Twitter and the Changing Research Conversation

Is the scholarly conversation moving away from Twitter? We took a dive into Altmetric data to see just what is really going on…

Over the past three months, Altmetric has been watching the situation at Twitter with interest. As this is an attention source that is important to our customers, we wanted to go beyond the anecdotal evidence and understand how Twitter use has evolved over the years. We are lucky to have a huge database of historical twitter data, nearly 400 million tweets both with and without mentions of academic research. Using this data, we looked across the last five years to understand the ways in which the conversation on Twitter has changed. We discovered that since 2018 there have been huge changes in how scientific research is socialized and discussed, but not in the ways you may expect.

The most talked about topic on Twitter today is Twitter. Since November, people have announced their exit from the platform, or a shift in focus to prioritize a different platform. People are asking the world, and us personally, if Twitter is “dying”. In asking the data, the answer is – not in a way that is observable. We have sliced the data by volume, by subject area, by demographic and by virality, and by all measures we are seeing similar levels of activity to those observed throughout 2021 and 2022, and much, much higher levels than 2019 and earlier.

Twitter and COVID

I started as Director of Product for Altmetric in April 2020. Starting a new job is always scary, but starting a new job while you are locked in your house thanks to a global pandemic definitely adds a new level of complexity. Nevertheless, I was excited to get to grips with a new domain and a new set of customer problems to solve.

Shortly before I joined, a report was published by Imperial College London called “Report 9: Impact of non-pharmaceutical interventions (NPIs) to reduce COVID19 mortality and healthcare demand”. This report became one of the first big “talking point” papers of the pandemic, used both by people who wanted to stress the potential risks of letting COVID spread, but also by people who believed it was evidence of COVID being “fake news”.

Prior to this report being published, the highest all-time Altmetric Attention Score for a research output was for A paper on neural networks, with a score of 11,194. Twitter made up a large amount of the attention that study had received, with over 56,000 tweets. The Altmetric Attention Score for the Imperial report is 55,806 at time of writing, and the paper has been mentioned (i.e. linked to) from 145,000 tweets. At its peak, the Imperial report had an Attention Score of nearly 63,000, but saw a precipitous drop thanks to the introduction of Twitter’s COVID misinformation policy and the subsequent removal of tens of thousands of accounts (and their associated tweets).

We saw this increase in attention to scholarly research reflected across all disciplines, but particularly in the biomedical and health sciences. All in all, we went from collecting 1.9 million tweets a month to over 3 million a month.

What is even more interesting is where the increase in attention was coming from. While Twitter activity greatly increased across the board, the largest increase was seen amongst users who are not in the academic domain at all. Prior to March 2020, the Twitter profiles we saw mentioning research consisted of roughly 66% “uncategorised” users, which is users that are not considered a part of the medical, academic or research communities (you can learn more about our approach to identifying demographics in this webinar). This has increased to 71%, with biomedical sciences in particular increasing even further to 82%.

Initially we expected this to be a temporary state, calming down to the prior levels once the world had developed a deeper understanding of the pandemic, and the general public became more informed. However, it now appears that this change in scientific discourse is permanent.

Twitter’s changed landscape

In the midst of the pandemic, it was almost impossible to take the time to reflect on just how profound an effect it was having on the scientific literacy of the general public. However, with recent events causing people to wonder about the future of Twitter, and whether it still contributes to the public discourse in the way it once did, we wanted to look at a broader horizon to understand if the community perception of the platform was reflected in the data.

Before 2020, your average internet denizen would be a step removed from the research, instead relying on the work of science communicators and journalists to find and translate important research into understandable insights. Prior to the pandemic, there was already a trend towards improving access to research. The push towards Open Access, the introduction of intuitive platforms like Dimensions Analytics, and the increased popularity of preprint servers meant that people could get their research in front of vastly more people, far faster than was possible by going the traditional peer review route. These improvements, alongside an unstable macro-level environment with confusing, conflicting rules, lead to people trying to find their own answers that supported their argument.

While this increase in overall awareness of current research has been good in some ways, it has been increasingly damaging in others. The “bubble” around academic Twitter has effectively popped, meaning that the work of an academic can make a larger impact, but also draw much more potentially negative attention. It also increases the likelihood of academic output being misinterpreted or misunderstood, especially in cases where people are sharing preprints without an understanding of the differences in peer review between these and published works.

Nowhere is this new trend more evident than in the attention for a July 2021 paper from the American Journal of Therapeutics. The paper, about the efficacy of the drug Ivermectin in fighting COVID-19, immediately became a hot button topic worldwide. Ivermectin was a controversial option in treating the symptoms of COVID, and while some tweeters mentioned the paper to argue the point that the treatment works, far more used it as evidence of a greater conspiracy connected to the pandemic.

The reaction to this paper was so immediate that the spike it caused is perceivable even when looking across the data for the whole year. When broken down by demographic, the effect is even more extreme. This kind of network effect was never observed prior to 2020, and is very much an artifact of how much more engaged the general public is in our scientific work.

Twitter in the Future

Taking a longer term view, Twitter data has never been more interesting. Academia is facing a number of reckonings, from the OSTP Open Access announcement, to the rise of systemic plagiarism and the emergence of AI authorship. The general public have made their way into the academic echo chamber, and while those fresh eyes have brought a lot of misinterpretations and confusion, they also offer a unique perspective into the value of modern academic research.

The future of Twitter is still uncertain, though probably no more or less uncertain than in any social media platform in today’s macroeconomic climate. However, the data in aggregate shows that Twitter still has huge value as a platform, and it still has a long way to fall before we would be back at the activity levels of the real “Old Twitter”, which for us is Twitter before the pandemic. Even if the platform were to disappear tomorrow, the stories that our historical Twitter data tells can help inform publications and research strategy far beyond social media, and help us learn how our work can have impact beyond the academic sphere. 

You can learn more about exploring Altmetric’s social media data in this webinar.