data above, @Pistachio tops the list and @amous is nowhere to be found due to the number of people who follow @Pistachio back. This makes a lot of sense given her role in the developer ecosystem, and interestingly enough, @davemcclure comes in third on this list.

The graphs below show each of the perspectives mentioned as well as some trends of the “closest” people. The different levels relate to how far down the social graph the relationships are. There’s definitely some more fun to be had here, and I’ll post some more thoughts when I get back from Chirp! In addition, I’ll be taking a look at how the closeness changes after the event. My prediction and the answer to @dougw’s question: yup, and hopefully I can prove it.

Top 10 Connected People at Chirp