Web 2.0 Blog – Discovering Innovation Opportunities using Social Media

Posts Tagged ‘Networked Audience

In my last post I started to examine the claim of the cluetrain manifesto that a more networked audience is more intelligent or at least a better detector than an individual. The #Mumbai victim list twitter distribution illustrated 4 ways which a network can apply truth filters and 2 ways in which the network affects might work against detecting falsehoods over the short term.

One recent tweet from Deb Lavoy questioned whether crowdsourcing will always generate good ideas, because after all a mob is also a crowd. Mobs are famous for poor and emotionally driven decisions and actions rather than intelligence and innovation. So how do we prevent crowdsourcing from becoming mobsourcing? Do connections between audience members, which a mob seems to have, mean better decision making?

Mobsourcing vs Crowdsourcing
Of course this is dealt with quite well in James Surowiecki’s Wisdom of the Crowds and I dont mean to say that these observations in any way change his conclusions. I am just using this blob as a scratchpad to prove it to myself by trying to understand the actual behavior in an audience which produce the effect.

First what is our working definition of intelligence? From the cluetrain theses, it seems they meant the ability to filter out inauthentic information. So basically, a networked audience is a lie detector in a way because it will filter out inauthentic sounding information which I talked about in the last post and came up with 4 elements which work in the networks favor of filtering out false information and 2 which can work against it in the short term. From a Surowieki analysis, we found 2 problems: Can you trust the aggregation mechanisms which are very informal in twitter? Are the contributors acting independently or does the emotions compel them to go along with the mob?

Of course the problem is as in the case of a lie detector, what if the originators of the lie, believe the lie to be true? If our audience is large and well networked, we hope it can confirm the factual nature through multiple sources and if there are differing views, we hope the network would prefer the one which is most aligned the individual members’ realty or partial knowledge as Surowiecki says. But the more emotionally relevant the information (as we saw in the case of the Mumbai list), the faster it spreads, even from a single source. So it seems the more emotionally compelling the information is to act on, the less likely it is to be verified by the network effects. The audience becomes a mob and acts as an amplifier for a single source which strikes a deep chord with the influencers in the network. So it seems a way to keep the crowdsourcing from becoming mobsourcing, is to slow it down and force deliberation while maintaining independence. I guess Surowiecki would say that the emotional element pushes people toward conformity or maybe as he quotes Arthur Schlessinger’s comment on the Bay of Pigs planning, it is urged to assume consensus. I wonder if the motivation to want to conform is because the need for belonging seems to rise in the face of many emotions such as fear.

Surowiecki also mentions the need for people to express their objective independently of others as a prerequisite for effective crowdsourcing as well as having a diversity of experience. So in the moment of an attack, the emotion felt by most in the audience would tend to give them a shared overwhelming experience which I would guess would seem to lesson the amount of diversity and independence in the audience though strengthen their common focus on the problem of finding out what is happening.

Of course the task of the Mumbai list of victim names seemed to completely authentic. Also the twitter seem to correct false information by outing a college student pretending to represent the indian government. So with these tenets of crowdsourcing working against, why did it seem to make the right decision?

The question before the audience was is the list authentic, and the shared emotion, need to conform etc actually seemed to heighten the need to pass truthful information about victims as well as not to be a source of inauthentic information. It also gave the normally busy audience motivation to spend time on the subject and try to be useful. So these normally problematic influences in crowdsourcing actually heightened the communities interest in verification in the list and is probably why the impersonator of the Indian government was investigated and quickly found out. Maybe the people needed to do something and couldn’t do anything else but work to discover more information. Is that why it worked.

OK so emotionally charged crowds are good ones to make decisions? Seems to depend on the question and the emotion. I wonder what the crowd in the moment would have thought an appropriate response would be? I think the answer to that is a strong case for inserting deliberation into decisions of actions in most cases.

So it seems to avoid mobsourcing, we need to increase the time it takes to make decisions on actions but in this case and may be others, the ability to spread the information quickly doesn’t seem to do harm when it comes to crowd based lie detection.

Side Note: I guess those guys who invented the US Senate had some insight into rules of callaboration, since they purposely made what at the time was considered almost lifelong terms in order to make the Senate a more deliberative body. (Watch for a future blog post on how the American Revolution was a social media product…Saw PBS Liberty recently and can’t shake the similarity to changes being brought about by modern social media experiences to those which made the revolution successful.)

Note: I have not verified the list of the Mumbai victims referenced here nor do I take a stand on its authenticity, accuracy or morality. I think its a very interesting example of how a networked audience behaves differently than an non-networked audience.

The mention of the increasing intelligence of a networked audience is what really struck me about the 95 theses at cluetrain.com. Mass calloboration, a way to deliberately request help from an audience is talked about in more detail in James Surowiecki’s Wisdom of the Crowds and emerging intelligence from networked audiences in Howard Rheingold’s Smart Mobs. The premise in the 95 theses is that once the early free flowing social media sites got started we were permanently on the path to more networked audiences. And that the networked audiences behaved differently and according to cluetrain, more ‘intelligently’ than the former broadcast centered audiences. This is very appealing, but is it true? I decided to do a series of posts trying to understand this in depth.

So is this a networked audience more intelligent? It’s not hard to get consensus that it is definitely different. The problem with intelligence is there are so many definitions. At our last workshop I cited the google doc which listed the injured and dead from the recent Mumbai tragedy. This appeared on my twitter feed while the terrorist attack was still ongoing. While I don’t know for certain who was responsible for the list, it seemed to be created by netizens not journalists nor government officials. Before we go on I want to acknowledge the 2 big objections usually rise to this sort of unofficial information:
1. Is it accurate? Probably not 100%.
2. Is this how such sensitive information should be disseminated? I am not sure there is a good answer to that. It probably is different for each family member and the entire answer might turn on whether it is accurate.

My point in talking about the list is that it is unarguably a different type of audience behavior during a terrorist attack than we are accustomed to seeing. As to intelligence, the information which was sent sped through the network very quickly and was believed. So to the extent the networked vetted it for authenticity which is the intelligence process we seek out from the official news media, this could be perceived as a type of intelligence.

The list was disseminated around the world within hours, then was apparently reported on by one of the biggest news agencies on the planet as a source of highly relevant information. (I have heard a report that this list and twitter was referenced by the BBC during the early reporting (unconfirmed). ) I would argue it was the most personally relevant information during the attack which is why is flew through twitter.

So regardless of whether the list should be posted, I think no one can question, that the posting and dissemination of a victim list during a terrorist attack is a significantly different behavior made possible and largely due to the fact that the audience is networked.

I trusted the Mumbai list to be authentic and mostly accurate even though I only received it from a single source and only heard mention of the BBC reference only much later. And while I had only heard about it from one source, I had reason to believe it was being tweeted by many twitterers.

One reason for the trust that I gave the list is that the community through which it flowed is preestablished and highly active. I received the link from Shashi (Social Media Swami) at Network Solutions and he is a very active twitterer. I know he must have believed and probably received it from sources which he has heard from before. So length of time can be offset by trust in the community before this event and how well the audience members value their reputation. The social technology and community itself can affect how well one values the reputation. Twitterers value a large following and usually their goal is to achieve it. They can loose that following very quickly, because it is easy to start unfollowing someone. Of course if it is highly valuable to them to disseminate the message, then it is more likely they will risk reputation.

The message also plays a role in whether it is likely to be believed. The more detailed a message is and the more ways it allows verification, the more authentic it sounds. The Mumbai list quickly reached over 200 names, so it seemed to be authentic and verifiable. Also we hope that someone who would attempt to release false information about such an emotionally charged subject, would not be a part of our community.

There was a presumably false piece of information which flowed through twitter at the same time as the victim list. Apparently a college student in Boston originated a message asking people to stop twittering about Mumbai while the attack was still going on and claimed to be part of the Indian police or government. It is likely that this person thought they would be doing good by disseminating this message. Initially probably a lot of people accepted this statement to be coming from the Indian government, but it was within hours reported on Twitter as being inauthentic.

It would hard to imagine much value in disseminating the Mumbai list and it would be a very high risk to reputations if it was false. It traveled through a network which can punish false messages quickly but it is highly emotionally relevant would also make it go through the network quickly as well. It also provided ways to verify itself eventually, though not immediately. These seem to be the elements which helped it to be accepted as true.

So elements which would made this message seem more authentic are:
1. Pre-established Community
2. Valuing of reputations.
3. Flowing through the community through multiple pathways.
4. Detailed information which could be verified later.

Elements which made it more likely that a message would go quickly and not be carefully verified
1. Timeliness on the importance of delivering the message.
2. Highly emotionally relevant.

So do these combined components make the audience more intelligent by trusting it to be a lie detector?
Of course lie detection is not the only type of intelligence we want to see in a crowdsourced or collaborative activity. We want original ideas which solve real problems or at least original ways to send information in a more relevant fashion (i.e. better marketing).

By the Surowiecki analysis, it seems this kind of event comes up short in a number of ways. Independence of contributors is definitely a problem in online social networks. Surowiecki himself questions how collborative the blogosphere is in a Ted talk.

Aggregation is also a bit tricky, its hard to see if the networked really aggregated the list or it was merely passing already aggregated information and people assumed it was organized by individuals?

Certainly there was a strong focus on a common goals of getting important information spread quickly and discovering more facts.

Well.. I still believe the list is real….Am I part of the problem?