Everyone likes a plan with three steps.
I have been thinking a lot about the Underpants
Gnomes lately, and here's why. If you invert the first two phases, you get a pretty
good model for tech-only agile:
- ???
-
Collect underpantsDo Agile Technical Practices - Profit
It has the reassuring three steps, but something important is missing.
Let's talk
about the company which attempts to roll out agile with a primary focus on the technical
practices. That would be...almost the whole world. Look at the first line of
the Agile
Manifesto--it says the goal is "working software," not "VALUABLE working
software," although the v-word does come in at the top of the list of "principles" which accompany the manifesto itself.
The IT
department generally acts first within a company to bring in agile trainers and concepts, and
then everyone works together to bring the business people along as well. But if they
can't play ball, then so be it. Off we go to build a burn-down chart.
In
tech-only agile, the business is Tonto to IT's Lone Ranger. Business is Arthur to
IT's "The Tick." Business is the sidecar to the Agile IT
motorcycle.
It makes sense from a job market perspective. While "agile" is magical on an IT
resume, it is not in the top ten skill sets which recruiters expect to see in a successful
product manager's vita (see this general job
site, chosen at random, as an example).
"It's all good," we like
to say, but it really isn't. All too many tech-side agile team members (and
worse--all too many business people) envision the agile project as something that starts when development
starts. "Wait," someone says, "we need a backlog! Yes!
A backlog!" And someone quickly puts something together, and off we go!
We're agile! Faster than a speeding bullet, we've
- pair-programmed
- some unit tests accompanying (and preceding development of)
- modules of no more than 100 lines apiece,
- which will be refactored along the way and
- can be unit, functionally, and scenario tested automatically
- at check-in
- with a build process that takes less than ten minutes,
- pipelined through to some "investigative testing" pre-production environment
- for manual exploratory evaluation,
- plunged automatically into production, and
- later made available through feature toggles.
Except nobody flips the toggles. Nobody knows exactly what they do.
Would this happen in
a lean start up? Never. Would it happen in a giant corporate enterprise attempting
an "agile transformation?" You bet it would.
- The business didn't want to spring for an in-room "product owner" and the "subject matter experts" were unavailable to the team, so the team had to make a bunch of guesses.
- The collocated team seated at state of the art pairing stations arranged in a well-ventilated team room knew from the second hour of discussions that what the sponsors wanted from the software would cost ten times what was budgeted, but management told the team to keep going to avoid people yelling at them. "We can always cancel it later," they said.
- These are the same people who complained that the team designed a Cadillac on a Yugo budget.
No-tech agile, of course, is just as scary as tech-only agile. Perhaps I will need to write
a companion piece. But to me the heart of agile is building the relationship between IT
people and business partners to allow mutual exploration of how IT can strategically empower the
business. The most important part of the project isn't the part where the programmers
show up--it's the part where the project is actually funded. It's a more or less
sacred goal of the team to make good on that investment. The sequence is what is important
here. Pay attention to what you're doing and why, THEN embrace the right technical
practices to get you there in the best way.
Tech-only agile is the emperor with no
clothes. Or at least no underpants.