“Can I trust Wikipedia?”

posted by eric

I hear many people today ask whether they can “trust Wikipedia”. Often they contrast Wikipedia with encyclopedias written by professionals, often they complain about the quality of certain articles, and often I have the feeling they simply don’t get it.

Thinking about this, I’ve come to realize that there’s no simple answer to the question posed in the title of this post, and I’d like to explain why. First, however, I’ll need to talk briefly about trust.

Trust is something perceived. There is no individual or institution that intrinsically “has trust”. Someone can be trustworthy, a source can be credible, a system can be predictable–but it’s up to you to decide whether to trust the person, source, or system in question. Another important property of trust is that it takes time to build–and it can be lost in a split second.

A look at society reveals a vast amount of overlaying and intersecting institutions–many of which are created to deal with issues of trust. Broadly spoken, these institutions serve to abstract the mechanisms for building trust away from the basic “face-to-face” level to a level that works on a larger scale. By doing so, these institutions bring stability in a specific context–whether it’s a market, a currency, or a physical space.

In game theory, the prisoner’s dilemma illustrates what happens in a situation where there’s no way for different parties to establish a trust relationship prior to a transaction. Since there’s too great a risk in trusting, each party chooses not to trust, which leads to a total payoff that is lower than it would have been if both parties had trusted each other. This problem is generally referred to as a social dilemma.

Now, as Howard Rheingold argues in a recent talk on the future of collaboration (don’t mind the techno, it’s an interesting talk!), a major challenge for institutions of the kind mentioned above is to turn the prisoner’s dilemma game into a something called an assurance game (also known as a stag hunt). The key is to develop mechanisms for establishing trust relations between actors in a system, so that they can start doing transactions based on trust, which in turn increases the total size of the pie. Institutions, seen from this perspective, present solutions to social dilemmas.

So, how is this relevant to the question of whether one can trust Wikipedia?

Because Wikipedia–just as its older counterparts–is an institution. But Wikipedia presents a new solution to a specific social dilemma on an unprecedentedly large scale.

One could argue that Wikipedia is still an encyclopedia. But its organizational principles are so different from those of its traditional counterparts that we have to adopt a whole new mindset in order to even begin developing trust in it. Consider Chris Anderson’s answer to whether one can trust Wikipedia:

“The answer is not a simple yes or no, because it is the nature of user-created content to be as messy and uncertain at the microscale, which is the level at which we usually expercience it, as it is amazingly successful at the big-picture macroscale. It just has to be understood for what it is.” [From the The Long Tail]

It gets more complicated still. The trust I’ve got in the institution Encyclopedia Britannica is not identical to the trust I’ve developed in the institution Wikipedia with its NPOV, its Benevolent Dictator, and its Long Tail of diff-checking contributors. Wikipedia is different, it’s a new game, it constitutes a novel paradigm. It will take time for the institution Wikipedia to build trust with its users–and comparing that trust with the trust people have in other encyclopedias might in the end make little sense!

And so it looks as if at this point, the only reasonable answer I can give to the question of whether you can trust Wikipedia or not, is a very definite “Probably”.

The Long Tail of Trust

posted by eric

I just finished Chris Anderson’s The Long Tail. Some thoughts: Links on the web are essentially a way to redirect attention in a very fluid way, hence the web with its multiple social layers is an enabling infrastructure for an emerging, full-fledged attention economy.
That much is clear, but what drives the clicking? What exactly shapes peoples clickstreams on the micro level? I think a significant driver is trust in various forms. Think about it: Almost each and every little cognitive process that forgoes a click is actually about evaluating trustworthiness, solving a mini trust dilemma if you will. Some links need almost no evaluation at all, whereas others need a lot more.

At any given time we are positioned somewhere on a long tail, surfing up- or downwards. Our attention wanders along the myriad of mini tails of long tails. In our excursion, we put our trust in algorithms, people and institutions.

It seems that trust, in various shapes, plays a part at virtually all levels of long tail markets. The key to making these markets work is to make them accessible to consumers. At least 99 per cent of the music in the product tail of a music store are not going to be interesting to a particular music consumer, and hence needs to be filtered out in some way. One could argue that the most effective filters of this kind today are trust driven. This is especially true when there is no significant head to drive demand down the tail.

What do I mean by that? I’ll let Chris Anderson explain. He contrasts the failure of MP3.com with the success of iTunes Music Store in the chapter entitled “The Short Head”. A key factor in the success of ITMS, he argues, is that the store’s long tail of music has a solid head (all the major labels) to serve as a sort of foundation for the tail. ITMS thus provides consumers with familiar points of entry in the form of artists and labels they know.

From a “trustoconomy” perspective however, myspace is far more interesting than ITMS, since it solves the problem that brought down MP3.com. It enables a long tail market with a much less significant head to function through a trust-driven social network. Users travel the tail in a serendipitous wander guided by trusted peers and niche authorities, rather than by mainstream musical “heads”. Last.fm works in a similar way, although with less focus on people and more on algorithms.

So does this mean that the future online music store should be a social network? I think it does–and the key is leveraging The Long Tail of Trust.