Digging through Twitter to learn about Meerkat

Like many others, I’ve been following the rise of the Meerkat social live video streaming app. I think it’s a great app. It brings Twitter alive in a way I’ve never experienced social. And it shows us a two-way TV future that was beyond our imaginations just a few weeks ago.

I wanted to learn more about Meerkat’s adoption rate and how people are using it. To me Twitter, with its very public content and available APIs, provides a way to get at data about Meerkat, specifically the Tweets that the app publishes when a user streams out a video.

A little over a week ago, I began using the Twitter search API to download data about as many of these tweets as possible. I’m searching for the string “mrk.tv”, which is the Meerkat domain that’s in the Meerkat URLs in the text of the tweet. I’m pretty sure this isn’t quite all the tweets during that time period. Twitter says that the search API doesn’t pull everything. And Meerkat users sometimes delete tweets.

But the data I’ve pulled looks like the overwhelming majority of the Meerkat streaming tweets. It contains all the Meerkat streams I did myself or watched as far as I can remember. And even if it’s not all of them, I think it’s a pretty good representative sample. It’s every Meerkat streaming tweet I could find from March 4 to March 15 (GMT time). I worked to filter out any duplicate tweets, retweets, anything like that.

I think I’m staying within the terms of service that Twitter attaches to use of its APIs. Also note that this is the best I can infer from Twitter data. Meerkat itself certainly has to have very good data for all of this. I’m just guessing. They know. I am looking forward to Meerkat releasing its API.

I’ve posted selected (fields, not records) data I pulled here in case anyone wants to look at it: https://dl.dropboxusercontent.com/u/444378/Meerkats_3-4_to_3_16.xlsx

Here are a few quick highlights from the data I have:

Meerkat stream tweet growth

In the data I was able to pull, the number of Meerkat streaming tweets I found increased by and average of almost 17 percent a day. I don’t know what the going measure of “explosive growth” is, but that seems pretty good.

Here’s a bar graph of that:

Streamers vs. watchers

Meerkat implied it had about 156,000 users as of Saturday. In the data I collected, I could find tweets for 21,600+ distinct Twitter users who had tweeted Meerkat streams during my March 4–15 time period. I found an average of 2.3 streaming tweets per use. I found tweets from 2,200+ users who had used it to stream in that time period 5 or more times. The median number of streaming tweets I could find per user is 1.

Here are the Twitter handles is the top 10 users by number of streams in the data I collected:
1. dmgrossblatt: 88
T2. CurlySafia: 81
T2. LeleleeleleeLee: 81
4. SteveMahoney_: 76
5. armano: 74
6. azeem: 63
7. Emmerschmidt: 62
8. ericscastillo: 60
9. sarita: 58
10. JaronRayHinds: 57

What are people streaming?

It’s really hard to know other than by watching a lot of video. The best I could do was to filter and aggregate descriptive text that many users enter when they stream and that is included in the tweets. I put that into a word cloud generator. Here’s the result:

A few notes on this word cloud:

1. As one would expect, Austin and SXSW are prominent in the word map.
2. The word “Driving” might be disturbing to see, but I’ve hung my phone from my rear-view mirror a couple of times. It didn’t feel that unsafe. I was so focused on describing my driving and what I was experiencing that I felt focused and serene. And I’ve watched drivers in Belgium, Russia, Canada and motorcycle riders in Gainesville, Fla., and California, and it never felt I was watching something unsafe.
3. It’s nice to see some foreign words in the map.
4. And it’s no surprise the word “dog” is prominent. Who can resist streaming them or watching them?

Some general thoughts on Meerkat

1. Meerkat has unified (at least temporarily) small-scale television broadcasting and made it available to everyone (who has an iPhone). THIS IS INCREDIBLE. The discovery problems that are obvious aren’t a concern. What we should recognize is that shows us an even more obvious path to discovery and schedule of millions, perhaps billions, of “broadcasters” and their “shows.” Sites like MeerkatStreams.com are already starting to show us what’s live now already, and Meerkat’s schedule feature is very clever. I will be shocked if we don’t see rapid improvements in discoverability, search, and management in the weeks and months ahead.
2. Meerkat brings Twitter ALIVE. Instead of words and pictures on a phone, you see real people doing real things, and you can talk them. I don’t know how else to put that.
3. It lets people around the world and the world itself in ways you could not otherwise. I watched a scooter ride around Gainesville, Fla., and then minutes later watched a Glacier National Park stream and narrate live video of snowmelt-swollen streams. I thought of how I’d like to be prompted to watch live video from the Chattahoochee River National Recreation Area where I so often hike.
4. People have been really, really nice on Meerkat. I wonder if a two-way realtime communication makes us like this. If so, we could all use a little Meerkatting.

Some things Meerkat needs to improve

I doubt there isn’t anything on this short list Ben Rubin and his team aren’t already working on. But these would be my priorities:

1. Get the comments out of Twitter. They aren’t adding anything to Twitter and they have people confused and upset.
2. Fix the interfaces for both typing and reading comments. It’s hard to see what you are typing. And the comments are often cut off in the viewing screen. And sometimes they just don’t work. But leave the Retweets and Favorites as is.
3. Meerkat has to do something about stream meta data and streamer-to-viewer chatting. As I mentioned above, it’s very difficult to know what a stream is about until you actually watch it. Or what it was about after it was watched.