Google+ vs. Facebook vs. Twitter vs. ... Part 2: The Danger of a One-Way Fast Lane

[#1]

Michael Wu, Ph.D. is Lithium's Principal Scientist of Analytics, digging into the complex dynamics of social interaction and group behavior in online communities and social networks.

Michael was voted a 2010 Influential Leader by CRM Magazine for his work on predictive social analytics and its application to Social CRM.He's a regular blogger on the Lithosphere's Building Community blog and previously wrote in the Analytic Science blog. You can follow him on Twitter or Google+.

---

Earlier this week, I talked about some of the unique network properties of Facebook. Today, let’s see what we can learn from Twitter in terms of its network properties and behavioral implications.

Twitter

Unlike Facebook and Linkedin, Twitter only requires unidirectional consent to connect. As a result, it has a greater growth potential. Despite all the limitation (e.g. 140-character messages) of Twitter, it grew with an astounding rate. It spreads so feverishly that it created a communication network that has lower degrees of separation than the social networks in our physical world. The average path length between any random pair of Facebook users is about 5.73, which is on par with the six degrees of separation in real life social networks. But the average path length between random pairs of tweeters is only 4.12. This means networks that require unidirectional consent could lead to a smaller world, where people are closer together (i.e. shorter average path length). Consequently information spreads faster on twitter. In other words, the Twitter platform is more viral!

That being said, I must say that the growth in membership is a bit overrated. It is only one metric for measuring the success of a social platform, and unfortunately, it is a rather poor metric for measuring success. Mainly because the growth in membership doesn’t imply those members will continue to use the platform. In fact, if you search around, you will find multiple sources demonstrating the following: despite the stellar growth of total membership on Twitter, the actual number of active members is not very impressive at all. Not only are a significant fraction of Twitter members inactive, many “active members” may not even be human (i.e. they are bots). The continual usage of the platform depends on the relative utility it brings to its users compare to its competition in the market.

So what utility does Twitter bring? IMHO, Twitter is merely a giant instant messenger with 140-character limit (per message) that is open to the public. The unique value it brings is the simplicity of the platform, which requires almost no effort to adopt and use. Don’t underestimate the power of simplicity. From the gamification perspective, simplicity (or ability) is one of the three crucial factors that drive human behaviors, which includes adoption of new platforms and loyalty to existing ones.

The Price for Rapid Growth

However, Twitter has its limitations too. The biggest problem created by a unidirectionally-connected network is that it makes the information content less relevant. Consequently the twitter stream is often flooded with noisy tweets that have low signal to noise ratio. Twitter has subsequently implemented lists to help user curate the tweets they received. By viewing the tweet stream within a list, users can filter out irrelevant contents and focus on a subset of their following that interests them at the moment.

I’d like to emphasize that lists is a receiver curation mechanism. That means it relies on the information receiver to group the people they follow into the appropriate lists in order to filter out the irrelevant noise. However, lists prove to be insufficient, as new Twitter curation tools continues to be developed (e.g. paper.li, storify, etc). Either people are not putting their following into the proper lists, or receiver curation is not strong enough as a mechanism to surface the relevant signals.

Another serious drawback of unidirectionally-connected networks is that the connections on these networks are much weaker. The high growth potential is an advantage of unidirectional network, since it allows users to build their network quickly without the consent-to-connect step. However, the side effect is that people can also re-build their network easily. So users would have an easier time switching from a unidirectional network and start all over again elsewhere. This means the Twitter network is rather brittle, and consequently it’s less cohesive and sticky. This probably contributed to the low ratio of active-to-total user base on Twitter, as well as the large number of spam bots on the platform.

Conclusion

So what lessons did we learn from the Twitter network today? The rapid growth that is enabled with a unidirectionally-connected network has a price. Unidirectionality made Twitter more viral at the expense of:

1. Weaker relationships: the network is more brittle, less cohesive and less sticky
2. Noisy content: low signal-to-noise ratio (SNR). Although Twitter created lists to deal with the noise, it is a one-sided curation mechanism (i.e. receiver curation) that may not be strong enough to filter out the noise

Lastly, we must not overlook the greatest strength of Twitter. That is, the simplicity of the platform.

Next week, we will take a deep dive into the network properties governing Google+. We will understand why Google made the choice it did, and why they chose to enforce the real name policy initially. There are reasons for all these as there are reasons for its astounding growth and the observed noise level on the platform. Stay tuned to find out why! In the mean time, let me know what you think.

[#2]

Hi Mike!

Great insights! 

1. What about median lists volume?

2. Have you analyzed activity by using Tw lists?

3. Have you some data about rebuilding activity - very interesting "hypothesis"?

[#3]

Hello Andrei,

Thank you for the comment and question. I apologize for the slow response. I've been very busy in and out of town.

Even though I have not analyze much data on list usage, I know people are doing that just from colleagues and friends. It is not too hard to abandon a twitter account and start fresh. I do not have population data on this, but it is just from observations. So I do not want to over generalize. However, the mechaniscs and dynamics I described in this articles does support these observation.

Thank you for asking the question. See you next time.