Saturday, May 24, 2008

Solving Second Life® User Interface and Churn Dilemmas

This is my first post from Windows Live Writer and it's an important post, so I am throwing caution to the wind.

For some time I have been thinking about the Second Life user interface and the loosely linked (perhaps?) churn problem. By thinking, I mean I haven't been in my usual "lean forward" mode, but rather in a more casual lean back, read, stare off into space and ultimately get distracted by something shiny mode. This is a good mode to consider seemingly simple but actually subtlety complex issues, and it allows my mind to wander off into open spaces and into more systems thinking. Lean back thinking is a luxury that we are rarely afforded in the current market of rush to publish micro-bursts of information and ideas. You also have to call your mind back occasionally and ask "What have you done for me lately?" and hope you aren't met with a blank stare.

So much for the frightening imagery of how I think, what's important is today my mind brought back an "aha!" (no, not that Aha).

I've been trained to go ahead and give you my answer right now, here before you go away leaving me with a tl:dr and a pat on the head. So I will, but you won't like it and some of you won't even read past my proposal except to skitter down to the comments and declare your outrage. Nonetheless, here goes:

The Second Life interface and attendant new resident churn problems can be solved using Artificial Intelligence and bots.

Yes, that's what I meant. Two of the most recently hated and feared parts of the Second Life experience can, in fact, improve not only the interface but reduce the new resident churn rate.

Despite the din of outcry across the Second Life resident blogosphere about the client, I have yet to run across a well-formed proposal for redesign. But, what my wandering mind did find was some correlation among sundry bits, explained in the following paragraphs.

At least once a week I mill about looking for the latest discussions, rants and otherwise about the Second Life interface. Honestly, I was looking for a "fix" - a miracle client release - but none were found. Instead I found Nicholaz Edition fanatics, the Linden Lab Viewer roadmap complete with the ability to skin your own (or rather, skin at your own risk), and reams of discourse some of which Digado summarized in his post about Second Life's New User Experience.

But for me, the most thought-provoking discussion arose on Dusan Writer's post about killing sacred cows. From Dusan:

….well, here’s the thing, I ended up wondering if some of the things I take for granted are errors of conception.

Call them the Sacred Cows of Second Life. And maybe it’s time for me to kill one or two of them. For myself in any case.

And the first is that the Second Life interface sucks. Which will lead to the next sacred cow, which is that the newbie orientation experience sucks. And the sacred cow isn’t necessarily that these things aren’t true, but that we should actually do something about it.

Dusan highlights his personal experience and frustrations with the complexity of the interface and ends up asking an important question. (emphasis mine)

All of which is to say that the interface is NOT user friendly, at least for someone like myself who’s unfamiliar with half the terms and look, it took me an hour to learn how to talk when I first got to SL, and it must be even harder now - do I click that little chat balloon or the other button and what’s the difference between “Communicate” and “Talk” or whatever the buttons say, all you really want to do is ask someone where the fun is.

But here’s the question - so what?


So the SL viewer seemed insanely difficult to me - so many buttons, so much to learn, and this on TOP of learning where to go, how to talk, who to talk to, and what the culture was all about, the norms, the attitudes, the list of things to see or places to go.

The answers to Dusan's inquiries change regularly, such is the dynamic nature of Second Life. This leads me to believe that the user interface will never be static, but rather will and must organically evolve with the Platform. It's important to note that this is a Platform that includes the technology, features, services as well as people, places and things to do.

Furthermore, the tasks and questions that Dusan is asking are all answerable (at least at the first order) using structured knowledge-based information. However, it's much easier to just ask someone how to do something than to have a priori knowledge that such information exists, find it, then read and understand it.

Those ideas were some of the trinkets that my wandering mind gathered up and placed into the leather bag of cognitive gathering.

Still more seemingly unrelated tidbits included the recent Linden Lab blog post regarding the removal of popular places and the implications of traffic. Many have speculated that these changes were implemented to discourage the use of compute intensive camping bots, that artificially raised the traffic results for parcels within the world. The argument is that removing the incentive of camping bots thereby removes the behavior. Note that is does not, however disallow the use of bots, it merely discourages the use of camping bots for purposes of gaming traffic measures.

Finally, two bits of video crystallized for me what I believe to be the only real solution to the evolution of a user interface against an evolving platform. The first was a video of over a year ago of Hamlet Au's conversation with Social Autopoiesis, a fully formed interactive non-player avatar (a bot). You need to see this full screen, and the YouTube version is blurry, but you can get a Windows Media version here (wait for WMP to launch). NOTE: Includes a few not safe for work words/language.

The second and most compelling video was this one from the research group at Rensselaer Polytechnic Institute in Troy, New York highlighting the work they are doing in Artificial Intelligence (AI) . The researchers are using Second Life because it's a controllable environment, which is both conducive to AI testing and development. One might assume that the Second Life Orientation Islands are a similarly controlled environment, with the advantage of a common set of goals and objectives.

Are you still with me? Are you still in lean back mode, or are you now leaning forward? Before I throw out a Q.E.D., let's review the key points:

The user interface is perceived to be complex and "broken" because the learning curve is steep for even basic tasks including finding people, places and things to do. However, the Platform (technology, services, features, people, places and things) is still evolving rapidly and a moving target is difficult for even the best Human-Computer Interaction (HCI) designers to hit. Furthermore, the diversity of residents and attendant needs in Second Life complicates the use cases for the interface development beyond that which is productive. Customization (skinning) like that which is offered by World of Warcraft interface will allow users to adapt the features that they need, assuming they escape the initial vortex of Orientation Island.

Bots have not been eliminated from the Second Life landscape, and the use of bots for fraudulent reasons is on the Linden Lab radar. Residents and researchers alike have demonstrated the most basic and fundamental uses of chat bots and AI in Second Life. The application of chat bots and simple AI for the first hour in Second Life is perfectly aligned with what new residents need. Some way to ask and be told the answers to basic questions is a logical way to present knowledge based information and possibly more important, is a way for the Linden Lab team to gather data on what tasks are most difficult to grasp, or what information is unclear, and what gets asks most often. Prokofy outlined a few of these most basic knowledge needs in "The First Hour in Second Life". These simple new resident questions are not outside the ability of Pandorabots (maybe you'd like to try your own to test me?), but real value can be derived from early AI such as that presented by Rensselaer.

We have assumed we must "fix" the interface, I am suggesting that is an exercise in futility given the above. You don't "fix" it, you teach it and evolve both the teaching and the interface using actual feedback based on experiential application. So allow me to repeat:

The Second Life interface and attendant new resident churn problems can be solved using Artificial Intelligence and bots.

These bots can be called upon at anytime to answer questions about how to do something in Second Life. They can be stationed at common areas and Help Island, and most importantly they shape the first hour of a resident's existence.

So what does the new Orientation experience the eyes of Grace look like? It goes something like this:

1) Pick your name, your starting "look" AND a serene place of your choosing: such as tropical, forest, desert, city park.

2) You rez and are greeted by your personal artificially intelligent tour guide, the lovely Graceimator.


Graceimator 1.0: "Hello, avatar_name. I am here to help you get started learning about Second Life. We are going to walk about this (serene_place) and learn a few basic tricks at learning stations. Along the way, you can ask me anything, and I will answer the best I can. Do you have any questions before we walk over to the first learning station? Talk to me by typing in the white space at the bottom of your screen, then hit Enter or press the "Say" button at the end of the white space."

Avimator 1.5: "Yes, will I ever get out of this Orientation and meet some people or am I stuck with you until I get frustrated and leave?"

Graceimator 1.0: "You will leave Orientation a happy new resident."

[Insert your personal view on amazing new resident indoctrination here]

One last note, the picture above was taken at the serene but unfinished "Organic" build in Second Life. Unfortunately I didn't see M Linden hanging out there.

Share Some Grace:

blog comments powered by Disqus