Will Unethical Editing Destroy Wikipedia's Credibility?
Stay up to date with the latest headlines via email.
Wikipedia is making a tremendous contribution to the democratization of information. But the release of WikiScanner has pointed out some flies in its operational ointment. It also reminded me of a joke about a man wanting to know what 2 + 2 equaled. Everyone told him four until he came upon an accountant who whispered, "What would you like it to be?" Nothing personal against accountants, it just seems that we have become so jaded by spin that we believe nothing is absolute. How then do we separate information that is truth from lies, damned lies, and statistics? Wikipedia has an opportunity to play an important role in answering this question in a way that reaches millions of people worldwide.
Wikipedia has been attempting to get to the truth by requiring the use of facts, not opinions, in its entries and relying on the integrity of open-source editors to adhere to its rules. As WikiScanner is demonstrating, this is not enough. More transparency safeguards should be put in place. But more importantly for the long run, Wikipedia will need to resolve some kinks in its understanding of the links between facts, neutrality, and truth.
Wikipedia seeks entries that are written from a "neutral point of view" (NPOV). Every editor has a point of view, so Wikipedia has some basic guidelines for editing that include a prohibition on creating or editing an entry about one's self or organization and a requirement that editors present "facts" -- which Wikipedia defines as "piece[s] of information about which there is no serious dispute." WikiScanner is documenting that some editors have been blatantly violating these rules.
The predominant violation is that people and institutions from politicians to the CIA to Diebold to ExxonMobil to the Democratic Headquarters have been anonymously changing their own entries or the entries of their opponents, to make them more positive or negative, respectively. These acts are clearly inappropriate, but, as a problem, they appear to have some ready solutions. Adding additional levels of editor identification will make Wikipedia more transparent and will likely make these rule violations more obvious and less likely. WikiScanner works well for this, and Wikipedia should encourage its use. More aggressive administrator oversight will help, too. It appears that Wikipedia administrators have been stepping it up, actively investigating suspicious edits and locking downs some entries with severe problems. Additional steps, like coloring young passages, might also become necessary as the extent of the violations emerges.
But another editing practice, what WikiScanner creator, Virgil Griffith, called "white washing" is more problematic, because it violates the logic, but likely not the letter, of Wikipedia's guidelines. In this way, it challenges Wikipedia's reliance on factual accuracy both as neutrality and as a means to truth.
White washing is where someone replaces negative or neutral adjectives -- words or phrases -- with more positive synonyms. Here's an example of the conundrum that white washing creates for the idea that one can achieve truth through neutrality derived from facts. In May 2005, someone at a Wal-Mart IP address changed a sentence in the Wal-Mart entry about employee wages. The original paragraph, with the key sentence in bold, read:
As with many US retailers, Wal-Mart experiences a high rate of employee turnover (approximately 50% of employees leave every year, according to the company). Wages at Wal-Mart are about 20% less than at other retail stores. Founder [Sam Walton] once argued that his company should be exempt from the [minimum wage]. (Palast, 121).
The new entry edited by Wal-Mart became this:
As with many US retailers, Wal-Mart experiences a high rate of employee turnover (approximately 50% of employees leave every year, according to the company). The average wage at Wal-Mart is almost double the federal minimum wage (Wal-Mart). However, founder [Sam Walton] once argued that his company should be exempt from the [minimum wage]. (Palast, 121).
There are two problems with these changes, and neither of them has to do with the facts. The facts are accurate, and that's actually part of the problem.
According to Wal-Mart documents, Wal-Mart paid its employees an average of $9.68 per hour in 2005. According to a well-documented report by Arindrajit Dube and Steven Wertheim of the University of California, Berkeley, Wal-Mart's average wage of $9.68 per hour was between 17% and 25% less than comparable general merchandise and large merchandise stores. So, the first statement is basically true. In 2005, the federal minimum wage was $5.15 per hour. So, the second statement is also basically true.
Leaving aside Wal-Mart's violation of the self-editing guideline, both sentences pass the undisputed fact test. But they also violate the logic of Wikipedia's rule: undisputed facts equal neutrality which leads to truth. Both statements made $9.68 per hour mean something different. The first made it a criticism of Wal-Mart as an exploitive corporation, while the second made it a positive attribute, portraying Wal-Mart as going way beyond its duties as an employer.
Both statements are accurate. They're also pretty meaningless, possibly misleading. Neither strikes a reader as really neutral, either.
How could this happen? And, what does it mean about the future of Wikipedia as a democratic source of reliable information?
The first, and more obvious, problem is that both statements are incomplete. Neither directly states the actual wage of $9.68 per hour. That both statements presented the fact only indirectly through describing its relationship to something else -- as a percentage of other retailers' wages and as a multiple of the federal minimum wage -- should be a red flag for spin. The simple correction is to require the statistics themselves, in this case, the actual wage in dollars and cents. Combined with greater editor transparency, this problem is easily solved.
This leaves a second, more difficult problem of incompleteness -- the lack of contextual frame. How do we understand what the hourly wage of Wal-Mart means? On its own, $9.68 per hour means almost nothing. That is why, it appears, that the first and second entries framed the context surreptitiously. They compared the Wal-Mart wage to that of comparable employers and to the federal minimum wage, respectively. By implying a frame, both editors made the frame for understanding the Wal-Mart wage seem neutral. This meets the letter of the Wikipedia rules, but violates its logic.
But this appears to be the fault, so to speak, of Wikipedia's guidelines, rather than the editors (leaving aside Wal-Mart's self-editing violation). Facts by themselves aren't neutral because they don't have an intrinsic meaning that is universally understood. As the philosopher Thomas Nagel put it succinctly, you can't have a view from nowhere. Facts require "frames" because they only make sense in context.
Current research in neuroscience and linguistics shows that we understand reality through frames composed of neural networks in our brains. These mental structures or frames, structure our ideas, shape our reasoning and impact how we act. They define common sense.
Frames operate through the words we use to discuss the world around us by linking together values, principles and ideal models of everyday things like fairness, a living wage and what a typical corporation does. Words and phrases trigger related frames deep in our unconscious minds that give them meaning. This is the mental process through which we understand what we hear and read.
This mental process is why the Wal-Mart edits are so enlightening. They show us that describing the Wal-Mart wage as being below that of comparable employers or above minimum wage can make Wal-Mart appear to be bad or good, without ever saying so. Depending on one's mental frames, one is already predisposed to understand the implied value connection. There is no factual neutrality because our brains are built to interpret. We assign value to information unconsciously. That is understanding. Without this ability, we would continually spend paragraphs explaining the context that is unconsciously obvious to most people in a few words.
For Wikipedia, reliance on facts alone to achieve neutrality that will lead to, or is itself, an understandable truth will result in a number of on-going problems:
- entries that are -- difficult to understand -- collections of dates and statistics.
- Indeterminate interpretations that vary widely from the editor's intent due to the prevailing political frames and those brought by each individual reader.
- the creatively implied contextual frames of white washing.
Adding more facts will not solve these problems. Wikipedia must re-think its reliance on the logic of its guidelines that link facts to neutrality to truth. In other words, Wikipedia's verifiability policy -- previous publication by a reliable source -- is no longer enough.
Wikipedia can address blatant rule violations by dishonest editors through more transparency and greater administrator oversight. These changes are straightforward and some are already being implemented. White washing, however, will require a more thoughtful examination of Wikipedia's process for arriving at truth. That examination should include a discussion, best on Wikipedia itself, of recent research developments in neuroscience and linguistics. These developments demonstrate the importance and necessity of frames in understanding facts. Further, this research demonstrates that the absence of an objective or universal truth does not mean the complete anarchy of relativism nor the dominance of spin -- multiple interpretations are possible, but they are limited, and some interpretations will be better than others.
The idea that a collection of facts doesn't equal neutrality and doesn't lead to truth could be Wikipedia's undoing, discrediting open-source information as a reliable democratic force. This problem, if explicitly addressed and debated, could also be another historical opportunity for Wikipedia. If the relationship between facts and frames is embraced correctly, then Wikipedia could bring a new understanding of information to millions of people.
Here's hoping it does.
Eric Haas is a senior fellow at the Rockridge Institute.