Freelance Web/Graphic designer, photographer & community campaigner
General

Shills, Astroturfers, and Sockpuppet accounts at the Central Fife Times

An investigation carried out into the identities of several online trolls that frequent the Central Fife Times has revealed a strong possibility that staff at the local paper are attempting to manipulate public opinion by masquerading as readers.

With the advancement of digital technologies it has become easier and more commonplace to manipulate public opinion and steer a narrative amongst citizens. Often the manipulation is to push political agendas, silence dissent, or promote products or corporations.

A variety of techniques are deployed from shill accounts, trolling, sockpuppets, astroturfing, and many more techniques available to disrupt and manipulate public opinion.

The usage of online manipulation tools is being used across the political spectrum. It is now widely known that both Trump and Clinton used bot accounts in an attempt to create a public narrative that favoured their political agenda.

Emerging Research & Tools

To counteract the spread of Fake News, Trolling, Shills, and Astroturf campaigns, a variety of individuals and groups have developed a wide range of tools to unmask the anonymous commentators, discover bot networks, and identify the corporations managing online manipulation tools.

One such area of scientific research is Stylometry, which is the application of the study of linguistic style. It is often used to attribute authorship to anonymous documents or those where the authorship has been disputed.

With the advancement in the fields of Machine Learning and Artificial Intelligence,  a number of tools have emerged. Most prominent is EMMA, a self-learning technology, described by her creators as a “machine learning and artificial intelligence to study the innards of each author’s writing style and attributes authorship on their basis.”

The creators of EMMA say:

“More than 50 math parameters stand behind every author’s writing identity. As thousands of tests on works of novelists, journalists, researchers, professors and students show, Emma proves to be 85% accurate in attributing authorship. This is the best result in the authorship attribution ever achieved by engineers worldwide.”

EMMA is aimed at a cross market of end users, including publishers seeking to protect their works from plagiarism, lawyers defending a client’s copyright, and journalists requiring assistance through the provision of ‘solid evidence on authorship manipulation and other misconduct’.

Methodology and Results

I decided to use EMMA to attempt to unmask the trolls on the Central Fife Times website’s comment sections and the Letters page of the print edition, with a specific focus on the most prolific ones: “TheSpectator”, “Campaigner55”, “Inside Left”, and “Inside Right”.

These four commentators especially Spectator and Campaigner are the most visible on Central Fife Times and often make personal attacks against anyone who dares to present a differing political viewpoint from Fife Labour.

I collated a majority of their comments into separate documents and created author accounts on EMMA. For control, I also created an author account for the Central Fife Times editor, Jim Stark, using only his editorial pieces as a control piece for the investigation.

During the processing of a letter from ‘Inside Right’, the A.I. EMMA made a very unexpected and interesting match, claiming a 100% positive identity with the editor of the Central Fife Times.

Surprised by the result, I abandoned the initial project and began to process other letters that have been published under pen names in the Central Fife Times listing the results below:

Pen Name Source Article (.txt file) EMMA Identity Match (%)
Blue Brazillian CFT-BlueBrazilian 86
Woodie CFT-Woodie 100
Gas Man CFT-GasMan-1 93
Meedies Minder CFT-MeediesMinder 80
Play Parker CFT-PlayParker 100
Memorial CFT-Memorial 100
Parker CFT-Parker 87
CACA CFT-CACA 100
Pathway CFT-Pathway Not Matched
Gas Man CFT-GasMan 100
Park Observer CFT-ParkObserver 66
Postie CFT-Postie 55
Drain Man CFT-DrainMan 85
The Gambler CFT-TheGambler Not Matched
Hub Supporter CFT-HubSupporter 87
Welfare Council Supporter CFT-WelfareCouncilSupporter 100
Fountain CFT-Fountain 98
Bin Laiden CFT-BinLaiden Not Matched
Inside Left CFT-InsideLeft 56
Inside Right CFT-InsideRight 100

From the results, there is a high percentage of “hits” indicating a very strong probability that Jim Stark (CFT Editor) is responsible for writing the majority of letters that appear online and within the printed version under noms de plume.

  • 6 out of 20 articles checked were a 100% match with the writing style of Jim Stark
  • 14 out of 20 articles checked were over a 80% match with the writing style of Jim Stark
  • Only 3 articles did not match the writing style of Jim Stark

It could be argued that some of the more inane and harmless letters are simply the editor trying to fill space for a paper that is failing to find enough copy from an ever dwindling readership for its letter pages.

The more sinister and dangerous aspect is that at other times, letters from invented users such as “Gas Man”, “Inside Right” and “Meedies Minder” have all been used, by the paper, to attack individuals and groups whilst masquerading as popular public opinion. This looks like a deliberate attempt to misrepresent public opinion and limit public debate.

If these results are accurate, they suggest that the paper is dishonest and guilty of subjecting readers to systematic deception. With a confirmed 85% accuracy rate claimed by the developers of EMMA, it is hard to avoid the conclusion that the Central Fife Times is little more than a propaganda rag.

The Central Fife Times can recover their journalistic integrity if they stop publishing letters from noms de plume in the paper and pseudonymous comments online as well as tackling the insidious trolling which afflicts their site.

Article written by:

Tech geekery, blogging, volunteering, graphic/web design, photography, digital activism and community-based campaigns are some of the things that I am passionate about.

Leave a Reply

Your email address will not be published. Required fields are marked *