Windmills 2: Fixing Scrum
Responses from Scrum officials to my recent Scrum articles here and on more private slacks have not been favorable. I think I’ll sum up and move on for now.
I’ve had a bit of support from Scrum Alliance trainers and reps, a pretty civil conversation with one of the individuals who “refactored” the CSD, and largely unproductive exchanges with the Scrum Alliance Chief Product Owner.
If I correctly understand the remarks from the latter individual, he’s angry because, in his view, I’m undermining their efforts to do a good thing. That’s not the case: I am criticizing the present work. We’re all trained not to take criticisms of our work personally, but few of us can do it. He also said something along the lines of “maybe if you started saying nice things, I’d consider your ideas”.
Oh, and I was assured that the CSD refactoring team was given “no marching orders”. I’m not sure what that means. The group has been in place since at least 2018, and it seems odd that they’d be doing all that work with no charter from the Alliance, no understanding of the Alliance’s expectations.
Enough about that. Let’s move on. My plan here is to summarize my views, based on what has gone before, and then to get on with my life. At present, I foresee no Windmills 3 article, unless something changes. Things do change, so it could happen.
Where I’m At
Im going to bend over backward somewhat to be fair to what they’ve done. I might even try to be a bit generous, but I also intend to come back to what I firmly believe.
I think the Alliance and I start from the same recognition: the CSD program has essentially failed in the marketplace. Few courses have been sold, and the CSD rolls contain a tiny fraction of the CSM rolls, when they should be three to ten times larger.
We even agree, I think, that the prior program was too expensive in time and money. Developers had to come up with high daily course prices for three to five days, and three to five days of time. And few development organizations will support that kind of time and money.
Chet Hendrickson said something bearing on this the other day. I’ll paraphrase:
When management looks at Scrum roles, they see ScrumMaster, and they check their organization and it doesn’t have any ScrumMasters, so they train some. They see the Product Owner role, and they don’t have any, so they train some. Then they see the Developer role, and hooray! they have lots of developers, hiring only the best, as one does, so they assume they are good to go without training.
I think there’s likely some truth to that. Management doesn’t know that developing in an iterative incremental style is different, so they are not open to the need for learning. And, as a rule, development teams don’t have much budget of money or time for training anyway. Hired only the best, you know.
One other point of agreement is that one of the two ways to get the CSD was to take ScrumMaster training first, and then three days of developer training. So it seemed like you had to be a CSM to be a CSD and that may not have tasted right to developers and their teams. I think that’s at least possible.
This effort started in 2018, and I’ve reviewed all the material that I can from the Slacks I have access to. I was active and sporadically vocal about my views over most of that period, although I was pretty quiet during the worst Covid era. So I saw what was going on and I’ll do my best to describe and explain it.
The approach taken, as I see it now, was to replace the then-current CSD certification offering with a new and better certification offering. The fundamental assumption was that the offering would be certification. That’s a natural assumption for the Scrum Alliance and its trainers and reps, because it is essentially the essence of the Scrum Alliance strategy: bring the value of Scrum to the world by offering courseware and other learning leading to certifications that the market will value.
This assumption is very much baked into everything the Scrum Alliance and its trainers and reps do. The certifications on offer start with the basic CSM and CSPO, but then there are the advanced levels, which can be attained by a combination of training and other learning. The advanced levels essentially mean that any training or coaching offering of a trainer or rep applies to obtaining another valued certification. “This course qualifies as a component of attaining the Advanced Certified Scrum Person” kind of thing.
The assumption is so baked in that there’s really no alternative that is even capable of being considered. We could talk about whether that’s right or wrong, moral or corrupt. For now, the fact is, everything they do is founded on certification.
Solve the Cost Problem
So we “know” that our certification program is too costly in time and money. What do we do? Simple and obvious. We refactor the program so that it delivers value sooner and at a lower cost. “Incremental Release”, they seem to have called it. The idea was fairly simple:
- Offer at least two levels of certification, CSD and Advanced CSD or something, and make the initial one, CSD, less costly in terms of time and money.
- Put the rest of the value, or even more value, into the A-CSD, and once they are brought into the certification flow, it will be easier to get them to invest the time and money in further learning.
A minor glitch was that that left existing CSDs in an odd position, since they wanted to name the initial one “CSD”. They fixed this fairly elegantly by promoting all the prior CSD holders to A-CSD.
There are of course many ways that the learning objectives for the new program could have been divided up. They seem to have chosen to put introduction of ideas in the first set, CSD, and the practice of ideas in the second, A-CSD. There’s nothing to say that you couldn’t teach a practical CSD to introduce the ideas, but the model seems to assume you’ll do concepts in CSD and practice in A-CSD. And since the course seems to be aimed at two days, there’s not much room for programming in there if you’re going to cover the material.
I’m prepared to be reprogrammed on my understanding, but supposing that the basic split is concept(CSD) practice(A-CSD), there’s the question of whether this is a good split.
Remember the basic idea: we want to deliver value to the learnings sooner, at lower cost to them. So it makes sense to have a two day course and then more learning. But should the course be concept first, then practice?
It seems clear that the refactoring team decided that it should be. I think that there is a strong argument about doing it the other way.
If offered the choice between a programmer who knew what TDD was, and a programmer who had actually done TDD, which one would be more valuable? I think the answer is pretty clear: practice is more valuable than concept, because practice includes concept plus application.
So I argue that we should prefer, if it were possible, a CSD with more practice over one with less. The question that arises is, of course, is it possible?
I’m not sure. If I had to introduce enough Scrum concepts, and there are plenty in the CSD learning objectives, there might not be time for anything else. But it might well be possible to design a course with two days of instructor contact that gave developers real practice in doing TDD and refactoring, and, at a stretch, maybe even incremental development.
It might be possible. It might not. I could imagine the team asking themselves if it could be done, finding that they didn’t know how, and deciding, OK, concept first, because they could see how to do that. I don’t know if that happened: I could easily believe it and easily believe that I could have been convinced that concept first was the most we could ask for.
Where that leaves me, since I am not bound by limits of space and time, is that the CSD to A-CSD program as now defined delivers less value than the one I can imagine, where there’s more front-loaded actual practice, delivering actual development value sooner.
To me, the purpose of CSD was to provide developers with the basics of the ability to deliver software iteratively and incrementally.To me, we have taken a single unit that did that, CSD, and divided it into two units, CSD’ and A-CSD, and removed the basics of the ability to the end. We deliver some value sooner and more inexpensively, yes, but it’s not the highest value. Developers need to be able to do it. If they can do it, they’ll come to understand it. And, in my view, they can’t really understand it unless they can do it, Lego refactoring notwithstanding.
So I think the refactoring is at best shipped in the wrong order, and at worst is not the right split, since it seems almost to force delivery in that order. Infrastructure first, then function, which is a flawed approach over infrastructure and function together. Sometimes necessary, but never desirable in my view.
Because in my naïveté I think better is possible, I am disappointed in the current result. This, of course, makes the people who worked so hard sad, and that makes Howard sad. And that makes me sad, because it has gotten in the way of understanding, and driven me to the feeling that when I end this article I’m going to turn the page and move on to something I can affect usefully.
Prospects for Progress
I am ever hopeful, though I am not much inclined to engagement in the present situation. I’ll mention some areas where progress should be made in my view.
- Full Responsibility
- I’d like to see the Scrum Industrial Complex take on a feeling of responsibility to all individuals trying to use Scrum, not just their own trainees and certificants. I’d like to see them using their resources in outreach to people working under Scrum via other than current high-cost training options.
This is probably asking too much. Certification is in their blood, in their genes. But they’ve created the entire ecosystem and I believe they should think of themselves as curators of the whole.
- Beyond Certifications
- Of course neither the Scrum Alliance nor its trainers and reps will turn away from certification: it’s an effective marketing strategy, central to their success. But
- Identifying Real Value
- I’d like to see a stronger focus on what’s really of value to the customers, which isn’t just more certifications, but the abilities that are implied to be present in certification holders.
- Delivering Real Value
- Given a serious focus on real value, there could and should be a pretty creative effort around devising ways to deliver that value sooner. I’d bet it’s possible. I’d also bet that it would have to look differently from current training offerings.
Here again, this is asking a lot, because the pocketbooks of the trainers, reps, and alliance itself is predicated on the current model. Better to start solving it now than to be eaten alive when someone else solves it.
- Focus on the Increment
- I’d like to see the Scrum Alliance specifically increase focus on the importance of delivering the Increment in every Sprint, and on the key fact that that always requires learning and behavior changes from the developers, no matter what the domain may be. Software, we happen to know some things that are useful to know. Scrum Beekeeping, I at least do not.
That focus would include adjustments to learning objectives, to training, to coaching, to all kinds of communications. It would take time, but once “Increment” is on the check list, a lot of it would happen readily over time. The cost would be quite low.
- Continue CSD
- I think the new CSD will likely bring in more individuals looking for the certificate, at this lower cost in time and money. I think it’s “skimming the cream” and will not reach deeply into the market, and I think it will be quite some time before we can even tell if it increases the uptake of the entire CSD/A-CSD learning. But it can’t hurt.
- Training Can’t Serve the Need
- They need to recognize that conventional courses as presently offered cannot possibly serve the millions of developers who need help in any finite time. New approaches are needed.
- Focus on Outside Learning
- They have the “SEU” program, “Scrum Education Units”, which offer credit toward certifications. In the case of software development, this needs aggressive focus to identify and bless useful offerings.
- Farm New Approaches
- They could provide encouragement and support for alternative means of bringing the necessary skills to this very broad market. This would, of course, require looking beyond “sell certification and training” model, so it is perhaps unrealistic to wish for it.
I wish for it anyway.
This is the same list, in a slightly different form, that I’ve been touting in recent articles, and in communication with the Scrum Alliance since at least 2018. Progress against the list has been, dare I say it, limited. I’ve not even seen recognition of the ideas much less sensible reasoning about them. It’s as if I’m on a wavelength that they’re not tuned to.
It wouldn’t be the first time. But I still think this is needed.
And for now, I’m going to turn my attention elsewhere, unless a door opens.
See you on another subject!