I'm back from VON this week and getting ready to go to CTIA next week and then O'Reilly's Web2.0 conference.  I had a couple of interesting experiences last week so I thought that I'd share them here.

I was given speaker credentials for VON with the assignment of going to the Unconference and "mix it up".  I hadn't been to an unconference before so I tried to get ready for anything.  The Unconference began by soliciting agenda ideas and getting a preferred list together.  It ended up that each presenter was given 15 to 20 minutes to present and most did it stand-up style without PowerPoint. 

I volunteered to present on emerging video applications, a presentation that I'd been doing various forms of for about a year.  Those plans went awry during the presentation just before mine when Ken Camp and Sheryl Breuker gave us a real-life example of their Social Media topic as Ken proposed on live Qik-stream.  You can read about it from Ken here http://kencamp.vox.com/library/post/were-engaged.html.  Congratulations to both of them!  Anyway, after the room came back to order, we took an opportunity to review the remaining time and topics to see if we needed to change anything.  I had really come to VON to talk about Comms impact on Social Networking, so with the Social Media talk being truncated I proposed that as a replacement for my video presentation.

So how do you follow a marriage proposal and still get anyone to pay attention?  Exactly!  I went right into a monkey story.  A few weeks ago I did a blog entry onto http://pulcom.ning.com/ that encompasses most of the story.  Rather than search around for the Qik-cast of my presentation, I'll just repost that blog entry here with a few edits to reflect new thought and this blog's context.  And here it is...



This was originally conceived as an article for one of the industry magazines, but now that it's written down, I see that it might be a bit further "out there" than other things I've written and presented (maybe excepting this one).  So I'll post it here where I think it has a better chance to be seen as a welcome relief during an all-night coding session

The trigger that started me writing this note was an article describing how chimpanzees used role models to define their troop culture. Things like what noises to use & what tool to use for doing a certain task.  It even emulates mankind to the extent that chimps will goof on a transplanted chimp that does things differently (can you say "salad fork"?).  Chimps are always funny, so after sending the link to a coworker who understands neither the funniness of chimps nor the hilarity of the Three Stooges, I sat back and thought about how this new data meshes with some study I'd been doing on Social Networking.

I'm not a psychologist, but I probably have several relevant pathologies. To really get past the Web 2.0 implementations of social networking aids (e.g. blogs, recommendation and reputation systems, network distance calculations), and to understand what should work and what is doomed to future ridicule, you need to read some stuff about the psychological research into group relationships and the communication modes that are key components of maintaining those relationships.

One of the coolest concepts in Social Networking is Dunbar's Rule of 150. How many people can you track to a level sufficient to know if they're free-riders, or to understand their current emotional state? The answer is about 150. That's it, 150, and no more. It's hardwired into our brains like an FPGA that's run out of gates. Dunbar described a relationship between this number and the size of the brain's neocortex relative to the total size of the brain that exists within primate species. So the chimp version of Fonzie will surely max out his network data array, but at a lower number than I will. That is, if I had anywhere near that many close friends.

BTW, the concept of Dunbar’s Limit has totally freaked out my son. He's also been researching Social Network analysis for a new startup he's involved in (and I'm advising for) and, like parental rules when he was young, is having trouble accepting this limitation.

Anyways... To gather enough data to understand someone's current emotional state, or if they're a free-rider polluting the commons, you have to communicate with other community members. In lower primates, one of the primary communications activities is grooming. But then a problem presents itself -- how many lower primates can you groom in a day and still hunt down some food? I'll challenge Martyn Davies on this question sometime in the near future, but anyway it's got to be a limited number. So if I can improve the efficiency and information density of the communications, what happens to Dunbar’s Limit?

Here's where an interesting coincidence happened. Around the time that language started evolving, our ancestors' neocortex started growing. Was it cause and effect? I haven't found anyone willing to go out onto a limb (even for a ripe banana) and say so yet, so let's just leave it as a coincidence. So now that I can communicate the same information much faster, I can fill up the larger data array that my humongous neocortex has created.

So what do we use those new skills for? Mostly for gossip and cell phone calls when amongst crowds of people, it seems. Actually, gossip is one of those communications that transmits valuable social information. How else do you find out who your society's free-riders are? I almost wrote "bad actors" as a synonym for "free-riders", but decided the pun was just too much.

Fast-forward to modern times when our "work tribe" and "neighborhood clan" are both likely to be larger than 150, and we're presented with a predicament as to how we break through Dunbar’s Limit to improve the efficiency and productivity of those Social Networks. We need to make those 150 memory locations part of a larger "virtual" social network memory system. That's where Unified Communications (finally) comes in.

I focus on three key aspects of a good Unified Communications product when I think about productivity enhancement: Presence, unified message stores and find me/follow me.

Unified message stores and find me/ follow me are closely related with one being real-time and the other being less real-time. You can sum both of them up into a simple statement: When someone wants to talk to me, I want them to reach me using whatever method our communications avatars negotiate (IM, VoIP, SMS, smoke signals) on whatever device I happen to be near at the time. These two features have a direct relationship with the efficiency of communications.

Presence communicates my current status and is somewhat akin to the communication of the emotional state that I mentioned above in the chimp paragraph. Presence also functions as an enforcer of communications etiquette as modified by my relationship to the person who wants to communicate with me.

How did etiquette work its way into this discussion? There is a subset of etiquette beyond “which fork goes where” that deals with improving the signal to noise ratio of the content we communicate. In traditional etiquette, rules such as “don’t talk with your mouth full”, and specific timeframes for writing thank-you notes make the content more understandable and time-relevant. Emerging etiquette examples include whether or not it’s polite to IM when you’re with a group of friends, and whether it’s appropriate to make a cell phone call to your best friend while traveling in a crowded train.

Presence should keep away people I don’t want to communicate with, and enable those with whom I do want to communicate to do so quickly and in the most content rich method that is feasible.

Unfortunately, that’s as far as most UC products take things today. What else could UC systems do that would increase work and personal productivity and also help us “virtualize” Dunbar’s Limit? I have some ideas:

• Map our social networks based on actual communications to enable network analysis and optimization (see www.eTelemetry.com)
• Enable role and skill based routing (note that some UC systems incorporate business rules for a limited version of this)
• Identify in real-time who’s solving a similar problem or doing similar work to minimize duplication
• Identify additional relationship oriented Presence models
• Automate the setting of real-time Presence (I call this Presence inference) so that more Presence detail can be used without additional configuration time
• Presence alerts. The Presence server should proactively alert me when three hard-to-schedule colleagues become available. A good Presence server would also initiate a conference or collaboration session when the condition occurs.
• Incorporate rating, reputation and recommendation systems to create virtual relationship data

That last idea requires some further explanation. In order to virtualize Dunbar’s Limit, I need to be able to immediately turn-on/swap-in a close relationship with someone I’ve never met before. I need to know whether or not they will waste my time and if they have information relevant to the issues I’m working on. Just like Amazon.com can recommend a book or a DVD, the UC system of the future will recommend that you take a video call from someone.

Setting up a reputation system is tricky when it comes to people. A book’s feelings aren’t hurt when you give it a bad review. Dehumanization through the creation and application of reputation data is scary. So how do we create a global reputation system without causing mass depression? One example solution that was recently published was used for detecting
SPam over Internet Telephony (SPIT). This paper by Vijay A. Balasubramaniyan, Mustaque Ahamad and Haesun Park of the Georgia Institute of Technology measures call duration and uses that data to calculate a reputation value. They were able to reach detection errors of 10% false negatives and 3% false positives. Sign me up!

I’ll forecast now that automated methods using complicated calculations are likely to be used in the future for calculating your worth as a human being. Does that make you uncomfortable?  How is that different from a credit score?  Is it the use of a complex, hidden algorithm or the authority of the agencies that has allowed us to accept the credit score?  Or have we just given up?  It's obvious that if I knew that I had a good reputation score I'd be more willing to opt in to such a system.  It's also obvious that if a reputation was hidden into back-office business rules that no objections would be raised. 

Here's another tie that Comms needs to make into business methods.  Comms is, in some ways, just another method to be linked to business logic, and the good UC packages will allow you to do that. 

I've heard that some call centers are adding recommendation scores to callers.  Note that we're not doing this at Dialogic at the time of this posting, nor do we have current plans to do so.  How can having a good reputation value with a call center help you?  One complaint I've had, and I'm sure you have run into it too, is when I call in to a technical support center and have to go through questions like, "Is it plugged in?" as read by someone who would be goofed on by any self-respecting chimp society.  If having a good reputation would get me past that person and to someone who could really help very quickly, I'd sign up for it.  If it ends up that I'm not as good as I think I am and my reputation value is low, then I hope that the call center would also tune itself to handle that interaction better and not tell me the many ways that they've found me lacking.


I'd like to hear other ideas on using reputation in comms apps.  Please send them in as responses to this blog entry.

The user interaction mode of the week is rich media Geo-Tagging/Geo-Cacheing.  I have identified and evangelized several types of services that could be derived from this basic interaction mode and it looks like a rich area for making money.  The only problem is that I haven't found a relevant monkey story yet.