Some data on human behavior in a Twitter community

Recently BloombergTV has decided to use the #BTV hashtag to promote special events on their channel. This post contains some data regarding one such event.

English: (Photo credit: Wikipedia)

Those who have read articles here for some time will be familiar with the ongoing experiment in digital community known as “the #BTV hashtag.

The short version is that, since 2007 Burlington Vermont has been actively using the hashtag #btv. In the years since then there have been several challenges ranging from automated bots polluting the channel to usage in Bahrain during the Arab Spring (and beyond). It’s a short, three letter little hashtag and this makes it attractive.

In addition, as any Twitter user will tell you: no one owns a hashtag. This means anyone can just start using one–even if it’s been in use. This is part of Twitter’s charm.

Skilled social media strategists and practitioners will first observe whether a tag is in use before developing and deploying any strategic or important work. Then they will orient their goals and businesses according to what they observe. After that, they will make a decision about the best way to proceed and then enact it.

Enough blah blah. Here’s some data from BloombergTV’s use of #BTV.

#BTV: An hour and one minute with BloombergTV and Burlington, VT

This data is from 12:15pm through 1:15pm. During that time 420 tweets carried the #BTV hashtag.

Bloomberg’s show was on from about 12:30pm to around 12:50pm December 4, 2012.

Bloomberg’s stated objective was to¬†“to prompt smart conversations” relevant to their programming.

The following graphs represent all Tweets containing #btv within this time period.

The above graph shows how many tweets occurred each minute in the time period which were either on-topic for Bloomberg or about Burlington Vermont. The yellow area covers the time period during which Bloomberg was airing their special program. Tweet counts are raw and include retweets as well as tweets by automated services and so on. On-topicness was determined by me, a sentient being. When in doubt I favored Bloomberg.

The graph above shows the total cumulative tweets occurring each minute in the time period. Again they are assigned either on-topic for Bloomberg or on-topic for Burlington Vermont. As above the yellow area covers the time period during which Bloomberg was airing their special program. Tweet counts are raw, include retweets etc etc.

A flat line in the above would be indicative of a flatline in terms of conversation, smart or otherwise.

Additional data available from the BloombergTV/Burlington data set

There are additional bits of interesting data that can be pulled from this selection of data which covers 420 tweets during the time period. For example:

  • Resonance: the scale and scope of retweeting behavior (essentially people saying “ditto” and passing along).
  • Total potential reach: the sum of all followers of retweet behaviors (though this is best used with care as only a fraction of one’s followers see a given twitter post).
  • Total human-like voices: many retweets originate from services which simply follow a news source such as Bloomberg and retweet everything they do. Filtering data for actual humans helps assess how much “smart conversation” could be occurring.

And several other interesting sets of metrics.

Eventually I’ll either calculate these metrics or release my data set or both. For now though, I’ll stop with the above two metrics because I think they can tell us a great deal when assessing whether Bloomberg was able to achieve their goal, whether they might have benefited from choosing a different hashtag or whether the hashtagging tactic was an effective approach at all.

Though I’m sure you can guess that my assessment is that Bloomberg’s method is ineffective, you may be surprised at my assessment for Burlington, Vermont.

I’ll save that specific analysis for a future post.

Leave a comment

Your email address will not be published. Required fields are marked *