Paid With Interest: COI Editing and its Discontents
Image credit: Jim Pennucci.
Everyone involved with Wikipedia has some kind of interest in what it says. In the classic formulation, its volunteer editors are inspired to empower a global audience by compiling information in an accessible format. Practically speaking, though, most participate because the project appeals to their personality, their sense of justice, or there's an ego boost in deciding what the world knows about their pet subject. Its readers care simply because they want to learn something. For the most part, this works very well.
Things are rather different when the motivation is financial. Most contributors consider editing Wikipedia to promote a business a morally different endeavor, and its readers, too, may be alarmed to learn some edits are made not to benevolently share knowledge with the world, but because the writer has a material stake in how the topic is represented. And yet the structure of Wikipedia makes this tension inevitable. The site's vast influence owes something to the fact that anyone can influence it, so when those described in its virtual pages decide to do exactly that, the result is one of Wikipedia's most challenging existential dilemmas.
Wikipedia's favored terminology for this is "conflict of interest", referred to in shorthand as "COI"—although other terms such as "paid editing" or "paid advocacy" are often encountered. COI is the subject of an official guideline1 plus numerous information pages giving advice to volunteers and paid editors alike, as well as a lengthy article2 in the encyclopedia itself chronicling the historical highlights and lowlights (mostly the latter). However, none of these resources really explain how COI has evolved over Wikipedia's two decades in existence.
Fortunately, this is a topic for which I have a rare insight: in addition to being a volunteer editor of more than a dozen years, I am also the founder and chief executive of a digital marketing agency that helps clients navigate their conflicts of interest on Wikipedia. From this perspective, I will outline the history of COI as I've witnessed it, attempt to classify its disparate participants, and share my own personal story, which intersects at all points. Finally, I will identify opportunities for potential research in this field, which to date scarcely exists.
I first became aware of Wikipedia through the American political blogosphere, which I covered for a Washington, DC-based news service in the early 2000s. Among bloggers on the left and right, the usefulness of linking to Wikipedia had become an uncommon point of agreement. I soon became fascinated with this audacious effort to impose order on the messy world of knowledge, not to mention the opinionated community responsible for it.
But the reason I finally started editing, prophetically enough, was because my boss asked me to. In 2006 had I abandoned journalism to join a digital public affairs firm, where doubtless I was the person in the office who brought up Wikipedia the most. The company's CEO had become concerned with the negative slant on a friend's biographical article, and wanted to know if something could be done about it. I investigated the matter, and decided something could indeed. But I didn't merely snip away the offending passage: instead I placed a note on the discussion page saying I would add a qualifying adjective to put the matter in context, and then I did just that. This instinct would come to serve me well in a way I couldn't have imagined at the time.
In the months following, I continued making small edits to articles of personal interest. Eventually, I created my first new entry: a biography of Tom Peterson, a retailer and pitchman whose homespun TV advertisements are cherished memories of Oregonians from the 1970s and 80s. During this time, I began voraciously consuming the various policies, guidelines, and essays that explained how Wikipedia made decisions about acceptable and unacceptable content. I found these statutes to be even more captivating than the articles they regulated—it was like discovering the secret rules governing all the knowledge in the universe.
As I gained confidence, my engagement with Wikipedia evolved along two tracks. First, I started attending offline events and making friends in the movement, eventually launching a blog3 about the community I had come to consider myself part of. Second, I recognized the possibilities suggested by my initial experiment. Many of my employer's clients were the subject of Wikipedia entries, which were seldom faultless. Reading the COI guideline carefully, it was apparent that while self-interested editing was discouraged, it was not outright prohibited. I was aware that others had tried and failed to thread this needle, but believed my prior experiences with the combative blogging community would help me prevail.
So I began carefully: I created a secondary account disclosing my employer and relevant client relationships, posted simple edit requests about client issues on discussion pages, and sought out editors willing to make the changes for me. Some ignored me or said no, though frequently enough they would agree, and occasionally thank me for being up-front about my COI. Some even granted me "permission" to make the changes myself, pointing to a section of the COI guideline regarding the acceptability of "making uncontroversial edits".4 So I returned to the original arrangement: first explaining my reasoning and then making the change. On the whole, this worked out surprisingly well—every once in awhile I would run into an editor who disagreed, but before long another volunteer would come along and help us find a solution.
By early 2010, I was convinced there was a bigger market for this service than my employer understood. That summer, I turned in a letter of resignation and embarked on a tour of DC PR firms, offering up my Wikipedia expertise on a contract basis. I built a book of business, one meeting at a time, until I had enough work to bring on my first employee, and then two more by the end of 2011.
At this point, I was still reluctant to discuss how I earned a living with fellow Wikipedians, fearing their disapproval. COI editing remained a controversial topic. Even if tenuously allowed, it certainly wasn't respected, and for very good reasons: the history of COI, up to this point, was largely a series of individuals and organizations getting caught doing something when they should have known better.
This history can be divided into four distinct time periods:
2001–2005 is the prehistoric era, before Wikipedia had attained a critical mass of public awareness. In this period, it was not unheard of for contributors to make self-interested edits, but the stakes were low, and the perception was they would simply write their own autobiography, or maybe an article about their friend's band. This is why one of the project's earliest advisories against COI editing was called "Vanity guidelines".5 The signature event of this era was the public embarrassment of Jimmy Wales, Wikipedia's famous co-founder, for editing his own biography in late 2005.6 This experience no doubt shaped Wales' views on COI editing, and his pronouncements on the subject soon took on a very disapproving tone, which would last through the following era.
2006–2009 is when Wikipedia began to understand the extent of the problem. This era properly begins in early 2006 with the cautionary tale of the first business focused on creating and editing Wikipedia entries for paying clients, called MyWikiBiz.7 Its founder, Greg Kohs, was soon blocked from editing by Wales himself, and would go on to become one of Wikipedia's most obsessive critics. With the help of a software tool called WikiScanner, editors soon learned anonymous edits were being made by governments, corporations and institutions around the world, demonstrating that the old saying "everybody's doing it" applied to this all-new context as well. Wikipedia's internal governance responded with varying degrees of success. A new "COI Noticeboard"8 helped to identify suspicious patterns of edits, though actual policy changes remained elusive. The period concludes with the failure of an effort to prohibit paid editing after a long debate over the summer of 2009.9 Few Wikipedians were great fans of the practice, but the severity of the harm was not clear to everyone, and concerns about unintended consequences prevailed.
2010 to 2013 was a period of tacit acquiescence and passive avoidance. Several paid editing controversies arose, only to subside without clear resolution, including the discovery of pernicious editing for unsavory clients by the since-shuttered London PR firm Bell Pottinger,10 and the "Gibraltarpedia" scandal,11 in which prominent editors manipulated site processes to benefit their territorial tourism board client. Yet another dispute happened to focus on my work and, as I will relate, this crisis arguably led to the era's two major positive developments. First, Wales' outspoken support for the idea that while COI contributors should not edit articles directly, they should be able to ask for help and receive it. Second, the development of new community procedures to facilitate and supervise this practice. But the biggest and most consequential event was the discovery and roll-up of a vast sock puppet network associated with a company called Wiki-PR,12 whose name became a shorthand for unethical COI engagement.
In late 2011, well into the third era of COI, I became involved in highly contentious disputes on two unrelated client articles which soon became the focus of a weeks-long argument that would become the catalyst for longer lasting changes.
The first client was a well-known regional restaurant chain, one already the subject of a low-quality entry, excessively focused on corporate wrongdoing—in my view contravening established guidelines about representing topics in proportion to overall coverage, known as "due and undue weight".13 The second was an automobile industry trade association having no article at all. In both cases, I proposed completely new drafts which I had researched and written, posting them to my userspace and seeking comment from unconflicted editors.
To my great surprise, the restaurant article rewrite was approved almost immediately, and moved into place by a volunteer editor. Too quickly, it turned out, as another editor soon re-inserted material about the company's numerous controversies, and slapped the page with a COI warning tag. Meanwhile, my draft for the auto trade association was given a lukewarm approval, so I took it live by myself, but the following day it too was affixed with multiple warning templates, this time by an editor who hadn't previously participated in the discussion. Here, my failure to describe the association as a "lobby group" came in for particular criticism.
I was stunned—with the restaurant chain, I had followed the hands-off protocol exactly. With the auto group, my position was more tenuous, but I had experienced plenty of success in similar circumstances. My first move was to ask for help from a couple of editors who had assisted me on other client pages. Alas, one began edit warring, which cultivated a few new detractors, who took to Jimmy Wales' talk page asking for his comment on my activities. Editors avowedly hostile to paid editing commandeered both articles, removing positive information they considered "puffery" and amplifying critical information I had tried to make less "undue".
Thus did a three-week period centering around New Year's 2012 become the worst stretch of my Wikipedia career, as debates raged on both Wales' talk page (though he remained largely absent) as well as a related thread on an administrators' noticeboard. I was an active participant, choosing to engage where I thought I could clarify misrepresentations, but erred on the side of letting the two sides go at it, of course hitting refresh constantly, watching with trepidation as new comments appeared. Some editors supported my position, complimenting my written content and willingness to defend myself, while others accused me of being a terrible threat to Wikipedia's future and asserted their intention to closely inspect every article I had ever worked on. One critic took the step of posting an email address for a third client to Wales' page, inviting incensed editors to give them a piece of their minds.
The tide began to turn toward the end of the first week of January. The uproar, which had initially focused on my actions specifically, inspired the creation of two new WikiProjects focused on this activity in general. One sought to watchdog paid advocacy, named WikiProject Integrity,14 and another aimed to create a collaborative space for working through COI issues, called WikiProject Cooperation.15 I kept a close eye on the former, while eagerly joining the latter. Off-wiki, a dauntless PR executive named Phil Gomes joined the fray, publishing a post on his blog called, "An Open Letter to Jimmy Wales and Wikipedia",16 and shortly thereafter created a Facebook group dedicated to the topic: Corporate Representatives for Ethical Wikipedia Engagement (CREWE).17 I joined it on the spot, soon becoming an admin and one of its most active members.
What about my clients? By this time, they weren't any longer. In both cases, our agreements had concluded with the placement of each article. My continued interest was not about specific contractual obligations, but my own sense of responsibility. Fortunately, the initial rancor subsided, and a few editors from the new WikiProject Cooperation helped to reassess both entries. A great deal of work was put into the restaurant article, and eventually it was accorded the highest possible recognition: Featured Article status. I waited for the clique of hostile auto editors to move on before submitting a compromise draft, which was approved without further acrimony. One month later, a heretofore uninvolved WikiProject Automobiles editor appeared from nowhere and added just two words: "lobby group". Since then, the page has remained virtually unchanged for seven years and counting.
To fully understand COI activity on Wikipedia, we must identify the different types of participants. These categories are broad and their borders porous, since everyone has a potential COI, paid or unpaid, with some topics. After all, volunteers too have their own outside relationships and affinities, be they an employer, an ideological cause, or a sports team. Nevertheless, we can usefully split these participants into two camps: those representing Wikipedia's interests, and those representing outside interests.
Let's begin with the editors responsible for preserving Wikipedia's integrity, sorted according to their views on COI and degree of interest in the subject:
1. COI-neutral Volunteers The vast majority think very little about this topic at all, but may stumble across obvious undisclosed paid editing, or be asked by a disclosed COI contributor for assistance. Most stay out of it, while some choose to get involved on a case by case basis, only to quickly return to their primary editing interests. The first editor to help me on the restaurant article, who quickly backed away from the deteriorating situation, fits into this category.
2. Anti-COI Volunteers A relatively small number of Wikipedians think about this a lot, usually because they are offended that some editors are compensated for labor they give away freely, or are concerned about the risks to Wikipedia's neutrality posed by outsiders focused exclusively on their own interests. Ironically, in the current era it is these editors who answer COI edit requests the most, figuring it better to encourage transparency than push this activity underground. The founders of WikiProject Integrity, and those who came after my clients' articles, belong to this category.
3. Pro-COI Volunteers Effectively zero Wikipedia editors are truly proponents of COI editing as such. However, from time to time one will stick their neck out and offer active assistance, but their involvement tends to have a short shelf life, likely owing to the stresses of working with sometimes pushy private interests, not to mention the disapproval of fellow editors. The brave members of WikiProject Cooperation fit here.
Next, let's consider the outsiders looking to influence Wikipedia's content, whether focused on their own interests or acting on behalf of others:
4. Novices The least sophisticated actor, and a bit of an outlier in this list, are those who don't know a lot about the site except that it can be edited by anyone, and decide to take Wikipedia up on the offer. They may genuinely not even know there are COI rules and do not spend much time pondering the ethical implications. After all, they are usually focused on a single page, and it's almost always about themselves or their own business. For most, their involvement with Wikipedia ends in failure, and that's the end of it. But some are irritatingly persistent, and they can waste a lot of volunteers' time.
5. Self-Interested Subjects Of greater concern to Wikipedia's community are the companies, organizations, institutions, governments, and prominent individuals who are either the subject of a Wikipedia entry or who perceive their interests to be affected by the information contained within them, and resolve to do something about it. They may start by assigning the task to an employee, or hiring an outside entity to handle it for them. Their level of sophistication varies widely: some may not take Wikipedia seriously until their typically-undisclosed efforts are rebuffed. It is this category which drives the demand for Wikipedia editing services.
6. PR Agencies As a first resort, article subjects will frequently turn to the PR firms they already have on retainer. While these companies do not consider Wikipedia a particular focus, as my former employer did not, they may perform the work if their client demands it. Some try to take care of it in-house, and they might even know something about the COI guideline, though whether they risk undisclosed editing or ask for help instead is largely a matter of personal or organizational ethics. Unlike the individual editing on one's own behalf, their zeal may be tempered by the fact that it's only one assignment, and they know their limitations. Most companies, contrary to the fears of anti-COI volunteer editors, will give up if it becomes too great a challenge.
7. Freelancers At the opposite end of the scale are the freelancers who have recognized the opportunity that lies in editing Wikipedia for pay, and who are notorious in the community for advertising their services on platforms such as Upwork or Fiverr. Few are particularly sophisticated, and are more likely to take on clients that don't meet Wikipedia's strict inclusion guidelines. Those who learn enough to make their articles stick and avoid detection may graduate to the next category, and very rarely the one after.
8. Black Hats Savviest of all are the SEO and reputation management companies willing to manipulate Wikipedia for their clients in knowing breach of the site's transparency rules. Black Hats are the poster children for bad behavior. When detected, like Wiki-PR and its successor firm Status Labs, their accounts are blocked and their names added to a list of known miscreants.18 So why do they do it? The downside risk is limited by their use of throwaway accounts, and offset by the large demand for their services—even if a project blows up in their face, someone else will be asking for their help soon.
9. White Hats Finally, by far the smallest category of all are those firms offering Wikipedia assistance as a standalone service, who disclose their clients on relevant pages, and who often (but not always) propose changes for volunteer review instead of editing directly. White Hats tend to be led by veterans of the Wikipedia community, and while this does not shield them from criticism, where disagreements arise they are willing to stand by their work. It is this last category to which my firm belongs.
Although the controversies around my clients tapered off by the middle of 2012, the wider discussions continued. Most significantly, Wales finally took steps to clarify his thinking around COI. For years he had merely offered strong reprimands to the guilty, but now he exercised his moral authority in the community to make a proactive recommendation, which he called the "bright line" rule.19 As he said in an interview around this time, the "rule is simply that if you are a paid advocate, you should disclose your conflict of interest and never edit article space directly. You are free to enter into a dialogue with the community on talk pages, and to suggest edits or even complete new articles or versions of articles by posting them in your user space."20
This wasn't necessarily a new concept, but it was the first time he had communicated this position so clearly. Attempts were made to standardize it as a policy or guideline, though this failed just like the efforts to ban paid editing. Nonetheless, wishing the problem away had conclusively failed, and no competing alternative emerged. Still smarting from the fallout of my own problems, I decided my company would follow the "bright line" forever after, even if the rule never became official.
WikiProject Cooperation was a lively scene in 2012, but the excitement soon faded. The project was viewed by some as too pro-COI, and was never made part of the COI guideline, so when the early participants declined through attrition, it atrophied. Yet the "ask for help" model has lived on another way, via the Edit Request system. Rather than an organized WikiProject, making an edit request is a multi-stage process whereby a COI editor includes a template21 with a talk page message, which flags the post on an administrative page22 collecting all such requests into an organized queue, and which volunteer editors may review on their own time. While the process remains relatively obscure, the COI guideline encourages its use, and it has become, like Wales' "bright line" itself, a passable solution.
Notwithstanding the improving conditions, it always rankled how PR engagement with Wikipedia only ever made the news in cases like Wiki-PR or Bell Pottinger, the resulting stories invariably failing to mention the guideline-compliant option. If all one ever hears about is companies who get caught editing anonymously, it doesn't automatically follow that one should instead declare a COI and ask for help—it leads one to either declare Wikipedia off-limits, as many agencies have done, or just try harder not to get caught.
While there will always be some who treat Wikipedia as a system to be gamed, I've long believed these actors represent a minority of PR professionals. What everyone else needed was a signal that there was in fact a way to do right by their clients and Wikipedia at the same time. Likewise, volunteer editors needed to see that there were thoughtful individuals in the assumedly reprobate field of public relations work who were capable of taking Wikipedia and its policies seriously. Because I kept a foot in both camps, I was in an ideal position to make this happen. At long last, I was going to actively seek attention for my work.
This was the genesis for the Donovan House meeting and subsequent open letter that became a major event of the fourth era. In late 2013, I started identifying people from both sides of the Wikipedia–PR divide to participate in an open and frank discussion about COI issues. With help from Wikipedia friends and an assist from CREWE, I received commitments from approximately a dozen individuals in total, counting global PR firms, academic institutions, and individual members of the community. I secured a conference room at the Donovan House hotel in Washington, DC, and we set the meeting for February 7, 2014.
The meeting was uncomfortable at first, but ultimately a success: having a face-to-face conversation helped everyone see that there were more points of agreement than disagreement, and reasons to think the pervasive feeling of mutual distrust could be lessened. The most important thing this group could accomplish, we concluded, was for these agencies to collaborate on a statement to release publicly, acknowledging that the industry had thus far failed to treat Wikipedia with proper respect, and pledging to do right in the future.
At the time, some concerns were expressed that the statement lacked accountability measures, criticisms I considered reasonable though not discrediting. But like previous efforts on Wikipedia, the moment passed and the urgency with it. Eventually, the CREWE Facebook page fell into disuse as well, with only intermittent spikes of interest following newsworthy paid editing controversies, which continue to occur, although perhaps less frequently. While it has not led to more agencies doing a great deal of policy-compliant, request-based, disclosed COI engagement at a high level, it has at least inspired them to do so at a moderate level, engaging infrequently, and not always effectively, but following the procedure nonetheless.
According to research by independent Wikipedians, activity on the Edit Request queue has increased steadily in recent years, with the biggest spike in new requests occurring in 2018.24 From 2012 through 2018, the number of desired edits posted to the queue rose in every year save one, and as of mid-2019 it appeared to stay on the same pace.25 The number of open requests has risen and fallen, but sustained efforts have kept the backlog manageable. While this research has been extremely limited, and attributing cause and effect may be elusive, I am confident our efforts played an important role.
In the latter eras of Wikipedia's COI history, the volunteer community and WMF have taken great strides toward confronting this challenge. However, I believe more work remains, and research is required to advance the discourse once again. Most importantly, rigorous studies can help to explain how well, or how poorly, the system actually works in practice. This essay has offered anecdotal evidence, but it is necessarily limited to my own experience.
Many questions are waiting to be asked, including: How do COI contributors find information about how to engage with Wikipedia, and what pathways do they take through the site? Why do volunteer editors choose to get involved with COI topics, or avoid them? What kinds of requests are being made by COI contributors, and how frequently are they approved? Are these outcomes consistent with Wikipedia guidelines? How efficient is the Edit Request system, and the COI Noticeboard for that matter? What opportunities exist to improve these processes? And how much undisclosed paid editing is there? As of this writing, there has never been a systematic effort to find these answers.
Conflict of interest will never cease to be a matter of controversy so long as what Wikipedia says continues to matter in the public sphere. A comprehensive review of the current situation would be valuable for editors who want to minimize disruptions, readers who want accurate information, and entities with a financial stake in what the encyclopedia says about them. Whatever one's motive for getting involved with Wikipedia, and whatever one's feelings about COI, understanding the role it plays now and might in the future should be in everyone's interest.