SoftwareEngineering

So Agile, they'll need another word

This last week has been the first of a project planned to take four weeks. It is a great opportunity for me, since the project's goal is to deliver on an idea I pitched, some months ago, to solve a problem with the way we test. I have been very pleased with getting this far. When I first proposed what I called 'a pretty radical change to the way we approach testing' I wasn't sure I'd be able to convince people it was the right thing to do. But as it turned out I had no real resistance. In fact so positive has been the response, that many have long forgotten that I got things moving in the first place.

Ironically I was busy constructing this plan when I was being told that I am 'not perceived as having enough sense of urgency on the day job' Apparently I should have been more head down and focussed on the job in hand. However I think that criticism was short lived, and short sighted. By planning the changes I wanted to put in place, whilst everyone else was too busy looking no further than the next day, I found myself with a plan and a proposal ready... just as the management team started to wonder what on earth people will be working on once the last version shipped.

My proposal got discussed, and sized. The team lead of the automation team, who will ultimately own the changes gave a sizing of four person months. And he was insistent that he get to pick the four. It's no good just assigning four random people and hoping for the best. He wanted people he knew would be able to deliver. He also insisted that these people be allowed to focus on the job. This sizing was for four people working full time - not being asked to fit it in around other priorities. This was a pretty tall order, and the fact that we got granted those demands is both a factor of having sold the long term benefits. And the management team having little better planned.

As fortune would have it, I got selected as one of the four. I am certainly the person most interested in seeing the idea succeed. And I'm grateful for the unusual opportunity to actually spend some focused time on delivering my idea. I've often thought it would be cool if submitting patents would come with the possibility of being given funding to develop a prototype... but no.

So here we are, end of week one of this four week plan. We are attempting to follow the 'Agile' method.

For those unfamiliar, Agile is nothing new as such. it is a rebranding and repackaging of various ideas that have been evolving for some time. It is - in my mind - an attempt to have a software engineering process that fits with real business needs. Gone are the days when clever people could lock themselves away to produce something clever, and just deliver when they thought it was ready. Many of the older software development processes just don't work in a world of 'marketing splash dates' and tactical deliveries. You have to be able to adapt to the market. Being half way through a 2 year development cycle when the market changes could be devastating. One of the main features of 'Agile' is that you construct just enough to deliver customer value every 'iteration' And those iterations are as short as possible. That said we're doing 1 week iterations which is very very short. But the idea is the same. At the beginning of the iteration you plan with your stakeholders/customers, what you will deliver in this iteration. Something they will be able to see, and use. At the end you demo that to them. This helps to keep you focussed on delivering just what you promised, and no more.

I can say after 1 week that it is *really* hard not to get drawn in to considering design and longer term goals. How should we do 'this' so that when we get as far as needing 'that' feature down the line it will work right, etc. Finding the path between bad design that will make life harder in the future, and over engineering a solution at the expense of delivering this iterations value. But the driving goal must be, that in the modern business world, you must be able to deliver value at any point. If the project got canned next week, we'd have something that would provide some function. It isn't everything people want. but it's also not just a pile of unusable code.

The aim for next week is to actually deliver enough to go live on our 'customers' website. It will not be ready for full fledged use. It will still have some serious limitations. But it will be able to do some key piece of the project goals. From there forward our delivery will be to update that live system with the latest set of capability. Every week our customer will get a little bit more of what they wanted.

This week when asked how it was going, one of the group replied, "We're so Agile they're going to need a new word". Which in part is a reflection that with this process you can apply it to greater and lesser degrees, depending on the project. With just five people in one room, and a single month, we can take it to the extremes of Agile development.

Another cool thing about the project, is that I am sitting in an office with four other guys (we ended up getting a fifth person for the first couple of weeks) And they are all super smart, capable developers. It's unusual to work in this way, it's not that I normally work with people that aren't clever. It's more that normally the way of working doesn't let you collaborate, as much as just do your own piece without that much interaction.

This week I've been blown away by how fast we got something working, and doing clever things. And it's really energising to work along side people that know how to do things I don't, and (hopefully) fill my own roll of expertise. It's an education looking at the code they write. I think nothing of throwing together the sorts of code and solutions I'm familiar with, and they all have different areas that they handle with the same ease. Together we seem to work pretty well. There are some differences of opinion on design, as is healthy. We all help to stop each other getting drawn into considering options and designs that are not part of our current deliverables. And so far we've not hit a problem that has stumped us all.

I haven't been this happy at work for a long time. I'm actually finding it hard not to keep on working during evenings and weekends. However another key feature of Agile is 'sustainable pace'. It's no good relying on people working every hour, they will burn out. As enthusiastic as we all are, I've been keen to remind us all that this project may be used as a template for others. And we owe it to future projects not to succeed only by working ourselves to death. We prove what we can do working just regular working hours.

Perhaps the strangest part of this agile project, is the acceptance that requirements are shifting. There is no hard and fast list of what we must deliver at the end of 4 weeks. There are key capabilities which form our first areas of focus. But at the start of each new iteration we review what, out of the possible options, is most important to deliver next. The expected output of the project will be the most important parts of a working system AND a list of the things we wanted to do, but were prioritised lower than what we delivered, possible enhancements that may form the basis of a separate project.

That's probably more than enough of the subject for now. However I will post updates as we go through as to how we are managing this Agile project. And if we have hit the much touted problem with Agile, that of hitting a point where you realise you've designed yourself into a dead end by virtue of not thinking far enough ahead.

Also I may post about Rational Team Concert.. for now I will just say it's awesome. (Yes I work for IBM, no I'm not paid to say that, nor is it praise I'd have for many other Rational tools).

Check it out at jazz.net