On Google Wave – Part 4: Collaboration

Continued from part 3: extensibility.

Google Wave in many respects is the ultimate collaboration tool. It can be best compared to Wikipedia but instead of striving for encyclopedic accuracy people can work together to strive for their own outcomes. In doing so they can add all kinds of content, including gadgets and robots that can give a helping hand.

For one thing, Google Wave blurs the line between instructions machines can understand (programming) and human language like conventions, expressions, forms of organizing and so on. Since robots are first-class citizens in wave collaborations they are free to participate and modify content as they see fit. This is quite a ground-breaking development, one we’re not very used to. We’re used to spellcheckers, but we’re not used to a robot intervening to alert us of that fact that there’s no point planning for a picnic since it will rain all day next Saturday.

Robots can for example take the current location of different participants in a wave into account, as far as that information is available. But the key point about collaboration is not so much smart robots, but instant collaboration between strangers which robots can facilitate in.

A wave is only accessible to its addressees and originator. Among those addressees there can be robots who receive every update as do human participants. Imagine every street in the world – or in your city – has a wave where planned road works and other obstructions are announced. If a given robot is a participant in all those waves it can index the scheduled closings of each street.

That same robot can keep track of street names that are mentioned in other waves where it is invited to participate. As soon as the robot can determine from the context of the conversation that a street name is mentioned during a certain period it can then add a message saying this street will be closed.

Things can obviously be taken further. Robots could index where somebody is scheduled to be in the future and draw a calendar automatically. This kind of extrapolation turns waves into harvesting fields for metadata which can again be used elsewhere, thereby enhancing collaboration incrementally.

Robots will be able to be this intelligent because of the structure of waves. Waves actually consist of wavelets which itself consist of blips (also called documents). A wavelet is a part of a wave which can be best compared to a e-mail in a e-mail thread or a message is a discussion forum.

Blips are interventions inside a wavelet, as is demonstrated in the video where participants can comment inside the messages of others. Documents form a tree structure inside a wavelet. Wavelets have a list of participants and a list of documents. From the documentation, it seems as if wavelets can’t be organized in a tree structure but form a list inside the wave instead.

This structure allows robots to walk through the structure of the wave to construct context, assuming that people will add content at that position in the wave where their interventions are most meaningful. This creates a new paradigm in the online experience: our conversations will become easier to understand for robots.

In conclusion, there will be a lot of opportunity to experiment with robots of many kinds. One issue that already arises from this overview is the question: when I add a robot to this discussion, can that robot be trusted? Is it appropriate if my discussion would be leaked to the outside world through the inclusion of an unfamiliar robot? Organizations will be able to deploy third-party robots to their own Wave providers thereby hosting them under their domain (<robotname>@<domain>).

In the next post I’ll look at version control in Google Wave.