My experiment (contract) has ended with interesting results. Subsequently I am shopping around for a new project. Have something interesting? If so please drop me a line.
I can now say I've seen scrum in action. Participated even. And for the most part, I think it's a good thing. Definitely helps maintain momentum and focus. But does require a full time scrum master to beat the drum.
I think one of the pitfalls is making sure the scrum process isn't itself the project deliverable. This trap seems to be easy to fall into if the project itself is a pilot for agile process.
Strangely this is nearly the opposite concern I expressed over a year ago in Two Directions of Computer Science.
On this project, it seemed many of the senior people, in one regard, rightly have started to focus their career development towards the process of developing software. That is, learning and applying scrum or other contemporary process.
Unfortunately their technical experience has only grown at the pace of new Java JDK releases. With the atrophying of (or lack of development of) advanced technical skills and planning, there is great risk "process" will stand in for technical design. Remember that agile process does not an architecture make.
Or worse, the same skills and solutions will be applied over and over. Not every application should be built on a business object, object/relational mapping, and RDBMS stack. And if it is, stop re-inventing the wheel, use modern tools, shun broken patterns, and keep the system complexity to a minimum.
I guess this bifurcation is necessary, and people must specialize while being aware of their limitations. And must not consider advancement in one discipline a universal sign of achievement across all of them, and in turn behave arrogantly.
I've also already made the point that projects must "iterate over risks".
Another perspective on that theme is to iterate over the business value. Rebuilding a legacy system in Java from C and Perl by itself provides little value, especially when dozens of people will need to be retrained and the system retains many of the same limitations. But building a system that delivers results in a fraction of the time the legacy system executes does deliver great business value. It may even change the competitive playing field.
See Google and their Map/Reduce, and the recent announcement by Yahoo, officially supporting Hadoop.
Anyways, it was a great experience. I met great people. And did a lot of traveling through Europe. I wish the team great success and hope things don't change too much for them.
Leave a comment