A template for building community on Twitter

One of the things I’ve been thinking about a lot lately is the question “Where do you listen to?” That isn’t a typo. The place you listen to informs the questions you hear.

I think the questions I hear make a big difference in what I think about and the problems I learn to solve.

The Twitter hashtag for the town I live in has presented a number of problems that have caused me to learn new things.

This is a post about where I listen to.

A town with a hashtag — Burlington Vermont — #BTV

Burlington Vermont has been using the Twitter hashtag #BTV since late 2007. The tag is the same as our airport code.

People in our town use it to talk about all sorts of things. Politics, the lunch menu at area restaurants, ski conditions, you name it. I created an archive of the #BTV hashtag that starts around 2009 if you’d like to get a sense of it (changes to the Twitter API have limited the capability of TwapperKeeper–the value of this link may change as a result).

The conversation on #BTV is much like the town hall meeting style of discourse that is common in Vermont but maybe not in other places. As a state that went through the whole civil union debate earlier than most and boasts a firmly independent US Senator, we manage to have opposing viewpoints without too much shouting or outrage or ad hominem.

This is where I listen to.

The #BTV hashtag: generating meaningful signal

For the #BTV hashtag to get established there needed to be enough signal to make it meaningful for the participants. Back in 2007 and 2008 this was relatively easy as it was mostly just early adopters of the tool. Like many networks, the utility of the hashtag increased with the number of participants–more knowledge and insight was added with every regular user.

Some examples of the utility brought to the hashtag include:

• Local media providing simple weather information
• Lunch spots providing a heads up on menus or specials
• Actual help from actual neighbors to actual problems, like shoveling snow
• Social Traffic Nav: how bad is I-89 at a given time, reported live from someone who is hopefully the passenger in their car

These sorts of things were really enough signal to keep the channel healthy. For many #BTV is open all day alongside their work. It’s a Burlington-wide water cooler and help desk.

The #BTV hashtag: eliminating noise

There were some early challenges with noise. Sometimes the noise was human generated. Perhaps someone getting online for the first time and not realizing what was a good fit for the #BTV hashtag. That was usually solved easily by helping the person to improve. Or waiting for them to lose interest. Not enough noise to damage the utility of the hashtag.

Noisy bots automating their way through the hashtag

Then there were a few automated services. One in particular that I remember was well intentioned. A student in Iowa had written a script to turn weather information from airports into Tweets. This would have been great except that the Burlington Airport weather station releases several updates every five minutes. Some of those updates say things like “The weather hasn’t changed in the last minute.”

The weatherbot was a real mess and was, to my memory, the first real challenge to the signal vs noise ratio of #BTV. This was solved fairly easily. I tracked down the script’s author and sent him an email giving him some gentle feedback.

Before I sent the email I took the time to listen to the bot. That’s right, I read a lot of the posts generated by this twitter bot to figure out how they worked. I looked for patterns of signal in the mess of noise.

Then I determined a potential text filter that the script author could apply so that Severe Weather Warnings would continue to flow to the #BTV hashtag (I figured that would be worthwhile) but “Hey it’s sunny outside” updates would not. When I sent the email suggestion to the author he was very happy to comply (and probably happy just to know someone was seeing his work). The resulting bot now limits its output to only one or so an hour.

The method I used to help keep the noise level down was this:

1. Observe that there’s too much noise in the #BTV hashtag
2. Listen to the noise to see if there’s anything of value, worth keeping
3. Determine a way to keep the thing that’s valuable and limit the thing that’s noise
4. Politely work to get the change I want

I didn’t spend a lot of time complaining about the weather bot. Focusing on a solution seemed better. I had to listen to find the solution though.

The solution had both technological aspects and social aspects. I had to be willing to help complete the solution.

This initial success became the template that I used for many many more interactions online via Twitter and other channels.

Human noise from outside Burlington

As you can imagine, having a short hashtag is useful–especially in a medium like Twitter which has a strict character limit. Burlington’s hashtag, #BTV, also contains “TV”–the universal acronym for television. Over the past few years there have been several occasions where different television viewing audiences began commenting in the #BTV hashtag.

For the most part, these television watchers eventually moved along. The reason for this seems to me that commentary about a television station is rarely engaging enough for people to actually have a conversation about it. Most of the television-based noise in the #BTV hashtag wasn’t even useful to the people who were creating it.

Conversation about specific shows is a different matter. Lots of commentary and back and forth on actual content. But on the producers and distributors of the shows, not so much.

There was once a time when the Bulgarian TV station started to use the hashtag. No one else was, just that TV station. Following the template I developed for dealing with IEMBot, I observed the “noise” and noticed that their own audience wasn’t using it to have any conversations.

I dusted off my Bulgarian language skills and dropped a note to the Bulgarian TV station, suggesting they focus on creating hashtags around their content. They had a good laugh at my Bulgarian, appreciating my efforts though. And complied. There’s still an occasional Cyrillic tweet that flies across the #BTV hashtag now and then. But it’s just not that much of a problem.

The original template held:

1. listen
2. analyze
3. make something people will like
4. help enact the solution.

I’m fortunate in that I’m self-employed so I can devote the time to help in this way. I’m doubly fortunate that what I learn by doing these sorts of things is valuable for my clients as well.

And then there’s Bahrain. And a revolution. And a television station.

On February 17th 2011 revolution began in Bahrain. At first, other than Vermont’s own independence movement, there seems to be no big connection between Bahrain and Burlington Vermont.

Except that, if you listen to the Twitter chatter, Bahrain’s national television station is bad enough that it is a topic of conversation among the Bahrainis. And the abbreviation the people of Bahrain chose to use was… you guessed it… #BTV.

It’s quite easy to see the before and after trend here. There is obvious lift in this chart: before Feb 17th Bahrain twitter usage is a flatline and after Feb 17th it cools into a regular thrum of activity.

Some of this Twitter activity is complaining about Bahrain Television. There’s even a very funny thread called #BTVConfirms which pokes fun at the way in which Bahrain Television apparently confirms outlandish events. Consider it sort of like a Bahraini version of Chuck Norris jokes.

The end result though, is a lot of chatter that has nothing to do with Burlington being in the #BTV hashtag.

Globalism and ad hoc taxonomies

Hashtags are made up. There isn’t any standard for them or any regulation or any rights to them or anything like that. Hashtags are just a way to organize conversations. And they’re made up, they’re ad hoc.

If the Bahrainis want to use a hashtag there’s nothing really to stop them from using it. They’ve as much right to it as Burlington does.

Eventually, as the “noise” generated by complaints about the service provided by Bahrain Television has begun to drown out the “signal” about Burlington, VT other hashtags are being used. Some people are using three or four different hashtags to try to stay connected to the local conversation.

The challenge with divergent hashtag use is that the total amount of signal also diminishes. If the people who generate valuable Burlington-related content all split across three or more hashtags eventually there is not enough signal to continue the sorts of meaningful connections we’re having online.

The weaker the signal gets, the more likely it is that the new hashtags get drowned out by whatever comes along to appropriate them as well. Eventually there’s no signal. Where the previous efforts to preserve a balance of signal to noise ratio relied on eliminating noise, the Bahrain event was threatening to disperse the signal.

Since I value the relationships I form via #BTV hashtag I knew it was time to do something other than wait for Bahrain Television to align their programming with the needs of their vocal Twitter audience.

My solution for this has been more technical than I would like, but it has been met with enthusiasm. I learned how to construct Twitter streams using geocode data and prepared a couple tutorials: one for creating a Tweetdeck location search column and another for creating a Hootsuite location search stream.

I used the same tactic as always, only this time I had to focus inward–on my own community–instead of outward on Bahrain.

1. Observe there is a problem: too much noise in the #BTV hashtag.
2. Listen to the problem: what can be salvaged amid the noise? what patterns exist?
3. Decide how to leverage the pattern.
4. Create a solution.

Those who have heard me present about John Boyd’s OODA loop will see the strategy clearly in these steps. The end result is a continued ability to operate–to maintain the rich local communication that we’ve enjoyed for four and a half years. Our continued operation doesn’t require anything of the Bahrainis, the solution makes conflicting use a moot point.

The solution I developed also allows us to aggregate signals from a variety of hashtags–ultimately increasing the value of the solution. People will adopt this solution because it not only diminishes noise, but increases signal. I’m pretty happy about that.

Some lessons

The strategy of observing and listening and hunting for patterns works well. But it doesn’t happen by itself.

At any point along the way it would have been easy to give up and just complain that the channel is noisy. Or move along to the next hashtag. Until someone else with a louder voice came for that hashtag as well–can there be a Pastor Niemöller’s lament for a hashtag?

In ethnomusicology (think mashup: anthropology + music) there’s a concept called communal recreation which focuses on how a community refashions culture. We have songs that appear timeless and learn as children and when asked who wrote them might say “no one.”

But just as Big Rock Candy Mountain was written by Haywire Mac (or was perhaps his recreation of “Invitation to Lubberland”), someone makes culture in a location. Someone creates a marketplace of ideas. Someone creates a market of conversation.

It certainly doesn’t require any sort of formal authority to happen. No one needs to give you permission to listen locally. No one needs to give you permission to create a conversation. But someone does need to work at creating signal. And sometimes someone needs to work at removing noise.

In my adventures in local listening I was fortunate enough to stumble on ways to do what I love: work with others to make things people like.

• A useful weather bot (though I’ll have to ask him to make another change so it gets through the filter)
• Improved conversations for a television station in Bulgaria
• A method of segmenting Twitter by topic and geography

These three things are a byproduct of what I set out do: maintain meaningful conversations with my neighbors.