I considered a big advantage of working downtown that I would be able to go to meetings of the Chicago Agile Development (ChAD). Yesterday was my first chance. A quick bus ride across the river and I shared the lift with Phil Armour, who was giving the talk (although I did not yet know it at the time).
The title was not really decided in advance but turned out to be “On Zeppelins and Jet Planes”.
Much of what he was saying was not all that new or radical for people who have an interest in agile development. The core metaphor of the talk compares a software project to shooting down an aircraft. It says that old projects of the past (on which most of the methodologies are based) were built to automate business problems that were like Zeppelins — large, slow moving and predictable. Today's projects address problems that are more like jet planes — smaller, unpredictable and much faster moving. It's easy for an agile developer to imagine his or her team being the intelligent, self-directing missile that will shoot down a target that a traditional software team would miss.
Along the way, he talked about his five orders of ignorance, from 0th (where you can prove you understand how to solve the problem) to 4th (where you do not even have a process for how to find the right questions to expose your ignorance). He also emphasised that software is a repository for knowledge with unique properties (for example that it can act independently of its creators).
All this was leading to the most radical part of his thinking. Software development is not about creating a product; it is about acquiring and codifying knowledge. Most knowledge is domain-specific — and getting more so. The tools we have, up to and including the top-of-the-line CASE tools, fail to address this. In future, this will have to change. The accountant running spreadsheets to do financial calculations is an early forerunner of what is to come. We are not only talking about the re-emergence of domain-specific languages, but fundamental changes in the way we do software development.
If the value is in the knowledge, the development team should reflect that. Phil imagines a team made up of domain experts, linguists, ontologists and psychologists. They would work with models that are at a higher level than current programming languages, but specific to their domain. (It was at this point that I could see doubts on people's faces. This is radical change.)
The talk went on after the slides ran out as the question-and-answer period gave Phil a chance to expand on what he was saying. He demonstrated a tool called DOME, developed at the Honeywell Technology Center. The main advantage of the tool seems to be that it allows everything about a modelling language to be configured, so it allows people to glimpse the domain-specific languages of the future.
[ Phil did disclose that he had been hired by the HTC to sell DOME as a commercial product. However, without marketing muscle they were unable to find serious buyers. The product is now available under the GNU Public License. So he is not just shilling a product for commission. ]
At around 9pm I decided to run for my train. A couple of minor observations:
- Phil Armour is not very tall.
- He is, however, an accomplished and obviously experienced speaker.
- The slides were slick and used effectively. [ Does PowerPoint have a point in life after all? ]
- He is English by origin and still has a detectable accent.
- The crowd was smaller than he had expected. Apparently fifty to a hundred people is not an unusual number. (More like fifteen to twenty last night.)
Overall, well worth the time spent. It's refreshing to hear someone rethinking the fundamentals, even if the conclusions are debatable. No immediate hints for the developer working in the trenches, but a lot to consider for the longer term. Phil certainly implies that we should be moving to higher-level tools and more domain expertise. At one point, someone in the audience compared developers to the scribes of old. These days, there are no scribes any more.
UPDATE: Dave Hoover was at the same talk and wrote up his own thoughts. Not only did he get the title right, he linked to Phil Armour's slides. (I have retroactively fixed the title and added the link in my original entry above). I'm really happy to see a slightly different perspective on the night.