<img src="https://trc.taboola.com/1321591/log/3/unip?en=page_view" width="0" height="0" style="display:none">

Fact Check with Logically.

Download the Free App Today

#KeepPrisonsSingleSex: How Botnets Pushed a Hashtag to Westminster

#KeepPrisonsSingleSex: How Botnets Pushed a Hashtag to Westminster

A Logically investigation has uncovered evidence of a Twitter botnet promoting the hashtag #KeepPrisonsSingleSex. 

By Rachel Muller-Heyndyk and Joe Ondrak

On January 10, Conservative peer Baron Blencathra proposed an amendment to the government’s Police Crime and Sentencing Bill which would move incarcerated transgender women into “separate, specialized units.” Women in jail, he said, should not have to suffer “rape and violence from big, brute rapists who have decided to identify as women and get sent to a women’s unit.” The amendment was withdrawn as it would have failed on the grounds that excluding trans women from female prisons would be “cruel” and “extraordinarily dangerous,” the Opposition noted, referencing the high levels of abuse and violence transgender people experience in prison. 

Former MEP Baroness Claire Fox suggested that the House should follow the #KeepPrisonsSingleSex hashtag, which had 37k mentions on Twitter that day, as she read aloud tweets from women in support of the amendment condemning trans women’s presence in prisons. Fox repeatedly referred to transgender women prisoners as “males," but this, she explained, only demonstrated the “muddle” of conflating gender and sex. 

Fox was right to observe that the hashtag had “created a huge amount of interest outside of [Parliament].” However, the Twitter conversation she cited shows clear signs of coordinated inauthentic behavior — the 37k mentions were not 37k people, but instead, at least in part, a swarm of bots. One of the tweets that Baroness Fox read out to the House of Lords was artificially boosted by accounts Logically identified as being connected to the botnet.

A week in a social media manipulation campaign

To analyze the #KeepPrisonsSingleSex hashtag referenced by Baroness Fox as it moved across Twitter, we set up periodic listening through Twitter’s API. This allowed us to pull the conversation “live” — we did this by listening in to those who used the hashtags #KeepPrisonsSingleSex and #MakePrisonssSingleSex from 9:00 am to 3:00 pm or when the network reached approximately 1000 nodes (whichever happened first). We focused on accounts, allowing us to see who had the most influence over the conversation during this period. Each node in the graph represents an account in the conversation.

Connecting the nodes are “edges.” These are direct interactions between accounts and represent tagging, retweeting, and replying. Because of this, while network graphs are incredibly useful at gaining visual insights into platform influence, it is necessary to dive into the platform and investigate what stance the account had on the issue. The final statistic in these graphs is “community” — here, an algorithm identifies accounts that are clustered around each other or shared distinct dynamics with each other. They are then color-coded.

The series of snapshots of a week’s worth of activity on Twitter shows clear evidence of a botnet at work. A spike of activity on Monday would be expected, as Baroness Fox and the campaigning group Keep Prisons Single Sex both tweeted the hashtag to their followers a day in advance of the debate in the House of Lords. However, rather than tens of thousands of individual accounts tweeting support clustered around prominent gender critical influencers, we found a small number of anonymous accounts spamming retweets back and forth to each other. Moreover, those accounts who were the center of conversation on Monday were nowhere to be seen the following days.


Nodes: 1018
Edges: 5875
Modularity: 12 communities

On Monday, the conversation came rapidly and tightly knit. The ratio between nodes and edges indicated a lot of interaction within the network. To emphasize this point, we only identified 12 different communities in the network, all of which were close to each other with high levels of interactivity.

Curiously though, beyond the accounts for Baroness Fox and other groups like NoXYinXXprisons (Keep Prisons Single Sex’s account) who were close to the amendment to the Police Crime and Sentencing Bill (in yellow), the accounts that exercised the highest degree of influence on the network were “name+number” accounts, which often indicates a bot.

These accounts never tweeted original content, instead rapidly retweeting and cross-retweeting content containing the hashtags #KeepPrisonsSingleSex and #MakePrisonsSingleSex. This rapid and high-volume interaction with a hashtag is exactly the kind of signal that Twitter’s algorithm needs to recognize a hashtag as “trending” as it appears as though it is being used by a lot of different people, even when it’s not.

Some of the accounts in the cluster retweeted the hashtag more than 200 times, with the highest-activity account more than 600 tweets. Analysis from TruthNest, a service that detects bot accounts, gave probabilities of 80 percent or higher that some of the highest-tweeting accounts in this cluster were bots. 

Other online researchers have alluded to possible hashtag manipulation in the past. Researcher David Allsopp, writing for Trans Safety Network, noted that just a handful of accounts were spamming the same hashtag up to 200 times per day. Notably, some of the suspected bot accounts he investigated in his report overlap with our investigation. 

Four of the accounts that Logically identified as being part of the botnet retweeted one of the tweets that Baroness Fox read out in the House of Lords. 


image8-Jan-21-2022-02-39-45-82-PMNodes: 1115
Edges: 4047
Modularity: 20 communities

To find out whether the “name+number” accounts were simply impassioned gender critical proponents who didn’t have a lot of good original content, we repeated the same pull as Monday. 

Here, we can see that the high-influence “name+number” accounts are nowhere to be seen. Some of them do appear in this graph, but because accounts are sized according to influence, it is indicative of a distinct halt in their posting that they are not visible. Instead, they are replaced with influential figures in the gender critical movement responding to Monday’s trend and Baroness Fox’s speech.

The network also appears to become more diffuse. There are more communities and the network itself isn’t quite as densely packed as Monday as accounts begin to interact in a more natural way rather than rapid and coordinated retweeting.


image1-Jan-21-2022-02-39-43-88-PMNodes: 567
Edges: 1490
Modularity: 35 communities

By Wednesday, discourse around the hashtag had begun to dwindle rapidly. The network pull for the window of time allotted only yielded 567 accounts, with a much more diffuse 35 communities. 

Most tellingly is that with the inauthentic activity now halted, dissident and oppositional voices are more visible. The account @noisybits belongs to Dr. João Florêncio, senior lecturer at the University of Exeter, who tweeted criticisms of the hashtag and the Bill Amendment.

The bot accounts have, by this point, completely disappeared from the network. If these accounts were naturally part of the gender critical community on Twitter, there would still be a level of activity, especially given the level of influence they exercised on Monday. While some drop in activity would be expected, as the amendment had been discussed and rejected two days prior, the network doesn’t just show lower activity -  it also shows higher diversity. If Monday’s activity were organic there wouldn’t be fewer communities, there would be more. Instead, there are nearly three times as many distinct communities on Wednesday than Monday, when the hashtag was trending. 


image4-Jan-21-2022-02-39-42-94-PMNodes: 137
Edges: 196
Modularity: 15 communities

By Thursday, the network barely exists at all. The top image shows the entire network, with diffuse and distinct clusters without connection. The account @pinkichiban tweets pro-Gender Critical content, but is isolated from the discussion in the left-hand section of the graph (enlarged in the lower image). This shows a continued discussion and critique by Dr. Florêncio, as well as accounts linking to and attacking him. 


image5-Jan-21-2022-02-39-42-82-PMNodes: 95
Edges: 184
Modularity: 11 communities

Finally, on Friday, the network around #KeepPrisonsSingleSex and #MakePrisonsSingleSex barely exists at all. Distinct and diffuse communities that do not interact continue to post, but the volume is at less than a tenth that it was on Monday over a far greater period of time. The bot accounts are not posting. The hashtag is not trending. The real level of influence and interaction the hashtags have on the platform is revealed.

There is no evidence that the organization Keep Prisons Single Sex knew anything about the botnet.

The Times and Maycock’s research

Baroness Fox drew attention to the hashtag, but also to a research paper. She said:

“Anyway, all of this is hearsay. It is just what I am saying, or what a tweet says or what, indeed, the MoJ says about operational success. The whole area would benefit from the Government commissioning some independent research. I wonder whether noble Lords have seen the research published in the British Journal of Criminology recently and reported in the Times.”

The Times’ piece, “Trans prisoners ‘switch gender again’ once freed from women’s units,” makes a number of harmful misrepresentations of one of the few available papers on the subject. The story hinges on research by Dr. Matthew Maycock, an academic at the University of Dundee. Maycock is a former employee of The Scottish Prison Service (SPS) and carried out the research to better understand incarcerated cis women’s attitudes towards transgender inmates, whose voices he felt had been left out of the dicussion on transgender incarcerated women. 

The paper reveals varied, complex views towards transgender inmates. Many told stories of forming close bonds with transgender women. Lucy, for example, explained to Maycock that not only did she feel “not bothered in the slightest” by gender transition, but that it had helped her to come to accept her own sexuality. Others said they were indifferent, but less intimidated by them than cisgender men or lesbian women. Some women, too, were far more apprehensive and mentioned their feelings of betrayal over former inmates whom they heard had transitioned back to their birth gender after finishing their sentences. However, by cherry-picking negative experiences in the research and excluding positive experiences entirely, The Times offers a distorted picture of trans women, in which gender recognition is used as a tool to intimidate and hurt others. 

Maycock’s research makes no policy recommendations, and instead calls for more research into how being in prison informs views on transgender persons. He writes, “In relation to prison practice, the views of the participants in this study further complicate the views of prison practitioners in relation to young women in custody [...] this paper futher challenges perceptions of risk and vulnerability in reference to transgender women in diverse and at times conflicting ways. Additionally, more could be done by the SPS to communicate to cis women in custody about risk assessment process and clarification about the gender identity policy more widely.”

The only two interviewees featured for the piece are Dr. Kate Coleman, the director of the campaigning group Keep Prisons Single Sex, and Marion Calder, of For Women Scotland, another Gender Critical pressure group. Both have actively campaigned against the inclusion of transgender women in women’s prisons and against the reform of the Gender Recognition Act (GRA). In this, and another related article, Coleman accuses the Scottish Prison Service of ignoring the “evidence” on transgender women: “The SPS did not consider the views of women in prison when developing their policy, meaning that this is essentially a live experiment. That they have seemingly ignored this evidence is shocking,” adding that the SPS was being subjected to “institutional capture” by the Scottish Trans Alliance.

Logically contacted Maycock, who declined to comment, and instead directed readers towards his paper. 

The SPS has also confirmed that it will not release a breakdown of the crimes convicted by transgender prisoners currently within the system, as it could pose a security risk. Nonetheless, the issue has taken up headline space. The Times published five pieces on transgender women in prison in Scotland between November 2021 and January 2022, all of which focus on trans women prisoners as a threat to cisgender women’s safety. Four of those pieces interview Kate Coleman and two of those pieces use Maycock’s research. 

On January 12, The Times issued a correction on a separate article stating that “In England and Wales 436 male-bodied sex offenders were classified as women from 2012 to 2018” admitting that the number of female defendants who are “male-bodied” is not recorded.  

The Times reported that there were 15 transgender prisoners in Scotland — 3 trans men, 12 trans women — and The Spectator reported on January 19 that there were 146 trans women in prisons in England and Wales, fewer than five of whom were housed in a women’s prison. 


Bot detection - especially for automated accounts on social media platforms such as Twitter - is as much an art as it is a science. AI pipelines and algorithms take into account the creation date of profiles, their follower to following ratio, the time they are active, the amount they tweet over that time period, and other metrics to assign automation scores. However, there is always the chance that some humans, as weird as we are, will have behavior patterns that flag as automated or bot activity. People may be dedicated to retweeting, do so at odd times of the day, or an inordinate amount. Other people may spam account handles or hashtags. In the case of this investigation, the finding remains the same even if the some of the accounts that were influential on Monday were human; their behaviour was coordinated and deliberately inauthentic, and done so to achieve a specific platform response.

But tracking the spread of misinformation is often more complicated than simply uncovering coordinated inauthentic behavior – it can be necessary to track what stories those networks are connected to, and how it can mislead uninformed readers. 

There is almost no publicly available information about transgender prisoners, and the little that is available remains protected under privacy laws, or nestled in politically-charged framings, or in labyrinthine public disclosure documents. Research on the attitudes of cisgender women prisoners towards trans prisoners remains limited to a single published paper. Misinformation regarding trans individuals is rife. 

When a trans person is incarcerated, the prison services individually assess each case to determine whether other prisoners or the transgender prisoner themself would be safe. It is not true that a trans prisoner, solely based on self-ID, would be able to be homed in a prison that didn’t match their birth sex. The Times’ reporting, in conjunction with the reach the manipulated hashtag gave it, could very easily mislead an uninformed reader into thinking that it was.

Related Articles