Live

Watch CBSN Live

Twitter vs. WikiLeaks: Algorithms Are Uncontrolled Corporate Spokespeople

Wikileaks has yet another victim. This time, it isn't the US government or the next expected target of a big leak, Bank of America, but Twitter. A number of sites like BoingBoing noticed that Wikileaks didn't appear on Twitter's trending topics. Twitter denied censorship and offered some explanations of its algorithmic approach to choosing trending topics.

But that was the problem. In a time of computers and automation, increasingly, software makes important decisions. Granted, programs use processes set by engineers, programmers, and scientists, but top management is generally uninvolved, even though applications may effectively direct the business in one way or another. Algorithms have become corporate factotums, trusted to work from approved guidelines at eye-crossing speed. But their complexity and intricacy of operations can have strategically important results that management cannot hope to follow, let alone control, and yet for which they are ultimately responsible.

Algorithms beget investigation

Just last month, you could see the impact an algorithm can have on a company, as the EU investigated Google (GOOG) for allegedly anti-competitive practices. Minor players in the search market complained that Google had demoted their sites in its search results. Google argued that search results were the natural result of the company's algorithms:
"We built Google for users, not websites, and the nature of ranking is that some websites will be unhappy with where they rank," [a Google spokesperson] said.

"Those sites have complained and even sued us over the years, but in all cases there were compelling reasons why their sites were ranked poorly by our algorithms."

Google was in hot water not for a decision by management or public statement, but because researchers and developers had created a system that they thought worked best. Even though founders Sergey Brin and Larry Page created the first versions of Google's algorithms, they are long past the time of active involvement in writing code.

Where did Wikileaks go?

Twitter's current predicament of many users being unhappy with the algorithmic processing of data and the accompanying results is similar. BoingBoing picked up the issue because of a post on the Safety First blog. The writer noticed that the hash tag (a convention used to associate Twitter users' messages with a specific topic) #Wikileaks didn't appear on Twitter's trending topics. The writer compared Twitter traffic for #Wikileaks with a number of top trending topics at the time and created the following graph:


Clearly Wikileaks was far more popular a topic, solely by traffic, than any of the official trends. The blogger assumed that Twitter was deliberately suppressing the news. As discussion spread through the blogosphere and other media, Twitter emailed the Washington Post with an explanation:

Twitter is not censoring #wikileaks, #cablegate or other related terms from the Trends list of trending topics.

Our Trends list is designed to help people discover the 'most breaking' breaking news from across the world, in real-time. The list is generated by an algorithm that identifies topics that are being talked about more right now than they were previously.

There's a number of factors that may come into play when seemingly popular terms don't make the Trends list. Sometimes topics that are popular don't break into the Trends list because the current velocity of conversation (volume of Tweets at a given moment) isn't greater than in previous hours and days. Sometimes topics that are genuinely popular simply aren't widespread enough to make the list of top Trends. And, on occasion, topics just aren't as popular as people believe.

According to Twitter, the trigger is a combination that includes the following:
  • The number of messages using the appropriate hash tag increases sufficiently over previous times.
  • It's not enough for the percentage of total tweets to increase. The topic must experience an increase in absolute volume, as well.
  • A topic must be "widespread enough," presumably meaning that messages relegated to too small an area would be considered as unrepresentative of conversations in general.

Show us the algorithm

People started second guessing, taking available measurements and numbers and trying to apply them. Someone who identified himself as Josh Elman, a former Facebook platform manager now working at Twitter, tried to answer concerns, but ultimately had to observe that he couldn't "go into the full algorithm."

Not only would it be impractical and exceedingly difficult, but it would also expose an important piece of corporate intellectual property.

However, the answer is unsatisfactory to people basing their views on a a combination of information and personal assumption (in this case, that Wikileaks should have been a trending term on Twitter). It sounds like a combination of general statements and a request to be trusted. "Why can't you just show us?" you can almost hear them say.

Do as I program

And there's the problem. The algorithms themselves embody a set of decisions basic to the business in which management may never have participated. The company can't reveal the details, and yet because Twitter has become part of the media landscape, consumers feel entitled to a thorough explanation.

But where does the growing umbrella of media end? At how a company like Yelp chooses to highlight some comments about small businesses while effectively suppressing others, because not everything can be t the top of the list? How Google's algorithms could allow an allegedly abusive web business thrive, even as former customers piled up negative comments about it? Google would up changing its algorithms to prevent such a scheme, but not before some asked if its search engine was responsible.

Algorithms have become brutally efficient employees that increasingly are the public face of corporations that constantly speak without managerial or even legal review. There should be no surprise that problems have begun to creep up, only that they aren't yet more prevelant.

Related:

Image: Flickr user juhansonin, CC 2.0.