IndustryYahoo Study Seeks Algorithmic Answer to ‘Six Degrees of Separation’

Yahoo Study Seeks Algorithmic Answer to ‘Six Degrees of Separation’

Yahoo researchers, in collaboration with members of the Facebook Data Team, are again challenging Stanley Milgram’s “six degrees of separation” theory, in an effort to answer the question as it applies in the age of online social networking.

yahoo-six-degrees-of-separationYahoo’s Small World Experiment, led by Duncan Watts and Sharad Goel, Principal and Senior Research Scientists at Yahoo (respectively), aims to conclusively answer the question of degrees of separation. The topic has been much debated and widely studied, but so far remains unresolved.

The subject of degrees of separation has been a societal fascination since psychologist Stanley Milgram first published the results of his own experiment in 1967, though it actually traces back to the 1929 short story Chains, by Hungarian author Frigyes Karinthy. “Six Degrees of Kevin Bacon” became a popular game in the ‘90s and spawned the Six Degrees social good project launched in 2007 by Bacon himself.

In a 2009 paper, Social Search in “Small-World Experiments,” Watts, Goel, and fellow researcher Roby Muhamad examined a common problem in small-world experiments: a high attrition rate, meaning so many chains fail to complete that estimates of true chain length become biased. They also highlight the differences between topological and algorithmic small-world hypotheses.

Algorithmic vs. Topological Degrees of Separation

The difference between topological and algorithmic distances between people is an important distinction when attempting to evaluate just how closely connected we have become. Search Engine Watch spoke with Watts to learn more about their study and how it differs from others completed recently.

“Where others have determined the number of topological degrees of separation, we are working on the algorithmic problem. We are now able to look at this on a bigger network than anyone has ever looked at before,” he said.

Topological small world studies take a look at the network as a whole and attempt to determine the smallest possible number of intermediate connections between any two people in the network. The recent Facebook/University of Milan small world study is an example of a topological study.

Researchers calculated the average distance between any two people by computing sample paths between two users, at a great scale: 721 million Facebook users, or over 1/10th of the world’s population. With access to the entire network, they found that the average number of acquaintances separating two people was 4.74. This gives the impression that the world has become smaller in this sense; we could say there are now 4.74 degrees of separation, whereas Milgram’s work famously pointed to “6 degrees of separation.”

Yahoo’s latest study examines the algorithmic answer to this problem. As a part of the network, you need to find that path by relying on your own social search skills, and those of the intermediate contacts. This more closely mirrors the real-life state of social networking. The 4.74 degrees of separation connection is the bare minimum between us, but would be a highly unlikely path to actually complete given that we, as the average user, don’t have access to the entire network. We can’t tell which 4.74 connections would lead us to our target person.

Instead, in a real-life situation, we would rely on our connections to choose a person among their connections who might lead us closer to that final person.

Yahoo & Facebook Working Together to Solve Algorithmic Small World Question

yahoo-small-world

An important piece of the study is the ability to know and understand the network as a whole. This allows researchers to compare the actual length of completed chains with the known topological shortest paths, evaluating social search difficulty and accounting for chain attrition. In plain language, what is the number of true degrees of separation when the experiment is performed inside a network, with participants left to their own devices to find their way from one connection to another?

We asked Watts how the study, launched in August, is progressing.

“The biggest problem people have had is that most of the chains don’t get to the targets, because people just don’t pass on the message. If just one person does that, that whole chain is dead,” he explained.

In Milgram’s earlier study, said Watts, each person was 75 percent likely to pass the message one, meaning the study had a 25 percent attrition rate. Over the course of the last decade, as Watts has studied small world theory, he’s seen the attrition rate rise to its current high of about 90 percent. They had hoped that by using Facebook’s messaging system, they could actually get the attrition rate moving in the other direction.

“There have been 28,000 chains initiated, and fewer than 4,000 have gotten to the second step. Almost 25,000 of that 28,000 have completely ignored the message and haven’t even clicked at it. Those 3,700 people send messages to 470. The next step was reduced to 65, then down to just 4. Just 1 out of 28,000 chains has gotten to 6 steps. Attrition is really making it hard for this to work and we don’t really know why,” he explained.

How You Can Join the Yahoo Small World Experiment

Yahoo is actively seeking participants for their Small World Experiment. Researchers have been careful to protect the privacy of users, who are given a target person (the person they are trying to reach) and instructions to choose one person from their social network who is likely to help lead them closer to that target. Users can check back to see the progress of their experiment, or sign up to become one of the target people others will attempt to contact.

Participants can help ensure the success of the study by choosing contacts who are not only likely to help lead them closer to their target person, but who will also pass on the message. It is also possible to participate in more than one chain.

“Our main goal is to answer the research question, but in doing so, we will learn some interesting things about why people succeed and why they fail,” said Watts. “Are there certain types of people who are very good at networking, or are there circumstances in which it is more likely to succeed or fail?”

Resources

The 2023 B2B Superpowers Index

whitepaper | Analytics The 2023 B2B Superpowers Index

8m
Data Analytics in Marketing

whitepaper | Analytics Data Analytics in Marketing

10m
The Third-Party Data Deprecation Playbook

whitepaper | Digital Marketing The Third-Party Data Deprecation Playbook

1y
Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study

whitepaper | Digital Marketing Utilizing Email To Stop Fraud-eCommerce Client Fraud Case Study

1y