Something unexpected has happened at community@brighton, the University of Brighton's equivalent to the Landing, running a modified version of the same Elgg software.

For over 3 years, UoB was running the old version of Elgg that powered me2u at Athabasca, making the switch to the new version in the summer just as we at Athabasca have recently done to build the Landing. In the old version, there were many thriving communities, approaching 20,000 active users (out of a potential 37,000 people), and lots of social networking going on. Some of that is still there and there is still plenty of activity,  but things have changed.

The number of active users has not noticeably changed in the new version, but the kinds of uses have changed significantly. Course-related and institution-related uses of the old system were always important, but were far from dominant. Now, they appear to make up the majority, while social uses are apparently falling. The reasons for this are under investigation, but some factors are clearly contributors, including use of other social sites such as Facebook, changes in the Blackboard system that handles formal course transactions, changes in system design that make it more suited to course uses and more. However, I observe several potentially worrisome causes, each of which is probably partially responsible and each of which demonstrates an underlying pattern:

• those of us promoting academic uses through every available channel have been successful. As a result, the system is seen as another institutional channel for teaching, not a place to get together and sustain or build communities. People don't feel that they own it;
• things were lost in the migration process. Many things. Content, personalisation, and more. People no longer trust the system and so don't feel that they own it as much;
• the former site was dominated by signs of user activity: that was the main thing you saw when you visited the site, latest posts, latest photos, most active groups, and so on. That sort of thing is still there, but now over a quarter of the new space (the most significant top left part) is taken up with a very cool animated tab set telling you things about the site and how to use it. This again implies a locus of control with the owners, not the users of the site.

While there are some obvious ways we can avoid falling into exactly the same problems here at the Landing, the generic problem remains: ownership. It is too easy to make the wrong decisions for the right reasons. Small design decisions can affect such systems in big ways. For instance, should we market the Landing aggressively to academics as a teaching tool and, if we do, what can we do to limit such uses' dominance of the system? Given that it is supposed to be a bottom-up system, should we even try? 

But, of course, it is not entirely bottom-up and never can be: administrative and design decisions inevitably shape things. For example, it is the nature of the beast that we have to show people what other people are doing: that's kind-of the point of having a social site! However, the fact that people see examples of others using the system in teaching (for example) will encourage them to do the same while sending subtle messages to those wanting to do other things that they might be doing something 'wrong' or at least different. Through a perfectly reasonable design decision, we therefore cause a cascade of interactions that lead the system one way while, had we chosen to push a different use, it could just as easily have moved another way. It may not be a terrible thing: it is great to offer formal teachers a new and more learner-oriented environment in which to work. However, the domination of an ecosystem by a single species works to the detriment of the whole ecosystem in the long run, including the dominant species. This kind of system thrives on diversity of primary uses and there are many ways to support learning and learning communities apart from in the formal setting.

We can get around these problems in various ways and I've written about some of them - deliberately parcellating the environment, actively intervening, showing random novel items, intentionally pushing other uses, making it easier to move to different spaces, changing the relative weightings of novelty and popularity, introducing new tools, increasing personal control, adding delay and so on but, in the process, the risks of allowing the top-down to dominate the bottom-up are ever-present. It's a delicate dance that we don't want to lead but cannot avoid leading, even if we do nothing.

We need an evolvable system that is agile and adaptable (ideally adaptive) to changing needs and we need it to reflect, not overly-determine, the wishes and needs of its inhabitants. At the same time, it is a learning system which, by definition, means it is meant to bring about change. The behaviour of the system, including (most importantly) its inhabitants but also the algorithms and interfaces that underpin it, is intended to affect the behaviour of the people who use it in a positive way.

The process and concerns here are a little like those of architecture except that, at a far greater pace than in almost any physical built environment, the building itself changes continuously according to how it is used, which in turn affects how it is used and therefore changes it. And, of course, if it is not used, it dies or enters hibernation.

So maybe the right metaphor is closer to gardening or landscape architecture - more Capability Brown than the formal variety. What seeds do we plant and where, how do we stop the spread of weeds, where do we prune, how do we fertilise the soil, what features to we add or enhance, how can we be sensitive to the natural beauty of the landscape and the ways the plants naturally grow, yet make it more beautiful?  Even a jungle is shaped by the land on which it grows.

As well as contributing to the design of the space, we are researchers of it. The kinds of research we need to do here, however, must use but go beyond intervention-driven methods that apply well to more controlled activities, or the empirical statistics that are richly informative yet simultaneously void of meaning. Somehow we have to capture the art, not just the science, the living, breathing dance of an inhabited virtual space in which all things exist in flux and the simple act of observation may be enough to change it. Indeed, we must celebrate the fact that our research will change it. Our research methods must be participative, rich and diverse. We must never let our vision overwhelm the desires and needs of the population and yet we must always be sensitive to the fact that, however subtly, if we are to be successful then we will necessarily contribute towards changing them.

This rambling spiel has centred on one important thing: that ownership is crucial in a social system, to give control to its inhabitants. But the space is limited when occupied by many people. The different needs of different sub-populations, many of which overlap, are pitched in an evolutionary struggle where the fittest, not necessarily the strongest or best, will survive. By and large, this battle is seldom the purpose of people and groups that are visiting social sites - quite the opposite, in fact. It is just a systemic feature that is determined by the limitations and boundaries of those sites, their surrounding contexts, and the ways that advantages are preferentially given through design, emergence, historical accident and context in a competition for crowd space and attention.

It must never be forgotten that this is only one of many owned spaces that does not need to be all things to all people. We are not trying to take over the world or replicate things that are better done elsewhere. On the other hand, there is no point in doing it at all if it doesn't attract the crowds and give them a space to connect. Our challenge is thus to construct a richly diverse, flexible environment where the landscape offers sufficient space and niches for all things to evolve and everyone can be in control of their own niches. However, given that we as designers cannot be neutral or inactive in creating that environment, we must always be aware of our ultimate purpose: to make a space where learning and learning communities (informal, non-formal, formal, incidental, peripheral, connected, disconnected) are supported, nurtured and fertilised.

We are creating a world of iteration and recursion that is, simultaneously, both controlling of and controlled by us. This is not your grandparents' information system!