Fluid personal communication in the era of Facebook, Skype and Google+


This week’s announcement of Skype’s acquisition of GroupMe, and the recent introduction of Facebook’s Beluga-based Messenger, are part of something much bigger than group text messaging: The landscape of personal online communication is changing. The very communication paradigms we’re accustomed to — email, text messaging, chat and wall posts — are starting to be blurred and redesigned. In the next generation of social media interaction, users will communicate online in ways that better mirror their organic interactions in real life. Welcome to the age of fluid personal communication.

In a thoughtful recent post Om said:

“…instead of getting bogged down by the old-fashioned notion of communication – phone calls, emails, instant messages and text messages – [Google] needs to think about interactions…..To me, interactions are synchronous, are highly personal, are location-aware and allow the sharing of experiences, whether it’s photographs, video streams or simply smiley faces. Interactions are supposed to mimic the feeling of actually being there. Interactions are about enmeshing the virtual with the physical.”

I agree, but the concept of fluid communication goes deeper. Until now, personal electronic communication could be crudely divided into two types – active and passive. Email, chat and text messaging are the prototypical active forms. Facebook wall posts, tweets, and Google+ posts, are the prototypical passive ones. Until recently, these two types lived separate lives, but that’s changing, and with it are some of our basic assumptions about online communication. Here are some thoughts about the new challenges and how to deal with them.

Three issues combine to make the story interesting: groups, informational side effects, and the social contract inherent to social networks. They are all important, but the interaction among them is particularly interesting – and confusing.

Groups. If you think of the continuum between emailing a small group of people and posting to a newsgroup or mailing list, Facebook wall posts resemble the latter. But the recent introduction of Google+ Circles, and the renewed interest this has brought to the long-standing (if somewhat dormant) Facebook Lists, blurs the boundaries.  Is sharing a photo with my eight-member “Immediate Family” List on Facebook (or the corresponding Circle on Google+) much different than sending the group an email with that photo attached?  Experience suggests it’s not.

Informational side effects. Email has long recognized informational side effects — the distinction between To and Bcc is the best example of it.  When I address a message to Sally and Bill we’ve achieved common knowledge of the content:  we are all aware of the message.  When I send a message to Sally but Bcc Bill, something more complex happens: Sally and I achieve common knowledge of the content, and Bill and I achieve common knowledge of both the message content and the fact that Sally and I have common knowledge of it. There are other, more subtle informational side effects of To.  For starters, if Sally and Bill don’t know each other, then by emailing them both together I’ve made their existence common knowledge, and disclosed that I have a relationship with both. And obviously they now can communicate directly. I once had a banker who sent email to several of his clients using To, a gross privacy violation. The banker could have used Bcc in that instance, but in other cases that’s throwing away the baby with the bath water. For example, a company may need to send a message to its investors. It’s important that everyone know who all the investors are, but at the same time it may be inappropriate to reveal their email addresses.

Social contract. Careful control of who sees what is perhaps something of a corner case in email, and one can imagine various ways to deal with it; for example, in the company investor’s case, one could Bcc everyone, and includes just their names in the body of the message. But what is possibly a corner case in email becomes central in social media.

A social network is not merely a communication medium. It is first and foremost a place in which social contracts are established and maintained; and when you overlay a communication framework over such a social graph new things happen. The most noticeable complication arises when the social network requires permission to establish a social connection. For example, when I post on my wall, since Facebook adopts a version of email’s “reply-all,” my friends see each other’s posts. Is it appropriate for two of my friends who are not mutually connected to comment on each other’s comment?

So what does this all mean?

The lesson from this is that in the era of social networks we need to revisit communication conventions that previously served us well. In particular, we can’t take for granted the distinction between active and passive communication. Every developer of a new communication service should ask him/herself the following questions:

  • When a group is created, are the group members aware of it in any way? If yes, what exact information do they get and who decides it?
  • Is the communication style active, passive, or does it span the spectrum?
  • What are possible responses to a group message? Reply? Reply-all?  Reply-only-to-other-people-who-have-also-replied? Reply-only-to-people-the-sender-is-connected-to?
  • When a user communicates with a group, what information does each recipient have about the other recipients?
  • Who can initiate communication with whom? In particular, when a group receives a message, can any group member now communicate freely with any other member?
  • If I create a group and communicate with it, and the system permits the recipients to freely initiate new communications with the group, does it remain “my” group or have I now put it in the public domain?

Overlaying multicast communication on top of a social graph is tricky; you need to think about the informational side effects and to respect social contracts. This can get complicated, but the issues are real. To borrow from Einstein: Communication in social networks should be made as simple as possible, but no simpler.

Yoav Shoham is a professor of computer science at Stanford University and co-founder of Katango. These issues have been the subject of much discussion at Katango, but this should not be viewed as describing Katango’s strategy or product offering; the goal is to have a conversation among all of us attempting to improve users’ digital social experience.


Terri Griffith

Thoughtful post and comments. I focused on, “The lesson from this is that in the era of social networks we need to revisit communication conventions that previously served us well. In particular, we can’t take for granted the distinction between active and passive communication.”

Absolutely. No more is it enough to know how to write a good one page memo or brief email with useful subject line. Our communication landscape and group working conditions require us to be more thoughtful (stop-look-listen before jumping in) and to learn to mix across the human, technology, and people dimensions.

Andrew Filev

Good questions.
I don’t think universally right and wrong answers exist for those. There are obviously some trade-offs. Context requires action on the creator side of things, and action is often synonymous to friction, when it comes to usability, productivity and adoption. When you are building a product, you also have to balance powerful features with simplicity and productivity. At Wrike, we build “work graph” that connects and organizes the work, and spans organizational boundaries. So the questions you raise are often at our radar. One of the things that seem to be simple, powerful and common in a business world is a distinction between employees of your company, and everybody else. Typically, social contracts for those two groups differ, up to the point where your employees might all be under NDA agreements, giving you some protection.
There’s another interesting and related issue. The sheer number of communication options and the overlap in them brought unification front and center. But unification also brings the issue of duplication. If I see something in the Activity Stream, do I have to also see it in email? If it’s an important thing, I do need to see it in both places, but how do I track “reads”? It looks minor when we speak about two channels and one message. But if there are 5 channels and thousands of messages, the problem becomes more important.
The whole notion of “read” is interesting in itself. The typical social contract attached to email is that if you are my co-worker (or friend) and you send me an email, I’ll read it. This contract is intertwined with the software. You know and I know that my email client will show me unread messages and highlight them. But if you post an update into your Facebook stream, neither you, nor I assume that I’ll read it. Now, if you want to unify these two channels without any additional context (like @mentions), you will break people’s expectation on the quality of content, its urgency and its signal-to-noise ratio.

Premium Anon

First ever highlight of these things I have seen, thanks! To me it is plain obvious that facebook/linkedin/twitter all screwed up here, in many many ways, the biggest is that they include themselves, as an invisible idiot, in all groups by claiming they OWN everything anyone says. They are all doomed walled fiefdoms and people will not put up with this or get enslaved.

Rooly Eliezerov

When it comes to new technologies, an important aspect is users’ migration dynamic.

Yoav Shoham

@ronald nailed it; context is key, and the more of it we can capture online, the more fluid the communication.

While I’m at it, apology to my logic / cs / game theory friends; in the editing process the concept of common knowledge got mangled a bit. But little harm done; you don’t read these sort of things anyway…


ever tried to build a context driven language. I’m working on a context run time, seemed an easier starting point… Seemed.
Don’t know if you keep up with neurology there are some interesting developments lately which could be very interesting, last 2 years.
BTW, I don’t think we have to build on neuron level(there is no magic), we just have to get the abstracts/concepts right [math] then we can build on any level of abstraction. Our tools are a major limiting point to translate/implement/test what we know, hence my question.


What about Information/context loss. In real live we have a lot of context[including body language, eye contact …] which we use to adapt. With email I can send one email to a persons work email addr, the same email send to the same person private email addr. can get a different response. Example, persons sees how he/she is addressed and can respond accordingly.
Since circles don’t convey the context I use them in, they are build around information loss. Which results in what you call site effects, but site effects are mostly all the time based on missing context which results in information loss since they are related.
Meaning requires context, see your list. Build context aware application. And no OO categories/hierarchies are not context[past, present,prediction]. Means we have to rethink a few things besides social communication.

Comments are closed.