Re: Spurring the adoption of unit testing (Re: [scrumdevelopment] Re: Is Scrum a block to Agile?)
- Hi Ron,
I've tried to summarize my comments here; I've inlined comments to
address specific questions not addressed immediately below.
While I perceive the situation a certain way, it would appear that
others on the team don't. Maybe I have it wrong, maybe they do. Maybe
my pain threshold is lower (too many Death Marches perhaps?), maybe the
team's is too high (in my opinion). Maybe it is purely inertia.
Anyway, real data, along with a discussion with my boss about what he
wants from the team should help objectively identify our problems and
focus our improvement efforts.
I perceive our "initial" quality (that is, the quality of the code
leaving the desk) as a pain point, because I think we'd go faster if the
initial quality was higher. I guess I also think this is a bit of low
I was planning on talking to my boss this week about things we may want
to consider as we become more agile, I think your article "Agile, Top
Down" would be a great catalyst to focusing this effort, thanks for
pointing that out. I'll definitely bring it along to discuss.
Ron Jeffries wrote:
>> Defects rarely escape the building, so perhaps my boss doesn'tQA keeps the defects from leaving the building. They are good at
>> see this issue as being as important as I do.
> Ah ... what's keeping them from leaving the building? Could you
> perhaps measure the cost of late vs early discovery, using things
> like the charts we talked about below?
testing early and often. They are also good at identifying where their
tests are weak and addressing them, e.g. they are trying to address the
lack of automated acceptance testing.
>>>What would happen if the boss asked for a chart on the wall, showingI'd prefer not to surprise my boss, I can't imagine that he would be
>>>defects by module? What would happen if there was another chart near
>>>that one, showing module by programmer? What if there was a chart
>>>relating test density to defects?
>> I'll take a look at producing charts, and talk to my boss
>>about whether he wants them up or not.
> Or ... just put them up? Make them factual, not directed at
> individuals, and see what happens?
upset, but you never know.
>>I think the problem is that we all have our own offices.Yes, I just don't think yet. :)
> Indeed. There are those of us who believe there is no agile without
> togetherness. Recent U of Michigan study shows 2x productivity when
> you're in the same room. Would that be of value? ;->
>>> I think he thinks the way he works is fine. He hasn't tried unitNo, I don't think he does, so data is the key to solving this problem.
>>> testing to any great degree so he doesn't understand the benefits nor
>>> how to go about it.
> Does he really know how much worse his defect rate is? Or does he
> imagine that he's really much the same as everyone else? If the
> latter, data might help ...
I also think he doesn't understand using unit tests to help pinpoint
problems. As an example he recently "fixed" a defect, but his
explanation basically consisted of, "I don't know how this fixes it, but
I changed xyz". I expected that asking "so how do you know you fixed it
if you don't know what the the problem was?" would cause him to rethink
his solution. He suggested QA just retest it.
So he and QA will enter this little cycle where they'll determine
exactly where the problem is and point it out to him and then it will
really get fixed. :(
- I've been away from this list for a while, I hope the following is
Bob: your comment hits home:
> It requires someone with the visionSome reflections: (not particularly in relation to your note, Bob)
> and a mastery of the corporate culture and politics to get
> something like Scrum going strong. To bring others in at their
> own pace without starting a holy war. And it's a full time
I think we need to stop expecting that every ScrumMaster has this
skillset, or even the personality to develop it. Just as we
encourage our own team members to leverage one anothers' strengths,
we SMs should identify and support those who "do enterprise change"
well, and let them lean on the rest of us to help make their
I find I'm very happy doing the SM gruntwork, helping a change
leader with my observations, intuitions and supporting work. I'm not
at the place where I want to (or should) lead the effort. I feel
that getting clear on my place in this is an important step toward
contributing effectively on the enterprise scrum team. There are a
number of different, important roles to be played by SMs with
different personalities within such an effort.
So I'm wondering: How can we help each other identify our own best
niches, and foster better teamwork among SMs? (Cat herders seem to
make very bad cats :-)
We can't learn this from the outside - we need to get our feet wet,
and "inspect-and-adapt" as we go. That means we'll continue to make
mistakes and recover from them. But I think that opening up such a
dialogue among us could help us learn and recover faster and teach
one another better. I know that the Gathering wants to address this
need to some extent - but once or twice yearly seems a long interval
for our inspect-and-adapt cycles, to me.
Once again, I'm musing in public. Well, perhaps someone else is
doing the same and we'll find some kind of synergy here :-)
--- In firstname.lastname@example.org, "bobschatz" <bschatz@s...>
> Being one of the folks that replied privately to Tobias, I'll put
> thoughts out there.a
> Change is incredibly hard. There's no doubt about that. Scrum has
> way of unearthing all of the ugly things that have worked theirway
> into the development AND business processes of building software.Schwaber
> There are people and organizations that can adopt change faster
> because they are either in enough pain, or they all know that
> traditional ways just won't work. I discussed this with Ken
> over two years ago when I urged him to prioritize his own backlog.people
> Some organizations with "play" with agile until they are convinced
> in their own time that it will work for them.
> Others will never do it.
> One of my fears three years ago was that we'd have a bunch of
> running around trying to make a living off of Scrum. Jumping intoand
> organizations, holding classroom sessions with process manuals,
> moving on. I live with Scrum every day in building products and myI
> company depends on it's success for survival. I spend everyday
> observing and thinking of how it could be better the next day.
> Sprint after Sprint; year after year. I've been in daily Scrum
> meetings everyday for over 3 years. It's a challenge of a career.
> don't think of the adoption as a start and an end. It's more likean
> journey that you don't really see ending. You just focus on whatis
> at hand and deal with the success, failure, issues, changes, etc.vision
> that you must face each day.
> When I speak at conferences I talk about success factors that no
> consultant can deliver to an organization. Successful change
> requires outstanding leadership. It requires someone with the
> and a mastery of the corporate culture and politics to getsomething
> like Scrum going strong. To bring others in at their own paceas
> without starting a holy war. And it's a full time job.
> Because we are in a business where everything is expected at light-
> speed we think that our new techniques should be embraced by all
> because they're just that obvious.
> The truth is that the adoption rate will follow the same patterns
> other initiatives have. Some fast and successful, others slow orin
> failures. The important thing is to learn something along the way,
> and focus on ways to add value to making the organization better
> small steps. The good news is that every success in projects usingIt
> these techniques can, and will, help remove barriers for others.
> will also serve as a reinforcement for those who created thea
> success. At Primavera, I made sure we invited other companies into
> our environment to see how we were using Scrum. That gave my team
> boost in confidence, and the visitors now had a vision of what itperfect?...certainly
> would look like in their organization. Were we
> not. Scrum is not about perfection, it's about doing something ofdirection,
> value, getting feedback, learning, and trying it again. I've had
> many discussions with other executives in my position and helped
> them understand their role in being a strong leader of change. We
> all need to learn how to manage expectations. Not only of others,
> but our own as well. Agile adoption is going in the right
> and those of us that "get it" need to be available to coach andhelp
> others that want to join us in the journey.your
> I typically close my presentations with this advice:
> Be Patient, Positive, Persistent Don't give up!
> Bob Schatz
> Solstice Software, Inc.
> Chief Development Officer
> --- In email@example.com, Tobias Mayer
> <tobyanon@y...> wrote:
> > Scrum is touted as "can be implemented in a couple of days at
> organization" (or words to that effect. While this is true, whatway
> does it actually mean in practice?
> > My recent experience with Scrum being introduced in a top-down
> within a large organization has shown that "doing Scrum" takesScrum
> priority over good practices. After six months of doing Scrum,
> almost no difference is perceived in the underlying engineering
> practices at this particular organization. The promises from
> consultants that "the teams will change their practices after twoor
> three sprints" has not been fulfilled. There are many reasons forOO
> this, most, if not all, have a cultural and/or political basis.
> > I am thinking of one team in particular, who for months were
> suffering from vague requirements, inability to estimate, lack of
> decent upfront testing - certainly no unit testing, lack of good
> design and a well-componentized code base (and little training tobefore
> teach them other ways of doing things), over commitment, burn-out,
> and on and on... Stuff got released, but then it always did,
> Scrum. The quality did not noticeably improve. The team appearedto
> exhausted. While these concerns were all aired - repeatedly - it
> was only when the team finally announced "we don't think we want
> do Scrum anymore" that action was finally taken. The exec teamwent
> into "firefight" mode to solve the problem - or, as I see it, toput
> a good face on it again.satisfaction
> > When the act of "doing Scrum" becomes more important than the
> quality of the product being produced, or the level of
> of the workers, then "doing Scrum" becomes an empty, meaninglesscan
> piece of market-speak. The danger of making Scrum so easy to
> implement - and encouraging organizations to do this is that it
> become a sham. People on both the inside and the outside of suchorganization.
> Scrum projects will quickly recognize this for what it is. Scrum,
> of course, will be blamed for being ineffective. Yet another
> scapegoat will have been identified.
> > Scrum has not been altogether unsuccessful at this
> Where the current process was so broken that nothing was beingsolutions
> produced at all Scrum has helped teams to begin delivering. Many
> people on the Scrum pilot teams are enjoying the experience of
> better communication, and a sense of empowerment. This is all
> good. But the bottom line is, that the essential practices of
> creating good software - great software - are not being introduced.
> > Does anyone else have similar experiences with Scrum in a large
> enterprise (the problems I mention may not be confined to large
> enterprises, but it is likely they are)? What might some
> be? If you are at Agile2005 this week, find me and let's chat.
> > Tobias