Konubinix' opinionated web of thoughts

Agile & Scrum Don't Work | Allen Holub in the Engineering Room


  • External reference:

what I understand of Allen Holub’s opinion

In this episode of the continuous delivery podcast, Allen Holub gives his opinion about the current practices of agile methods and scrum.

He regrets that the ordinary use of agile has become meaningless and people tend to thrive for getting out of agile by describing some ideal that actually is the definition of agile (according to him).

They regret also that scrum and derived methods like SAFE1 have become dump stuff made to be sold to people that are not willing to do agile but still want to have some kind of certification.

Allen Holub indicates that the concepts of scrum are meaningless and oftentimes harmful, like the notion of accountability that entails a violent wording.

He criticizes the scrum priest state of mind2. He is fed up with the fact people defend the inspection and adaptation of scrum without realizing that those actually are already part of agile. Like if scrum had given something new, while it is some rebranding of agile with loss of meaning. In his opinion, this became like this when money has been involved.

He regrets that people are estimation priests and emphasizes the fact that we won’t know how to estimate a things prior to having some experience about the things.

He even says that scrum is contradictory when dealing with estimation and pretending being based on lean. If you repeatedly try something and repeatedly find out that it does not work, you should stop trying it.

In companies he came as a coach, they hired him to make better estimation. When he asks them why, they answer that they won’t spend a lot of money for a project unless they know how much time it will take. He then teaches them that this is the wrong question as well as how to make iteration and repeatedly adding value. When they ask him what is this wonderful method, he says it is agile. But the ordinary use of agile was so strong that they did not even believe him.

He is strongly against trying to define metrics, because, according to him, they tend to fall into the hands of the managers and therefore become victim of the Goodhart’s law.

Companies tend to confuse managers with leaders. They should have more leaders and less managers.

Then, even though scrum did a lot of harm to agile communities in general, the scrum guide only defined meaningless meetings in an empty method that per se are not harmful. It is what people made of it that was harmful.

They also discuss the slacker fallacy and the fact in performance reviews people are considered the cause of the issues. This contradicts the idea of agile methods that needs empowered people (also, see often the system is perfectly designed to generate the result it did).

More generally, he thrive for leaving more room for intuitions. For instance, he deplores the fact HR hires people without considering future coworkers’ opinion.

Also, he thinks that informatics does not needs solid mathematical knowledge but rather solid social and psychological knowledge3.

My opinion

To me, he describes something very similar to the tragédie de la définition d’un méthode. From the initial definition to the ordinary use that converges after a long chain of references, there are several differences. Then people tend to rationalize their opinion using the initial text, becoming method priests.

I cannot tell whether I agree or not about whether scrum is a good method or not because, as Allen Holub says, people tend to use the word scrum to do other things that what is described into the scrum guide. Therefore, it would need many experiences of scrum by the book to have a sensible opinion about whether it is useful or not.

Whether he is right or wrong, it looks to me like he developed some kind of aversion to some words, like accountability, metric, scrum. I agree that those words are very loaded, but in some sense he stopped the game of communication there and apparently stopped trying to understand what could be good in those notion, because of his bad experiences.

For instance, it sounds like he despises the notion of metric, even when it may be used to describe exactly what he is suggesting: asking around and gathering answers. I think he means that people tend to blindly trust numbers and formulas and make then suddenly become objectives, rather that educating the system 1. Also, it looks similar to the decision hygiene of don’t use scales when measuring something. Yet, the apparently systematic reject of the word itself, while a good heuristic is to me the sign of a polarization, due to the too many bad experiences, and making his arguments about this particular subject doubtful.

Notes linking here

  1. described as Shitty Agile For Entreprise

  2. However, he defends some of his arguments citing the agile manifesto, making him look like a kind of « agile priest_». To me, he might well be victim of the priest fallacy↩︎

  3. to me, he falls here into the curse of knowledge. The fact he apparently has some mathematical knowledge gets him away from understanding what it is like not to have it↩︎