Toxic Algorithms: Hashtags in Science Communication

SciComm tools or harbingers of destruction?

Hashtags (#) are increasingly used within the scientific community to promote and disseminate research on social media platforms. However, studies investigating the dissemination of accurate scientific information across social media platforms via hashtags have yielded mixed, and sometimes dangerous results in terms of both visibility and engagement.

Hashtags as Archives

As a user-defined tagging system, hashtags enable researchers to catalog and update their work, and to search efficiently through social media spaces. For example, official institutions such as the World Health Organization can use hashtags to effectively report information on public health recommendations, pandemics, and vaccination campaigns.

“COVID19” is a hashtag included in reputable information sources like the World Health Organization.

Furthermore, many articles offer “advice” to scientists on how to use hashtags to promote their research—largely because journals often do little to publicize authors’ work themselves. Such advice typically involves creating a social media account, selecting relevant hashtags, and posting about one’s findings.

However, many scientists have relatively small followings and therefore struggle to reach targeted audiences. Building a substantial, engaged follower base requires significant time and effort, and in doing so, researchers also expose themselves to potential online harassment.

Furthermore, an analysis of 656,992 #SciComm hashtags from 28 May 2020 to 28 May 2021, indicated that hashtags alone aren’t enough to improve scientific article visibility. The inclusion of both a hyperlink or image was associated with a several-fold increase in the number of retweets and views. Furthermore, another study found that the presence of a visuals (“present information in an easily digestible, visually appealing manner for social media”) were the only significant predictor of social media engagement.

Hashtags Exposing Sentiment and Intention

Hashtags themselves have been the subject of research in social media content analysis, as they often carry sentiment and help structure online discourse. They provide unprecedented insight into public psyches.

Word cloud from Marcon et al. 2018 that scraped breastfeeding‐related content on Instagram using the hashtags (#breastfeeding, #breastmilk, #normalizebreastfeeding, and #breastisbest) in 2017. Word size reflects frequency of occurrence, and connotations and sentiment towards breastfeeding were concluded to be fairly positive.

Tones from hashtags can be classified in terms of ‘positive’, ‘negative’, ‘neutral’, ect, and the prevalent tones could be inferred to reflect public perception of health topics. This can have profound communication implications for public policy. Shifting hashtag popularity can also indicate temporal trends, and how hashtag sentiment evolves across major events in public life. For example, they illustrated how public sentiment evolves in response to major events, such as the COVID-19 pandemic.

Hashtag co-occurrence networks are powerful tool, in which the text of social media posts are ‘scraped’ and entered into a software program. They can reveal centralized nodes and clusters for topics (#breastisbest), or poster identity (#momsunder25), providing insight into how information circulates within digital communities. Hashtag co-occurrence analysis (breastfeeding v formula debates) is valuable for identifying relationships among topics and patterns of content within scientific and public communication on social media:

One study (Ghahramani et al. 2025) analyzed co-occurring hashtags from 73,288 tweets scraped from #AmericanHeartMonth between January 2019 and March 2023. The study found that hashtags such as #AmericanHeartMonth and #OurHearts could effectively amplify campaign messages related to heart health and could be leveraged to design hashtag strategies that enhance visibility and engagement.

Keyword network from Haunschild et al (2019) for hashtags in climate-change related tweets.

Hashtags & Misinformation

While hashtags might serve as an archive, who is moderating and curating these archives? Any user can add a hashtag to a post, whether relevant or not. Turns out, nobody, really.

Hashtags have been weaponized to spread scientific misinformation, particularly during the first years of the COVID-19 pandemic, by both automated “bots” and human users. Hashtags associated with political movements such as Trumpism and QAnon, as well as those expressing anti-Asian sentiment, were more commonly generated by bots. Within vaccine-related discussions dominated by automated accounts, clusters of hashtags related to misinformation—such as #fakenews, #billgates, #pandemic, and #china—were among the most frequent.

According to NPR, Instagram’s algorithm recommended posts containing false claims about COVID-19, vaccines and the 2020 U.S. election. Recognized health organizations, including the Centers for Disease Control and Prevention, the World Health Organization and the Gates Foundation, however, did not contain false claims.

Screenshot from Episode 4 of “Twitter Breaking the Bird”, common hashtags in 2020-2021.

According to the CNN documentary “Twitter: Breaking the Bird” early developers of Twitter quickly recognized how hashtags and tweets could cause users to “flock” toward particular conversational topics or even in-person events. At first, this was “cool” but took nefarious turn.

In 2020, hashtags such as #MaskOff (discouraging mask usage while COVID cases surged) frequently trended on the platform, often with minimal oversight or established infrastructure for content moderation. In the period leading up to the U.S. Capitol attack on January 6, 2021, hashtags like #StopTheSteal and other misinformation-related tags gained significant traction and mobilized people. Rapidly trending hashtags could mobilize online communities and influence real-world destructive events.

What Can STEM Institutions do?

Major academic journals like Nature have surprisingly strong social media followings, with hundreds of thousands of Instagram followers. Yet, much of their content relies on clickbait-style headlines such as “Trump links autism to Tylenol: is there any truth in it?” One commenter noted that this sensationalism undermines scientific authority:

I don’t remember Nature being financially desperate to have to rely on sensationalistic posts such as this one to have more attention. As a scientific journal you should do your job and say immidiatly in the title that this decision is wrong. There’s no scientific back up to prefent pregnant people to take this drug.

Similarly, Nature posts lighter content like “Shake It Off: Taylor Swift’s Changing Voice Shows How Our Accents Evolve”. This is a far cry from the rigorous research that built Nature’s reputation.

Cell’s Instagram often includes scientific diagrams and more informative post captions. However, post text often relies on vague hashtags like #research and doesn’t make an effort to curate its own hashtags. While the page frequently showcases journal covers, it rarely promotes individual research projects. In contrast, Science Magazine’s Instagram highlights eye-catching articles with screenshots and uses more specific, targeted hashtags. However, human-interest stories abound like “As a PhD Student with bipolar disorder, I’ve found strength in perceived weakness”) or posts highlight glossy topics like dinosaurs, pets, pumpkins (How easy is it to fudge your scientific rank? Meet Larry, the world’s most cited cat #CatsOfInstagram).

Corporate structure of the scientific journal can highly influence the content displayed on Instagram, leading to a lack of cohesive strategy in science communication styles. There is also the annoying Instagram trend of directing the user to the bio for the link instead of providing the actual URL and/or link in the post, which increase click rate of the academic article.

College and university social media accounts also tend to miss this opportunity. Their feeds are often filled with promotional content—campus weather updates, college rankings, or faculty awards—rather than substantive engagement with ongoing research. Most follow few accounts compared to the scale of followers, potentially missing a prime opportunity to shape communication channels.

For these academic institutions I surveyed, many of the accounts they followed were internal—i.e, MIT Engineering follows mainly MIT accounts, or science journals mainly follow other science journals or federal institutions.

• Nature: 227K followers 243 following

  • Inconsistent hashtag use (some posts have 0, some many)

  • Mainly following other academic journals, health institutions, or museums

• Science Magazine: 433K followers 124 following

  • Diverse hashtag use

  • Mainly following journalistic outlets

• ASM Microbiology: 65.2K followers 97 following

  • Hashtags mainly related to #MicrobeMonday

  • Mainly following other academic journals, or health institutions, or museums

• Cell Press: 55.7K followers 126 following

  • Diverse, targeted hashtag uses

  • Mainly following other academic journals, federal institutions, or museums

• MIT Engineering: 113K followers 179 following

  • Almost no hashtag use

  • Mainly following MIT related accounts

• Yale Medical School: 51K followers 282 following

  • Limited or generic use of hashtags (#Classof2029, #research)

  • Includes link to research article in post

  • Mainly follows Yale related accounts

• Iowa State College of Agriculture: 4,232 followers 279 following

  • Limited hashtag use

  • Mainly following Iowa accounts

  • Few posts on science topics

    

    2025 Instagram images from the above-mentioned academic institutions and journals.

If STEM institutions want to increase scientific literacy and visibility, they must shift their digital presence from self-promotion to genuine science communication. Too often, this responsibility is left to already overburdened researchers and authors. Institutions should take a more active role by curating their own hashtags, developing consistent communication strategies, and engaging with other scientific organizations online. These are elite institutions with large followings and huge name recognition and respect, that could be better leveraged for more effective scientific communication. By doing so, they can generate stronger networks, amplify research visibility, and make scientific knowledge more accessible to the public.

Conclusion

Hashtags have transformed scientific communication. They can serve as archival tools, sentiment indicators, and/or mobilizers. While they can enhance visibility and engagement in scientific and public health communication, they also pose risks when used to amplify misinformation or polarization.

As social media continues to evolve, understanding the structure, sentiment, and ethics of hashtag use is of utmost importance for encouraging accurate science communication.