I've been using Twitter for about a month of so and I like it so much that I think that it should replace IM, but that didn't happen until I got Twhirl. The client makes all the difference. Knowing that, there is some potential for applications of this IM technology, if the client is built well and integrated into something reasonable. So I'll ruminate about how I would apply Twitter to enterprise applications.
The obvious application is workflow. There are times when you need to know exactly when something is happening. The moment it happens that's all you need to know, and once it has happened you don't really care so much. An IM is a perfect thing, but what's better about Twitter is that I can think of it as a notification system - which is to say that the thing on the other end of it can be a process, not necessarily a system. The difference between a twitter stream and a set of emails is that, well an email is a memo and a tweet is a doorbell.
I have hundreds of backend notifications in my Google mail going back over a number of customers. This is clutter that I don't need and don't want to be responsible for cleaning up. But if I had something like a massive log file that I could access someday... That's a tweet stream. So the enabling of notification systems to which I can subscribe and automate is a very nice way of thinking about making communication more efficient when a significant number of people want to know:
When is that Websphere server going to be back up?
When did the last defect run on the number three manufacturing line?
When did the Cleveland office submit their closing numbers?
When did the review board approve the preliminary budget?
When did the ambulance pull up to the emergency room?
Twit agency is a pretty cool idea. I would attach it to any number of complex workflow systems. Loan approvals, budgeting cycles, manufacturing processes. Any time people collaborate over distance and there are critical path milestones that are of interest suggests to me an opportunity to informate with twit agents. The more complex the schedule, the more nuanced the dependencies, the more I want it.
At a level above these transactions are opportunities for meta-scheduling and planning and feedback. At the human perception level, the crowd will be moved and grow to get some sense of when certain events flash over time, but keeping records of such events can allow for computer aided analysis. I expect that I will get a tweet anytime there is a serious traffic accident on my primary route to work by subscribing to the traffic net on my section of the freeway, but after three months of daily notifications, I don't have a good sense of whether or not I should simply plan to take a different route or play it by ear. That's the kind of thing a longitudinal analysis would give.
Not bad eh?
Comments