Waiving (hash)flags: Some thoughts on Twitter hashtag emoji

There’s a new Ghostbusters film coming out in July, which has already attracted a lot of coverage for various reasons, not least the response of male 30-somethings unable to accept the idea of four female Ghostbusters. This week, a new trailer for the film was released, and as part of the accompanying promotional buzz, Twitter unveiled an icon to accompany the #Ghostbusters hashtag in tweets:

Hashtags have long been a particularly distinctive element of Twitter, even with their adoption on other platforms (and the # symbol’s use as discursive marker long before Twitter). Connecting tweets on a common topic, the hashtag offers a means for concisely denoting theme, emotion, and more, within the 140 characters available for a tweet.

As with other aspects of Twitter, hashtags have seen changes as the platform and communicative practices develop over time, as part of the evolution of how Twitter is used, what the platform allows, and how it works. Platform-wide, such changes range from support for different forms of visual media, changes in how Twitter displays replies and conversation threads, retweet options, search results, algorithmic feeds, Twitter policy on abuse and harassment, and more.

Icons accompanying hashtags, like the Ghostbusters logo, are part of this wider evolution, and of various visual developments on popular social media — but these icons, originally known as ‘hashflags’, also offer several points of contention about social media activity, user-driven practices, and the role of the platform.

What are hashflags?

Twitter’s use of small icons appended to hashtags started in 2010 during the soccer World Cup, when users putting hashtagged abbreviations for competing countries (e.g. #AUS, #USA) in their tweets during the event saw the relevant flag appear as well. The flag visual alongside the hashtag meant the obvious punning portmanteau of hashflag.

Despite the popularity of the World Cup on Twitter (where large volumes of hashtag use meant large volumes of the hashflags appearing), the platform didn’t pursue further opportunities for icons for a few years. The flags reappeared for the 2014 World Cup, and since then similar hashtag icons have been unveiled by the platform, for special events, television and sports, politics, and more — the excellent archival work at hashfla.gs provides a full overview (see also the Twitter account Hashflaglist and Emojipedia).

The icons have also been dubbed hashtag emoji or Twitter (custom) emoji, among other terms, and Twitter’s own blog has moved towards referring to them as emoji (especially since many of them are not flag-based anymore). However, these are not emoji in the sense of the Unicode Consortium’s emoji: they are specific to Twitter, so if you use the same hashtag on Facebook or Instagram the icon will not show up.

Furthermore, just as Unicode emoji are rendered and supported differently across devices and platforms, Twitter’s hashtag emoji are not universally viewable — they will only appear when viewing Twitter.com and on some but not all devices (and not necessarily when seeing embedded tweets on other websites).

Examples

Hashflags have been emergent in response to major news (e.g. #LoveWins following the US Supreme Court decision on marriage equality in 2015), events, holidays, and campaigns, and also the result of partnerships with brands (e.g. Coca Cola) and production companies (several icons were released ahead of The Force Awakens). See also:

More examples (current at 20 May 2016)

The icons are associated with specific hashtags: when a user includes a relevant tag in their tweet, the image automatically appears alongside the tag. In some cases, the flag changes over time — e.g. the US ABC television channel released a series of emoji connected to hashtags for Scandal (#Scandal), Grey’s Anatomy (#GreysAnatomy), and How To Get Away With Murder (#HTGAWM), where the relevant emoji changed each week.

Sounds cool! But why does this matter?

The pictures can be pretty sweet, and can add a new element to a tweet — although this may have been more of a concern in Twitter’s text-centric past than in its current form, where the visual has greater support (including link previews, Periscope streams, Vine loops, and GIFs).

However, these icons come directly from Twitter (whether in association with other companies or not), rather than from the community. Unlike emergent and user-led behaviours of hashtagging, where hashtags are the user’s choice (and language is pliable and offers great potential) — after all, the push for the hashtag itself was a user-driven innovation — Twitter determines what is and isn’t worthy of hashflag recognition. The list of hashflags to this point is varied and seemingly haphazard; I leave it to you to determine if the Beatles catalogue being available on Spotify is the most pressing recipient of a hashflag (Spotify money behind it apart), for instance.

The motivations and processes behind choosing and creating hashflags are unknown, which is of course in keeping with other elements of Twitter and other social media platforms — see all the talk these last few weeks about Facebook’s Trending News algorithm and feed, and the processes behind determining what goes there. Twitter doesn’t have to tell its users what it is doing and why — even if its user base would like that, and has views about this.

Because this is being brought from the top down, though, hashflags as platform practice also risk overlooking — or even downplaying — user behaviour and cultures. Of course, brands, broadcasters, political parties, and more promote their own specific hashtags and encourage users to feature them in tweets, as well as the diverse interepretations and alternatives users might use instead (or as well). However, by featuring hashflags, Twitter (and/or brands and broadcasters), essentially, endorses one specific tag as the one for a particular topic.

For example, an Iron Throne icon currently accompanies #GameofThrones on Twitter, but not #GoT or fan commentary-driven tags like #DemThrones (both of which, crucially, are shorter than #GameofThrones). Although I do not have numbers for which hashtags are more popular than others, and this is no doubt a consideration when developing the icons, the fact that multiple hashtags are popularly used for covering the same topic is somewhat negated here.

Furthermore, the icon is time-limited; Twitter deactivates them once their period of relevance is over. This means that a) later tweets with the relevant hashtag will not feature the icon; and b) the icon no longer appears in tweets that were published during the period it was active.

There is also variable recognition of Twitter’s multilingual user base: some icons accompany multiple hashtags representing different languages, such as the red packet emoji used for Chinese New Year earlier in 2016 which appeared alongside 16 different hashtags in English, Chinese, Simplified Chinese, and Indonesian. This type of approach is the exception so far (both in scope and linguistic diversity), as unless relevant to particular international/non-English events, English-only hashtags dominate (and over-rule other variants) — last month, Twitter expanded #IVoted to include Spanish and French hashtags (among others) ahead of the New York US presidential primaries, which is nice but also came after several months of primaries without such extensive recognition.

There is a question then of whether Twitter is (deliberately or inadvertently) imposing homogeneity, seeking to centralise coverage under a single banner, leaving alone the fact that many comments on a topic might not feature any relevant hashtag — a hashtag alone does not represent all coverage of a given theme. For metrics and analytics by Twitter and its partners, focusing on a single hashtag offers a simple focus for studying user activity and engagement — but this is not necessarily how people use Twitter. This homogenity also privileges a particular interpretation of a term over other uses of the same tag, irrelevant to that of the icon, e.g. putting football-specific iconography on hashtags that could be used in other contexts.

Two recent/current examples (close to my own interests) highlight these ideas:

#eurovision

For Eurovision 2015, Twitter enabled hashflags (in the sense of actual flag icons, within the Eurovision heart logo) for the competing countries. In 2016, though, Eurovision was accompanied solely by an emoji of the contest’s heart logo (possibly because the expanded Unicode emoji repertoire meant that more flags were available that way?). The Eurovision emoji was also limited to the #eurovision hashtag:

While the producers of Eurovision did promote #eurovision as the contest hashtag, the #esc (Eurovision Song Contest) hashtag is well-established as an alternative (or additional) marker, sometimes more popular than #eurovision in different countries. There are also other country and broadcaster-specific hashtags and year-specific variants, which might rely less on English — previous instances of this are analysed further in research that Stephen Harrington, Axel Bruns, and myself carried out into social media and Eurovision (see this presentation and article). While covering all variants is highly unlikely (and possibly implausible), putting the emphasis on one hashtag over other options gives the apperance of ignoring users communicating with other markers (or legitimising one over the other).

#ausvotes

Similarly, the upcoming Australian federal election has a green and gold ballot box alongside the #ausvotes hashtag. As my colleagues at QUT and I have noted in our social media and politics research, #ausvotes has previously denoted Twitter coverage of Australian elections in 2010 and 2013, and has inspired state-specific variants like #qldvotes and #wavotes. But this does not mean that no other options are used, such as #ausvotes2016 (distinguishing it from the previous #ausvotes instances at federal level, not counting state elections, Senate re-run elections, spills, and so on — Australian politics is a veritable soap opera). Other options, like #Election2016 run the risk of being too vague as other elections are taking place in 2016 elsewhere in the world, although an emoji now accompanies that hashtag (as seen above).

But they’re just fun pictograms, right? Is there a problem? Why so serious — do you hate fun?

Well, one distinguishing feature here is that hashflags are not a choice, if a user wants to use a relevant hashtag. During a hashflag’s active period, Twitter appends the icon automatically, and it can’t be removed from the tweet if the user just wants to use the hashtag unadorned. Sure, this might not be a problem or concern for many, but it doesn’t necessarily reflect a user’s wishes or decisions.

Furthermore, the way that the hashflags are supported by Twitter means that an active emoji appears in any tweet with the related hashtag. This extends to long-published comments: an archival tweet with a corresponding hashtag will also feature the icon even if it didn’t use them originally (since that wasn’t an option). For instance, the Iron Throne pictogram was only launched in April 2016 for the sixth season of Game of Thrones, but it still has the ability to travel in time on Twitter:

Conversely, a hashtag might also be used because it has a relevant emoji, and this short lifespan works against it. By only featuring the icon for a short period, Twitter risks denying or overwriting user intentionality (would users employ particular hashtags for TV shows if there weren’t the flags associated — especially since using the hashtag is the only way to unlock the image?). Or indeed, users might actively seek out alternative hashtags (or not use hashtags) for the reason that they do not have an associated icon.

This platformed approach to impermanent content has broader implications. While social media might be experienced in the moment, the longevity of content is important for long-term analysis of how these platforms are used, how communication and user practices develop, and many more considerations. Back in 2012, Axel Bruns and I wrote about Twitter as the ‘first draft of the present’ (compared to journalism as the ‘first draft of history’), due to the instant response to, and coverage of, breaking news and events, as a form of collective sense-making; Axel and Katrin Weller have been exploring this idea recently with regards to archiving this social media content, too: preserving this content for posterity and for analysis is important. (There is also the question, too, of whether or not archives of this content (including those acquired through automated scraping and API queries) and analytical software support the hashflags).

How hashflags work then means that the lifespan of content varies — and the platform itself is actively changing the archive. For the user, a published tweet isn’t editable, just deletable, but Twitter essentially edits tweets of its own accord by changing how they are seen at different times — content within the tweet is temporarily visible, and then gone… (and, if the same hashtag gets a new icon in the future, what appears in older tweets is not necessarily the same icon as at the time of publication). Compare the screenshot above of my Eurovision tweet from the day of the 2016 final, and then the same tweet as seen on the Twitter website a week later (or any #LoveWins tweet posted in June 2015 which once did have the rainbow heart icon):

Wider considerations for social media communication

Twitter hashflags and custom emoji can complicate considerations of social media communication, including practices of hashtag non-use and resistance. They also are a very different feature to emoji more broadly (referring here to the Unicode library). Emoji are, ultimately, a visual language which provides many options for user creativity and personal preferences. Emoji can stand in for particular things (objects, ideas), without needing to be explicit (or at times because being explicit isn’t endorsed), whether alone or in sequence. Twitter’s hashflags, on the other hand, don’t allow for this flexibility of expression and interpretation. (While Twitter’s hashflags are separate to Unicode’s emoji, related debates remain applicable here, including representation, additions and changes, and visibility).

Twitter’s time-limited hashflags also fit into a wider trend of the fleeting visual on social media, such as Facebook’s support for temporary profile pictures, as a momentary change from a user’s more permanent image (particularly in response to tragedy, holidays, or events). But rather than the user deciding that this is an image that should only last for a week, Twitter’s custom emoji is the platform taking choice away from the user.

While perhaps a minor concern for many, this type of approach raises questions about how Twitter the company works, particularly regarding platform development and awareness and response to user practices. Whether in consultation with others, partnership with events or brands, or just determining this itself, Twitter is actively advocating for particular tags (and narratives) over others by promoting the tags with visuals rather than those without; by saying this is the tag, it’s kind of saying that, in its view, other users are doing it wrong.

The hashflag then lies awkwardly at the intersection of what Twitter knows its users tweet about — politics, live events, sports, popular culture (long-established topics, highlighting practices and particular strengths of Twitter) — and what it wants its users to tweet about as well (and how they should do this)… It is indicative too, perhaps, of Twitter’s response to particular trends.

Twitter is a commercial entity, and seeks to be profitable. As brands, shows, fans, and other content producers propose and develop their own custom emoji, GIF keyboards, and other types of (visual) social media communication content, Twitter is trying to maintain and interest (and extend) its user base. This includes drawing on popular practices and by working with brands and companies (for a price) to engage with users by promoting products by, for example, running special campaigns and access using this content.

Hashflags, like promoted moments and tweets, are ways for Twitter to make money, which ultimately is its goal, as well as engage its users — but what more do they actually add to the user experience or discussion beyond a small, fleeting image directed by the platform rather than the user (and which does not necessarily recognise user choices, practices, and cultures)?