The architecture of Twitter has never included any reference to the ownership of hashtags. They are not registered to a user or a group of users. In theory they are not controlled by anyone as any user of Twitter can use any hashtag any time they wish. Because of that, there is a risk that the original or first user intended meaning of the hashtag can be changed by other Twitter users. When this change of meaning is intentional it could be defined as hashtag hijacking – a form of cyber content attack, which takes place when a hashtag is hijacked by messages with undesirable content (Xanthopoulos et al. 2016). For example #myNYPD hashtag promoted by the New York City Police Department and intended for the public to share photographs of officers was hijacked and turned into an online protest as thousands of citizens appropriated the hashtag to highlight instances of police brutality, abuse, and racial profiling.
As a result, the ownership of hashtags needs to be defined in terms of association, rather than origination or formal administration. On Twitter it can be established by performing a Twitter search for it and checking who Twitter search algorithm associates with this hashtag. The higher the person/account is on search results, the stronger the algorithmic association. Whoever gets the top position can claim a kind of ‘ownership’. Just like the Page Rank algorithm on Google, which assigns weights to links between pages and then calculates their relative importance, Twitter’s search algorithm assigns weight to relations between hashtags and users. This concept was well described by Gerlitz and Helmond (2013) and their ‘likes economy’ created by Facebook and which is built on relational value, mediated by users’ participation. The weight of these relations helps to identify the ‘relational value’ in such search algorithms.
Algorithmic Ownership
The development of Web 2.0 helped to make this ‘relational value’ more ‘social’ than previously, when the value was derived from the relationships between web pages (PageRank). On Twitter, this relational value can be understood in terms of ownership through an analysis of the social connections between content (hashtag) and a user. This value is also constantly calculated by algorithms to generate the most relevant search results. Based on this, I developed the concept of the algorithmic ownership in terms of relational value. This is a form of distributed ownership in which content (i.e. a hashtag) is assigned to an owner (a person or persons) based on an algorithmic association of usernames (Twitter Handle) with the hashtag in the search results. This form of ownership has the following features:
- It is performative. It can only be discovered by performing Twitter search. It is impossible to discover algorithmic owners without performing the search because it is only in the search results that Twitter shows the profiles associated with the search query/hashtag. This is a platform design limitation.
- It is distributed across users, attached to multiple, shifting usernames.
- It is fluid. It can change anytime depending on variables that the algorithm associates with the hashtag at a given time and the usage of a hashtag by Twitter users. It will be more stable for less popular hashtags, but its fluidity will be much greater for trending hashtags, which generate thousands of new tweets per minute.
- It is personalized; that is, not simply dependent on usage, but dependent on who is searching. Twitter algorithms personalise search results for users based on their behaviour, so it is possible that highly contested hashtags i.e. #MAGA might have different algorithmic owners depending on the profile of the user who is searching for it e.g. Donald Trump supporters might get distribution of algorithmic ownership, while those who oppose the president and their social networks using #MAGA in a different context, might see a different distribution ‘owning it’. Multiple people/organisations can be algorithmic owners of a hashtag in different locations and for different users who are searching for it.
The use of the term ‘ownership’ might be a bit misleading, as it is often associated with exclusive rights and control over property (an object, land or an intangible object). In case of algorithmic ownership there are no exclusive rights and no control over the use of the hashtag. However, algorithmic ownership does resonate with legal understandings of intellectual property, which define property in terms of relations between people in relation to something. This might be use (e.g. rent of land or a building), or exclusion (e.g. private property) and so on. In this case, ownership refers to the association made by the algorithm between users and a hashtag, rather than the actual hashtag, even though this ownership is not recognised in the formal terms of intellectual property law. What is owned is an association between users in relation to a hashtag. This is a form of ownership that needs to be constantly confirmed by other users.
The actions required to ‘confirm’ the algorithmic ownership might be in the form of posting a hashtag several times with a different message or even the same message (spamming). The fandom trending hashtags or Hashtag Games, could be also seen as algorithmic ownership games. Other ways of associating users with the hashtag are by replying to other users using the hashtag or simply performing a search, which might be acknowledged by the Twitter algorithm as an indicator that a user has an interest in this hashtag. No one knows what actions Twitter algorithm uses to associate a hashtag search with a user and how actions are weighted. Most likely these ‘signals’ are always changing, but that does not change the fact, that the ownership of that association between a hashtag and a user, needs to be constantly confirmed in order to be continued, and that the conditions of that ownership, while in part a consequence of user actions are also provided by Twitter.
Acknowledgement
The concept of Algorithmic Ownership was developed for my PhD thesis. I am thankful for the financial support provided by the Economic and Social Research Council (ESRC) which helped me to complete this research at the University of Warwick.
References
Piatek, S. J. (2020) ‘Hashtagability: A study of the potential of hashtags to do things on Twitter’, PhD Thesis, University of Warwick, UK
Xanthopoulos, P., Panagopoulos, O. P., Bakamitsos, G. A., & Freudmann, E. (2016) ‘Hashtag Hijacking: What it is, why it happens and how to avoid it’ in Journal of Digital & Social Media Marketing, 3(4), 353-362.