I have kind of a love/hate relationship with Twitter. I used to describe it as being like a cocktail party full of interesting people who had also brought their libraries with them. I have had lots of interesting conversations over there, and I am generally aware of breaking news before a lot of my (actual) friends.

On the other hand, the rampant abuse that takes place on Twitter was the spark for the startup project that I spent two years on, so I can't be considered an uncritical fan of the platform. I keep coming back to it, but I've never been entirely sure what it is good for.

This month's Storytelling With Data challenge was to curate a data set of your very own, one for which you would know the vagaries and source, so the fact that I had a question about how I was using Twitter made it a good place to start.

I had recently noticed that I was using retweet a lot more, and not composing a lot of original tweets. I thought that I used to write original tweets, so I wondered how my use has changed over time.

The advantage of using my own tweet stream is that I have external information about what was going on throughout this time period. I went back and looked at my other digital records to find dates of a number of events that might have had an impact, including the timestamp on the folder in Chrome (for the installation of the StayFocusd extension).

I did this analysis and visualization in python. (Code is available on my GitHub if you want to do the same thing with your own twitter feed.) The timeline overlaid on the first graph tells me that, at least from a "getting me off of twitter" perspective, installing the blocker on my browser was astonishingly effective. Although I was already reducing my use of Twitter, there was nearly three years of relatively flat low use... which confirms my initial sense that I'm not really writing tweets much any longer, since the low use corresponds with increasing levels of using retweet.


I think that the most surprising feature on these graphs is the trough in July and August.

This is not a concerted effort; I have never made a rule that I should use social media less in the summer. I just live in a profoundly temperate climate, and the summer days are valuable. I was surprised to see that this pattern has been so consistent that you can see it in the sum over the last nine years.

If you are interested in your own use, the code is freely available. It doesn't have an interface; you would need to make some modifications, but the timeline is in a dictionary form that makes it easy to add or subtract your individual event.

Originally published on LinkedIn