“These sliding alert windows must stop.”
That was the text of an instant message from an executive talking about some software I was writing, along with a few members of the WebAhead team at IBM. It was 2002 and we had released an internal project called IBM Community Tools (ICT) that combined instant messaging with Broadcast Messaging. The anonymous executive’s request was not without precedent from other users of the software – but they where not in the majority.
The Broadcast messaging tool used a Pub/Sub engine written by IBM Research that allowed for messages to be sent to a large number of users subscribed on a particular topic. These messages weren’t email, just bytes over the network. It was up to the receiving software to determine what to do with them. In the case of ICT we sent around XML messages that originated as requests from individual users for response from the community in the form of a poll, group chat or 1-on-1 chat. For online users subscribed to the topic (community) that the requester was using they were alerted with a sliding window that contained the content of the request. This window would slide into the user’s screen from the right, wait 10 seconds, and slide away.
This was based on a previously successful system, with a small cult following that alerted the users with an even more intrusive pop-up that did not go away until the user closed it. When we were building out the system for a lager audience we realized that that type of interface would not scale for the broader user population.
It’s important to mention here that by default all users were initially subscribed to the “everyone” community. This was the default community for questions that did not have a clear category. The software allowed for client side filtering so that instead of subscribing to messages for a large list of communities the users could list key terms that would indicate a message they were interested in. We fooled around with an adaptive filtering scheme at one point, but filtering in general was largely unused.
This meant that most people who were logged on to the system got the messages sent to the “everyone” community. It also meant that most people got messages that were really not relevant to them, but *most* users didn’t complain. It’s funny to think about it – in essence we built a spam tool and just put it out in the IBM internal wild and let it run loose. At first glance its surprising that anyone enabled it at all.
From the beginning there were people who “got it” and they were on board right away. For the people who considered it a livable minor annoyance, over time, we noticed a pattern, there was a very simple tipping point that changed people’s minds. If they used the system to find the answer to a question that they had no other clue about how to answer they were hooked. There seemed to be an instant transformation in their opinion not only of the tool, but their perceived inconvenience with receiving broadcast messages. These people then started to become evangelists, slowly changing the minds of others.
I believe that this principle applies in most social software and is more evidence for what Tom Coates mentioned in his talk. There must be percieved value to the user. In the case of ICT, that perceived value might not have come for some time and I am sure that, dispite its relative success internally, there might have been something that we could have done to show that perceived value from the beginning.