Jump to content

英文维基 | 中文维基 | 日文维基 | 草榴社区

Wikipedia:Wikipedia Signpost/2015-10-07/Op-ed

From Wikipedia, the free encyclopedia


Op-ed

Walled gardens of corruption

Kazakhstan, the world's largest landlocked country, straddles Europe and China.
I first became aware of the Kazakh government's impact on Wikipedia in 2012 when I learnt that Jimmy Wales was due to visit the Central Asian country to formally bestow the inaugural Wikipedian-of-the-Year award on Rauan Kenzhekhanuly, a former Kazakh diplomat. The ceremony, planned to take place in the presence of the country's president and prime minister, sounded like a high-profile event – almost like a state visit.

In the end, Wales' journey never materialised. There was a hubbub on his Wikipedia talk page, pointing out Kazakhstan's awful human rights record and Wales' personal links to Tony Blair, who has long been castigated for his multi-million-dollar consultancy contract advising the Kazakh government on how to polish its image in the west. Several press articles appeared, raising metaphorical eyebrows, and the trip never took place (see previous Signpost coverage).

Kazakhstan and Wikipedia: A marriage made in hell

In the meantime though, the embryonic user-generated Kazakh Wikipedia was systematically overwritten with material from the state-published (and thus censored) Kazakh national encyclopedia. And with this job and other work accomplished, Wales' Wikipedian of the Year was last year reported to have returned to official government service: becoming a deputy governor in his home country and founding a Brussels think tank, the Eurasian Council on Foreign Affairs (ECFA), widely judged to be a PR front for the Kazakh regime.

When UK Labour politician Jack Straw announced that he wanted to do work for Kenzhekhanuly's ECFA, human rights organisations were in uproar. Allan Hogarth, head of policy and government affairs at Amnesty International UK, was quoted in The Independent as saying:

Kazakhstan is a vast country, the world's ninth-largest by area, blessed with enormous mineral resources.
Alas, where was Hogarth when Wales announced his Wikipedian-of-the-Year award? Rightly or wrongly, I thought much the same words could have applied to Wales at the time. When two years later, in the wake of the Stanton Foundation and Belfer Center paid-editing scandal (see previous Signpost coverage), I learned that Belfer Center director Graham T. Allison, the husband of the Stanton Foundation's Liz Allison, not only had a friendship medal from the Kazakh president, but had also authored a foreword to the good president's book, Epicenter of Peace, this was hardly likely to make me feel more sanguine.

The Stanton Foundation, administered by Liz Allison, has historically been the Wikimedia Foundation's biggest donor. If it had been able to pressure or cajole Wikimedia Foundation staff into abandoning their principles in the Belfer Center paid-editing case, despite warnings from veteran Wikipedians like Pete Forsyth and Liam Wyatt, I wondered, perhaps the Belfer Center had had something to do with the Kazakh Wikipedian-of-the-Year award too? After all, the Wikimedia Foundation (WMF) received a record-breaking $3.5 million grant from the Stanton Foundation in 2011, mere weeks after the Kazakh Wikipedian-of-the-Year award.

Wales denied it when asked about it on Reddit, adding he had never heard of Graham Allison, the Belfer Center, or Liz Allison before – despite the many millions the Stanton Foundation has given the WMF over the past few years, and despite Wales' having assisted the Belfer Center's professor Joseph Nye with a 2014 "good government and trust-building" project realised in cooperation with the United Arab Emirates government (another human rights violator). Wales said he had not bothered to research professor Nye's more detailed affiliations and was unaware of them, just as he said he was unaware of Kenzhekhanuly's prior government roles (listed in his LinkedIn profile) at the time he gave him the award.

But the news of Kenzhekhanuly's official return to government service was enough to make Wales finally, with a delay of several years, "distance himself from the Kazakhstan PR machine", as one observer put it. Following accusations on Reddit that he had repeatedly been reported to have praised the Kazakh government, while never having spoken out publicly about the lack of freedom of speech in the country, Wales even went out of his way to criticise the Kazakh government for its internet censorship. At last.

The world's ninth-largest country, with tens of trillions of dollars in mineral resources

It was these events that sensitised me to Wikipedia's content about Kazakhstan. The country – formerly part of the Soviet Union, and only independent since 1991 – is still remarkably obscure to many people, despite being the globe's ninth-largest by area, and blessed with enormous mineral resources.

According to presentations by Kazakh embassies designed to attract Western investment, the value of these resources is measured in tens of trillions of dollars (yes, tens of trillions, not billions).

Kazakhstan has serious wealth; that and the president's strong stance against nuclear proliferation have gained him a certain amount of favour with pragmatically thinking Western leaders, even while human rights organisations vociferously condemn his regime.

"Tinkering with Wikipedia"

The Kazakh government is based in Astana, Kazakhstan's futuristic, purpose-built capital. The large yurt-like structure in the far distance was designed by renowned British architects Foster and Partners.
And this is where Wikipedia comes in: because when a country has a poor human rights record, it hurts investment. Reports appeared in 2012 that the Kazakh government was taking an active interest in Wikipedia, employing PR agencies to massage entries related to the country ("Kazakhstan: Top-Notch PR Firms Help Brighten Astana's Image", "Tinkering with Wikipedia part of Kazakh government's PR strategy?").

Looking at Wikipedia's content I noticed, for example, that the article Elections in Kazakhstan held a great amount of technical detail, but not a single mention of the fact that Kazakh elections are widely considered a sham. The president, in power since 1989, when the country was still part of the Soviet Union, won his last election with 97.7% support, owing to the fact that no genuine opposition leaders were allowed to stand – something the election account in Wikipedia's article on the President of Kazakhstan provided no clue whatsoever to.

The human rights situation in Kazakhstan, as portrayed by human rights organisations

Next, let's look at the article Human rights in Kazakhstan. Before we do, here is a summary from Human Rights Watch of the situation in Kazakhstan:

Here is what Amnesty International has to say:

Here is Freedom House's take, under the subheading "Dictatorship prevails elsewhere in Eurasia":

Freedom House's democracy score for Kazakhstan is 6.61 on a scale of 1–7, where 1 is best and 7 is worst.

All of this is about as bad as it comes, right?

The human rights situation in Kazakhstan, as portrayed in Wikipedia: Torture? What torture?

Organisations like Human Rights Watch sharply condemn the Kazakh government for its human rights abuses. You would not have thought so from reading Wikipedia.
You might expect to find these assessments prominently reflected in the Wikipedia article Human rights in Kazakhstan (permalink). Not so.

The article starts,

Well, that's great!

The entire second paragraph is devoted to a report titled "Looking Forward: Kazakhstan and the United States", authored by a bi-national team comprising three Kazakh academics as well as two Americans from Johns Hopkins University (which has been called out by ABC News for taking money from the Kazakh government for academic reports) and another from StrateVarious, a "global strategy consulting" firm which, presumably, also accepts payment for its services. It says,

We may not allow free elections, have closed down opposition newspapers, shoot protesters and torture dissidents, but don't give us a hard time over it! We're trying!

It follows this up with some of the most soporific writing to be found anywhere in Wikipedia:

Are you asleep yet?

The 2015 Freedom House world map. Kazakhstan is shown slate-coloured, i.e. "not free". The freest countries in the region are Mongolia, landlocked between China and Russia and marked in green, just to the east of Kazakhstan, and Kyrgyzstan, marked in orange to the south-east of Kazakhstan.
This is all there is above the fold. It is followed by several thousand additional words of soporific detail and inconsequential import. There is nothing about torture in detention and arresting people for exercising their freedom to assembly, but plenty more Soviet-style news detail about a religious congress, more praise for Kazakhstan with its "Rule of Law initiatives", "Justice Sector Institutional Strengthening Project", "human rights dialogue", "efforts against torture", and so forth. The press freedom section ends with the confident assertion that "Kazakhstan complies with the international human rights standards" (sourced to the state-published Astana Times). And buried somewhere in this morass is a story of some Hare Krishnas getting evicted.

Soviet-style states churn out a never-ending supply of articles detailing various palatable-sounding initiatives, high-minded affirmations of constitutional rights. All you have to do to win at Wikipedia is to cite them all. Even if someone, somewhere, inserts a critical comment, place enough fluff before and after, and the reader will never get there.

And this, some Wikipedians eventually began to suspect, is exactly what has been happening in this topic area.

Wikipedians smell a rat

On Talk:Kazakhstan, NeilN initiated the following conversation a few weeks ago:

And that is about the size of it. As far as I have seen, there have been few actual deletions of content, no edit wars. Whenever critical material has been inserted, it has simply been surrounded with a never-ending stream of boring, Soviet-style news. All sourced, of course!

See for example the sections on human rights, media and the rule of law (permalink) in the main Kazakhstan article, one of Wikipedia's 1,000 most viewed articles (ranked 664 at the time of writing, and averaging more than 4,000 views a day). Yes, there is some limited criticism there, but you have to look for it like a needle in a haystack, because it first tells you about a Kazakh diplomat's international efforts, Kazakhstan's participation in the Human Rights Council, its Human Rights Action Plan, a media support centre opened in Almaty, and the 2002 creation of a human rights ombudsman ... zzzzzzzzz

Wake up again!

So, who's doing it? Well, it's noticeable that there's a whole bevy of red-linked single-purpose accounts (SPAs) that only edit Kazakhstan articles and nothing else. They only make a handful of edits, at most a few dozen, and disappear again. They overlap to such an extent that some articles' edit histories are almost entirely composed of their contributions, with regular Wikipedians only making brief appearances to disambiguate a term, add a category or fix a typo, and then disappear again.

Quite possibly, most or all of these red-linked accounts are operated by one group or person. So, where is Wikipedia's much-vaunted transparency here? Just as in the Wiki-PR and Orangemoody sockpuppet cases, we see that Wikipedia actually provides readily accessible means to obscure and falsify an article's edit history. It looks like lots of different people have worked on the articles, but there is no way of knowing. It's like the old joke about the incredibly secure house that has dozens of specialist locks on its front door, but no walls.

The search engine manipulation effect (SEME): recent research shows that Google visibility has a significant impact on public perception, opening the door to manipulation. Wikipedia, increasingly integrated into Google's Knowledge Graph and Bing's Satori equivalent, is a key factor in what search engine users see.
I don't think anyone could look at the English Wikipedia's articles on Kazakhstan and conclude that anonymous crowdsourcing has worked well here. There is no "crowd". The topic area looks like a walled garden owned by a PR team, and has done for a long time. It works so well that a Kazakh embassy tweeted the Kazakhstan article to celebrate Kazakhstan's national day. An embassy of a country with a democracy score of 6.61 on a scale of 1 to 7. That occupies 161st place (out of 180 places) in the 2014 Press Freedom Index. Tweets the Wikipedia article about its country, because it is so wonderfully flattering.

The matter is currently at the conflict-of-interest noticeboard. But given that in recent years, search engines like Google and Bing have taken to displaying Wikipedia material directly on the search engine results pages, there is potentially far more at stake here than the English Wikipedia's suite of Kazakhstan articles, all of which have been compromised to some extent.

Manipulation of online information

A recent academic paper, titled "The search engine manipulation effect (SEME) and its possible impact on the outcomes of elections", demonstrates experimentally how great an effect even subtle changes to Google rankings can have on public opinion. Making praise or criticism more or less visible, even by just a tiny bit, makes all the difference, the paper's authors argue.

We are finding time and again that on controversial topics, Wikipedia is less robust, more vulnerable to manipulation, than the average of the existing reliable sources. For all its convenience, it is also a bottleneck, the most vulnerable link in a chain stretching from original reporting to information consumers.

Indian families who went to Wikipedia to reassure themselves about the Indian Institute of Planning and Management (IIPM) were duped: a corrupt Wikipedia administrator had, for years, deleted criticism and inserted praise, continuing the work of a small sockpuppet army that was active in earlier years (see previous Signpost coverage).

The consumer warnings that were available on the web disappeared from view in Wikipedia. This bottleneck effect is increased even further with programmes like Wikipedia Zero.

In 2013, Croatia's Minister of Science, Education and Sports, Željko Jovanović, warned his country's students not to rely on the Croatian Wikipedia, as it had been "usurped" by a far right fringe group that had filled large parts of it with falsified content.
Croatian internet users had to be warned off the Croatian Wikipedia by the country's education minister, because their language version had reportedly been taken over by fascists.

A related problem is that the public tends to trust Wikipedia content too much, as Oxford scholar and former Wikipedia checkuser Taha Yasseri pointed out a few weeks ago, and understands too little about how said content comes into being.

In part, this is what motivates manipulation attempts: if a manipulator's content sticks, it's successfully become disconnected from the person who placed it. To members of the public, it is now "Wikipedia" saying whatever it is that is being said, not that person. You can see that error in thinking in press articles sometimes, when journalists say things like, "Wikipedia updated its description of so-and-so", failing to understand and convey to the public that it was simply one – typically anonymous – person changing an openly editable webpage.

The authors of a recent Oxford Internet Institute study, "Digital Divisions of Labor and Informational Magnetism: Mapping Participation in Wikipedia", pointed out that for all the good intentions of Wikipedia, "In practice, we see how existing inequalities and imbalances don't just make places invisible, but also suffocate certain voices and perspectives."

You may have thought that Wikipedia would make it harder to suffocate certain voices. We are seeing evidence to the contrary. This shouldn't come as a complete surprise of course. In 2006, Jason Scott, in an almost prophetic speech, said,

Nine years on, lives are being affected. Commenting on the Wifione/IIPM case, Indian journalist Maheshwar Peri said, "In my opinion, by letting this go on for so long, Wikipedia has messed up perhaps 15,000 students' lives."

What now?

The Kazakhstan articles may or may not be corrected. The COI/N thread has elicited sympathetic interest, but very little change in article space. But even if the articles are corrected, the fact is they stood corrupted for years. And another set of equally important articles in some other topic area may be becoming corrupted as we speak. Jimmy Wales was advised of this situation last December on Twitter, and did nothing. Nothing changed by itself; if anything, the situation got worse.

There are only about 200 countries in the world. Quite a lot, but I wouldn't have thought too many for such a large community to keep an eye on. But if Wikipedia can't even notice and deal with anonymous manipulation of its content about the world's 9th largest country by area, a country with enormous mineral wealth and outstanding global significance, what will it do for human knowledge if Wikidata and Wikipedia content is plugged into Google and Bing to become the world's default answer to everything?

We should be honest with ourselves and the public: Wikipedia can be fun to participate in. It can be entertaining to read. It has some great content and can be really useful. Its sheer breadth can be awe-inspiring. But a reference work compiled by anonymous volunteers, the way Wikipedia is today, is too vulnerable to act as a substitute for the existing plethora of voices out there. Wikipedia is not the sum of human knowledge. It is a severely abridged summary, and sometimes, a very flawed one. Away from high-profile articles that receive diligent scrutiny, it is no better than the last thing said about a topic by a stranger in the pub.

And to answer the obvious question …

Of course I could have tried to WP:SOFIXIT. Six or seven years ago, I probably would have. But the thing is, if I as a single editor am all that stands between accurate coverage of just the basics of the human rights situation in a major country and said content devolving into a farce, then Wikipedia and the public have a structural problem that is far more urgent to address than a few dozen corrupted articles on Kazakhstan.

Citing SOFIXIT is like building a dyke made of a type of sand and telling anyone who sees water coming through that they should volunteer to remain standing there for the rest of their lives and shovel more sand on, when the more sensible way forward may be to point out that there are structural flaws in the design of the dyke that need fixing.

It is well known in industry that if you end up dealing with one crisis after another, fixing the underlying systemic causes is a greater priority than attending to the individual crises. Doing the latter may make you feel you are doing great things, but it just keeps you so occupied that you cannot see the big picture. So, what are some of the structural flaws?

Structural flaws

  1. The most fundamental flaw is unrestricted anonymity. Individuals can and do register dozens or even hundreds of pseudonymous accounts and use them for more or less dishonest purposes. Many Wikipedians argue that anonymity is not a bug but a fundamental feature of Wikipedia, but it is blindingly obvious that complete anonymity facilitates mischief and manipulation, incurring significant administrative costs and adversely impacting article quality.
  2. At present, there is no obvious way in Wikipedia to give a conflict-of-interest or paid-editing problem affecting one of the site's 1,000 most-viewed articles a higher priority and more eyeballs than a problem affecting an article viewed by three people a day. No one is responsible, and whether a particular article gets attention or not is a completely random affair driven by volunteer interest.
  3. Speaking more generally, I am not sure there are many or indeed any quality improvement measures in Wikipedia that are driven by empirical article traffic statistics.
  4. Wikipedia has three times as many articles as it did in 2007, while the number of highly active editors, the core community, has dropped by a third. There are more and more articles that do not receive proper scrutiny.
  5. The reader cannot tell the difference between an article like Barack Obama that is watched by hundreds of editors with rounded contributions histories and an article that was written by half a dozen single-purpose PR socks that have only ever made a few dozen edits in one narrow topic area. To the public, it's all "Wikipedia".

It doesn't have to be that way. There are statistics in the database – like the number of contributors that were involved in writing the bulk of the article, the average number of edits those contributors have made in other articles and topic areas, the number of article watchers, the number of edit wars and deletions of sourced material – that many PR efforts score badly in. It's an area that could do with further research. Ultimately such statistics should be analysed and reported to the reader in the form of an article "health index", perhaps as a simple colour-coded icon.

Quite apart from introducing the reader to the notion that some Wikipedia articles are healthier than others, the fact that the typical PR product will achieve a visibly and permanently poor result on the health index scale might itself discourage some types of PR efforts.

I hope the WMF will devote further resources to researching quality metrics and correlates, and devise ways of making them highly visible to the reader.


Parts of this op-ed were informed by discussions at Andrew Lih's "Wikipedia Weekly" Facebook group. Stimuli and encouragement received from contributors there are gratefully acknowledged.


Andreas Kolbe has been a Wikipedia contributor since 2006 and is a longstanding contributor to the Signpost's "In the media" section. The views expressed in this editorial are his alone and do not reflect any official opinions of this publication. Responses and critical commentary are invited in the comments section.