Editorial note: This post began a year ago in October 2012, prior to my acceptance of employment with Stack Exchange, Inc. The views expressed in this post are my own, from the perspective of a Stack Overflow community moderator, and not of my employer.
Geert Hofstede is one of my personal heroes as a cultural anthropology nerd. He defined and articulated something extremely abstract, the measurable factors of a culture that influence the behavior of those within it. By identifying these factors, you gain an understanding of how they can affect cross-cultural interactions and can take steps to ensure these interactions are positive.
If you deal with communities of human beings in any capacity, you really should at least familiarize yourself with his theory of cultural dimensions. Communities are very much a business, even if no monetary profit is gained. When people come together to get something done, they’re doing business. Whatever facilitates their interactions should eliminate as much friction as possible, or at least try to avoid adding more.
With 80% of almost 4.5 million questions answered, Stack Overflow gets stuff done. Our collective mission is to ensure that the crème de la crème of programming knowledge rises to the top for all to find when they need to get something done.
Near the middle of last year, some patterns emerged in the way that people use the system that we, your moderators, could not quite understand. An increasing number of users from very specific parts of the world seemed to be participating in ‘open air’ cross voting rings in a manner that was so obvious that users without moderation tools were quick to spot it. We’d dealt with sock puppet accounts being used to artificially inflate someone’s reputation in the past, but nothing on this scale and not nearly as blatant. While Stack Overflow is primarily about sharing knowledge, many people take the game around it quite seriously. People were getting increasingly angry, we were getting a little worried. This was far from a few isolated incidents.
Enter the concept of power distance
Put simply, a culture’s power distance index is a measure of how much its members expect power to be distributed unevenly. The higher the number, the higher the expectation. When distance from authority and power is widely expected, it’s of course generally the norm. One of the more amusing reflections I’ve read on it was written by Mike, a professor of Physics at the University Of The Philippines:
One of the disadvantages of a high-power distance culture is the use of mitigation when a social inferior is talking to a social superior. Malcolm Gladwell takes the example of a pilot and a copilot. Pilots and coipilots take turns handling the plane, and one of the surprising statistics is the higher likelihood of a plane crash when the pilot is handling the plane, compared to when the copilot is steering.
This sort of mitigation isn’t entirely due to the notion that subordinates should just do as they are told, it’s the notion that one does not ordinarily take the liberty of questioning someone higher in authority or stature than themselves. When a subordinate or underling must question the judgement of those perceived to be above them, they must do it in such a way that respects, if not reiterates the stature and position of the other. As this can be a rather tenuous and exhaustive process, one wonders how often it’s abandoned or never initiated.
Piloting a plane is a bit of an extreme example. Plus, flying stuff is hard, let’s talk about programming. How long could you last if you worked in a high power distance culture where questioning your superior, or even coworkers with more seniority than you was something that was very taboo unless done infrequently with the perfect amount of finesse? How long could you do your best not to look better than those you answer to, to the point that giving others undeserved credit for your work was something that just came naturally? It’s something that people working in these places have probably dealt with on more than several occasions:
|Country||Power Distance Index|
|United Arab Emirates||80|
All of us have experienced the above scenarios to varying degrees, the phenomenon I’m speaking to is actively expecting these circumstances. When the expectations are high, the practice flourishes and the gap is much larger than you’d see in typical western office politics.
A high power distance index (or any other metric) doesn’t necessarily describe the environment that someone lives and works in, it doesn’t tell you anything conclusive about the person. It only gives you insight into circumstances that might apply to a person living in a particular environment, allowing you to make decisions in better context.
That’s what we didn’t have, context.
We were doing it wrong
We found it difficult to believe that groups of a dozen real people would be participating knowingly and willfully in the types of voting patterns we were observing. Many actions were taken, some of which were extremely difficult to reverse and despite the obviousness of it all, almost all of our actions in these cases were contested. Not just contested, but often politely contested. The first few times we were a little taken back that someone would deny seemingly blatant abuse of the voting system when confronted with overwhelming evidence, and then it finally clicked.
As it turns out, we were wrong in many of these corner cases. Many of these users were in fact individuals just doing what they thought was right by supporting their superiors for the good of the team. Several of us got the sense that individually, these people really cared about the site and understood that what they were doing was actively harmful. All of them were coming from a country that had a power distance index of 75 or higher.
We immediately changed the way that we interpret the data, giving much more weight to ways that users might be connected and influencing one another in these situations. We also created a special template for reaching out to these users that serves two purposes:
- Asks them to please stop doing what they’re doing, explaining why it’s wrong
- Gives users that would rather not be doing it something to support them, and relieve them of the burden of actively dissenting. A message from us, the ultimate authority that polices the site serves precisely that purpose.
This is becoming a rather frequent message that we send, and so far it seems to be having the desired effect.
We could be doing more things better than we are
Power distance is interesting, but probably the least interesting index that Hofstede theorized. These metrics aren’t something that you can just plug into code, these are insights that human beings can use in order to make more informed decisions and add precision to their actions. It’s something we’re learning together in an effort to make sure that culture is a much less frictional barrier than language; this is fascinating stuff.
I invite you to take a look at the JSON dump I’ve made available and the context for each index explained on Clearly Cultural. What could you do with this data in your user interface? Is it rude to assume that the majority of people from a certain region and culture need a special kind of help to use your software properly? Isn’t it the software’s fault if users can and do use features incorrectly? How would you test this?
If you design user interfaces, I’d like to hear from you.