Scrum has been one of the standard approaches to creating software products for many years. It also is the subject of a lot of criticism. Many have had bad experiences with Scrum. Countless people hate it. Here are some examples from one Reddit discussion:
Believe me when I state there’s more where that came from. These 4 can be understood without additional context. Every day, I see people hating Scrum.
At the same time, Scrum has many proponents. Frankly, I’m one of them. And so are my friends at Serious Scrum. But I also wish to maintain a critical eye and ask the question: Why do people hate Scrum? What are their arguments and do they have merit?
I intend to go through the complete Scrum Guide (2020 version) to try to turn it inside out. Hoping to find a satisfying answer to the question: “Does Scrum actually work?”
I will start the journey with the topics “Purpose of Scrum” and “Empiricism”. They are the foundation of Scrum.
![](https://substackcdn.com/image/fetch/w_1456,c_limit,f_auto,q_auto:good,fl_progressive:steep/https%3A%2F%2Fbucketeer-e05bbc84-baa3-437e-9518-adb32be77984.s3.amazonaws.com%2Fpublic%2Fimages%2F31a0ede9-71f2-43c7-9ad0-4e13b059228f_5792x3861.jpeg)
The purpose of Scrum
The Scrum Guide says that the purpose of Scrum is to help create valuable products through adaptive solutions for complex problems. So:
Scrum exists to create products or services
The focus is on value
The framework is about adaptive solutions
It deals with complex problems
This helps us determine what Scrum is NOT:
It is not about the delivery of fleshed-out products
It doesn’t focus on working faster
It doesn’t work with rigid plans
This is important to note because:
I want to assess if all aspects of Scrum align with the purpose.
I want to know if specific criticisms of Scrum are based upon a misunderstanding.
Now that I discussed the purpose, we can continue to unpeel the Scrum Guide.
Empiricism according to Scrum
The Scrum Guide has the following definition for empiricism: “Empiricism asserts that knowledge comes from experience and making decisions based on what is observed.” It is in line with other established definitions.
This definition implies that knowledge doesn’t come from thorough upfront analysis. Nor from expert knowledge. It also suggests that you can’t make detailed long-term plans based on assumptions and analysis.
Empiricism has three pillars: transparency, inspection and adaptation. Decisions are made upon the perceived state of Scrum artifacts (Increment, Product Backlog, Sprint Backlog). Low transparency leads to wrong decisions.
You are supposed to inspect the artifacts to understand the progress towards the goals. When the goals are endangered, you need to adapt. This can result in updates of the Sprint or Product Backlog or an updated product.
Scrum’s three pillars of empiricism (transparency, inspection, adaptation) align with other famous learning cycles:
Probe-Sense-Respond from Cynefin
Build-Measure-Learn from Lean Startup
Collaborate-Deliver-Reflect-Improve from Heart of Agile
Observe-Orient-Decide-Act (OODA) Loop
To name a few. Scrum is one of the many approaches with a similar learning loop. The learning loop is an integral part of lean and agile practices. So it makes sense that the creators of Scrum claim it is founded on lean. On top of that, empiricism aligns with Agile.
As an interesting aside, Plan-Do-Check-Act has its origins in the scientific method (17th century). And Deming later changed it to Plan-Do-Study-Act to emphasize the learning. Which makes it even more obvious how much it relates to Scrum’s version of empiricism.
Empiricism and Complexity
To assess if empiricism is a sound response to complexity, I will use Cynefin. This widely acclaimed sensemaking framework defines 5 domains;
Clear
These are known knowns. You know that you know the response to a certain situation. There’s a best practice and the typical response is sense-categorise-respond.
Complicated
There are known unknowns. You know what you don’t know and you can come to a good response through analysis or expertise. Here, the response is sense-analyse-respond.
Complex
This is the realm of unknown unknowns. There are no right answers, only in retrospect. This is why you can probe-sense-respond.
Chaotic
Here cause and effect are unclear. This is why you need to take action. The response is act-sense-respond
Confusion
When in confusion, it is not clear which domain applies. Therefore there’s no proper response, except to break it down into consistent parts.
Empiricism is a response to complex problems
Scrum has a response called empiricism, which is about transparency, inspection, and adaption. This aligns with the complex domain in Cynefin. In both cases, you do an experiment, learn from it and then respond to the learnings.
So I conclude that the theory of Scrum is sound. It is aimed at being a framework for complex problems, with a response similar to probe-sense-respond. A team builds a product Increment and inspects it at the Sprint Review, together with the stakeholders. On top of the inspection of the increments, there may also be an inspection of other aspects, like the market, financial position or new technologies.
This inspection will bring you new insights that will lead to an adaptation of the Product Backlog. Priorities may shift or new items may be added due to the insights.
Critisim on empiricism
There are many known struggles with empiricism in Scrum. The majority will be topics of discussion when I address individual areas of Scrum, like the Daily Scrum, Sprint Review or Sprint Backlog.
This leaves me with the following major issues:
Teams don’t use (or only partly use) empiricism within Scrum.
Teams use Scrum to improve delivery instead of improving value while doing discovery through delivery.
I have seen Scrum Teams that used the artifacts and events to plan and execute their work in a traditional way. The major difference with how they worked before is that they plan more often. You can argue that this is wasteful. Why would you plan so frequently without reflecting on what you did and learning from that?
I also know Scrum Teams that only actively use empiricism during the Retrospective. They only look at improving their effectiveness as a team. But they ignore making use of empiricism for the product they create. They will not learn what works and what doesn’t.
When teams don’t use empiricism in a complex environment, they will fail to deliver the highest possible value. And this is the whole reason why Scrum exists.
The second issue is strongly related but still different. Now, the purpose of using Scrum is ignored. Instead of aiming on creating the highest value products through discovery, teams focus on improving on delivery of their product. They will aim to find better ways to deliver (faster, higher quality) their product. They will deliver as requested, not as desired. The entire premise of Scrum is that you don’t know which work will bring you the desired results. It makes no sense to improve the delivery of undesired features.
Teams that forget to focus on value will deliver as requested, not as desired.
Both issues are caused by a misapplication of the Scrum framework.
Verdict on the concept of empiricism in Scrum
Scrum is a framework to create valuable products in a complex environment. It is founded on empiricism. This is a logical response to complexity. The building blocks of empiricism in Scrum are transparency, inspection and adaptation. These are similar to other learning loops, like Plan-Do-Check-Act and Build-Measure-Learn.
The purpose and the definition of Scrum make sense. You shouldn’t use Scrum as a technique in traditional project management approaches. Or as a delivery framework within SAFe.
It also defies the purpose of Scrum to not inspect your product or the way you collaborate. Scrum without empiricism has no soul.
To conclude, Scrum’s premise is sound. It is a framework targeted at creating a high-value product or service in a complex environment. It is based on the learning cycle of empiricism.
I will continue my journey with the elements of Scrum to asses how these align with the premise.
Good read !!
Interesting read and works well when building a new system with new features when all is complex. I struggle a bit when it is not so complex any more, for example the system needs 300 templates generated as PDF with data coming from different sources. For the first 5 this is challenging for the developers. However, after number 10, this is just a little factory, doing the same over and over again. No complexity here. All clear and no real need to dive into the added value, that was already clear from the first rounds. And all 300 are really needed, which takes about a year to complete. Any suggestions how to incorporate empiricism here in a useful manner?