One intriguing aspect of Twitter’s evolution over the last seven years is the growth and then decline of the venerated hashtag. Perhaps the most defining symbol of the social media era, the hashtag emerged as shorthand for bringing the legacy world of categorical metadata to the social age. The brief posts of Twitter offered an especially potent use case in which the paucity of verbiage meant it was hard to keyword search for relevant content. Yet, as the hashtag is declining on Twitter, what might replace it in the future?
According to Google Trends, the English word “hashtag” first entered the public consciousness at the start of 2009 but really began to take off in early 2011. Google searches leveled off in mid-2013, intriguingly right around the time that the growth of Twitter itself leveled off. Search interest remained steady until early 2017 and has climbed steadily since.
Over the last two years, the most common word associated with worldwide English language searches for “hashtag” has been “Instagram.” Twitter is the second most associated word, but with 60% fewer searches and Facebook is seventh with 86% fewer searches.
It seems Instagram has revived interest in the venerable hashtag, replacing Twitter as the “home of the hashtag.”
The timeline below shows the percentage of all tweets in the Twitter 1% stream that contained at least one hashtag from January 2012 to October 2018 (using the Twitter-provided “entities” field encoded in the tweet’s JSON record).
Hashtags slowly gained in popularity through November 2013, at which point they suddenly began infusing rapidly into the Twitterverse. Somewhere in mid-2016 use of hashtags on Twitter leveled off and has steadily declined since mid-2017.
From a starting point in January 2012 of around 12% of tweets having hashtags to a high of around 48% in September 2016 to just under 40% as of October of last year, the hashtag is in decline across Twitter.
It is unclear what might be driving this decline.
Looking back on Twitter’s evolution, the rapid growth period of hashtags in the latter half of 2013 corresponds with the platform’s initial stagnation period where the daily tweet volume and unique tweeting user counts level off. Within a few months of Twitter’s growth leveling off, the hashtag explodes into popularity. This suggests it could be an emergent consensus behavior once Twitter’s user community stopped growing and began searching for better self-organizing behaviors.
Interestingly, the mid-2013 entrance of the stagnation period of Twitter also marked a precipitous collapse in replies that coincided with the rise of hashtags on the platform. As the growth of hashtag use markedly slowed in mid-2015 so too did the decline in replies level off. This raises the interesting question of whether Twitter’s early years, defined by interactive conversation, had less need for the organizing metadata of the hashtag since users were conversing directly with one another. As Twitter evolved into a more traditional broadcast publishing medium, the hashtag rose supreme as a way of organizing it all.
The decline of hashtag use on Twitter appears to closely coincide with the collapse of geotagged tweeting. The percentage of tweets with hashtags levels off precisely when the percentage of tweets geotagged with “Place” information begins to freefall, suggesting this may mark a broader transition point of the platform.
However, the strongest connection of all appears to be with hyperlinking behavior. The density of tweets containing hashtags and those containing URLs (either in the form of external hyperlinks or embedded media) is highly correlated, rising at the same time and falling at the same time. A manual examination of more than 1,000 randomly selected tweets did not turn up any errors in Twitter’s metadata fields compared to the tweet text itself. Tweets with hashtags or links in the tweet text were correctly noted in the metadata field and vice-versa.
This suggests that one possible explanation could be that the growth of the hashtag emerged as a way to describe shared hyperlinks on Twitter in which the shortened URLs of Twitter failed to convey sufficient information about the contents of the linked content, while users searched for easy ways of organizing those links, especially across linguistic differences.
Putting this all together, the hashtag appears to have entered a period of decline on Twitter, raising the question of what might replace it. In the end, however, the rise of the hashtag in the first place reminds us how much our social platforms resemble the past systems they claim to have upended.
