Monday, December 14, 2009

The Next Big Thing - Network Intelligence

Last week I read and listened to two stories that when combined painted an interesting vision of the future. One story shared the results of a contest sponsored by DARPA in which participating teams were to find 10 red weather balloons released at various locations across the United States. The second story was that of Google launching personalized search.

The winning team in the DARPA contest was a group from MIT who used social networking and a multi-tiered affiliate rewards structure to provide a financial incentive that aligned the interests of the team in pursuit of a common goal. In listening to the story on NPR's Science Friday, I recognized the power of leveraging a social network to refine information and create consensus around points of knowledge. I also thought of why multi-level marketing (MLM) compensation plans are so effective at building a channel. Participants were rewarded for recruiting members who were ultimately responsible for providing information that led to the successful location of the 10 weather balloons, so people were rewarded for quality and results rather than on participation alone.

After stripping away the financial incentive, the large social network created to find the 10 weather balloons was a use case of something that has been happening for years on a smaller scale online. Message boards and social encyclopedias are built and ultimately succeed as communities of interest gather around specific topics. The difference here was that the network was not a casual "ad hoc" network created to by a group of individuals trying to create the best encyclopedic entry on a topic of common interest or to share specific knowledge through altruism. The network MIT created to win the DARPA challenge was designed to achieve a specific goal, in part by providing a proper incentive.

The end result was what I am calling Network Intelligence, the aggregate intelligence of everyone in a network that surpasses the intelligence of any individual for the defined purpose of the network.

This is not so different from institutional knowledge; however, the concept of institutional knowledge tends to be historical rather than actionable. You could think of Network Intelligence as that portion of institutional knowledge that is transactional or relevant to a specific purpose that is being acted upon in the present.

Network Intelligence is also what will be taking Google's search to the next level once personalized search becomes "networked."

I was consulting with a friend on Thursday, and he was excited to show me how many pages on his sites had reached the #1 position for various keywords. I then remembered reading about Google's personalized search a week before. We went through his keyword searches on Google using our own computers and found we both received different Google result pages for the same searches. Prior to our meeting, my friend had spent most of his time reviewing his own sites and tracking internal changes over time. I hadn't spent much time visiting his sites in the past. Most of my searching using his industry keywords had been to conduct market research and sign up for offers from other companies.

Where he was seeing his sites and pages in the #1 position for very competitive keywords, I was seeing the same pages in the #3, #4, or #5 position. So our respective search results were indeed personalized based on what we had been searching and clicking on previously within the results page.

(I'm not going to go into all of the SEO and relevancy implications here, but there are now questions to be answered related to how new sites and pages are going to compete under this new methodology . . . at least until network intelligence is applied.)

BUT, what would happen if I had the option to have my friends' preferences and search habits influence mine? If, for instance, I had wanted to follow the same biases of my friend, who is an expert in his industry, could I have seen the results presented in the same order as his appeared?

In order to do this, Google, Facebook, or another aggregator of personalized information would need to create an algorithm to have the search results and preferences of friends influence the results of each other.

Another feature I want to have is to follow the search preferences of specific individuals who have opted in to allow their preferences to be followed. I could then (with the permission of my friends) see which search results were most relevant to them in specific areas. There are huge privacy implications here, but I think the service would work incredibly well as long as specific search terms for which my friends were searching weren't shared . . . or maybe it would be even better if they were.

I would love to know what my friends were searching for online. Oh, the implications on privacy and individual behavior! How would this impact a friend who becomes embarrassed because he is searching for nefarious content? How could this be leveraged for competitive research and corporate espionage? Allowing transparency and the sharing of search habits might change the behavior of those whose public values and private behavior is inconsistent because of positive social influence. Is that so bad? I think that could create one of the biggest safeguards against the viewing of illegal content and pornography. Would people be viewing it if they had opted into a service that let others know exactly what they had been searching for?

But I digress. The first implementation would simply be applying an algorithm to my search results based on the aggregation of my social network. I would refer to this as "social intelligence," but that concept has already been used by Daniel Goleman and others, who have spent the last 15 years defining different modes of intelligence. Network intelligence has already been used by computer scientists, such as Drew Major and others, to explain where "intelligence" resides in a network to manage request routing, caching, and other network concepts. But I think the concept of harnessing the intelligence of a social network (whether formal or informal) deserves the title of Network Intelligence.

The real-time internet, such as watching trending items on Twitter, provides some network intelligence, but the network is too broad to be useful for a specific purpose beyond identifying topics of interest. The REAL value of Network Intelligence is in the network itself--surrounding yourself with the most intelligent and thoughtful people on a broad array of topics to provide you with the most intelligent and insightful information on any subject of interest.

Now, here's the interesting point: How will network intelligence influence the evolution of social networks over time? What will happen to the relevancy of pluralism, multiculturalism, and the value of differing opinions and diverging theories? Won't someone just surround him or herself with like-minded people? Not if the intelligence isn't helpful over time.

Those who provide "bad intelligence" will over time become less relevant as the wisdom of the masses is refined through collisions with new thoughts and ultimately hardened through agreement. Network Intelligence is not static but a concept full of movement and change through "living" knowledge and wisdom. The network accommodates and assimilates new information that it finds useful and "loses" less meaningful information over time.

In the end, I believe the concept of network intelligence will drive us to the pinnacle of human intelligence that can be found through the reasoning of humankind. While I believe there is a higher source of knowledge that a human social network, I don't believe there is anything more powerful that can be created through the integration of human behavior and technology.

No comments:

Post a Comment

Thanks for commenting on my blog posting. I will try to respond in a timely manner to any questions or comments that keep the conversation moving forward.

Feel free to contact me at to correspond directly.

Warm Regards,