Following from my previous post, this is my response to a second thought-provoking article that criticises a codification of agile software development into rigid project management frameworks. In that article, Mike Hadlow asks why agile has come to mean just management practices (stand-ups, retrospectives, two-week iterations and planning poker), divorced from any base in technical practices. He bemoans projects in which non-technical people are given the role of Scrum Master, enforcing agile rituals without understanding what the team are actually doing.
I can see that Mike's scenario would be problematic. I have seen examples in our organisation where some non-development staff may have thought they could control an agile team (e.g. as a business analyst) but the structure of the agile team has (correctly) worked against them. In general, it is certainly easier to explain daily stand-up meetings and two-week iterations to non-technical people than it is to educate them in the ways of continuous integration and refactoring. But that has been OK for us, because the non-technical members of the teams have other responsibilities.
Perhaps it has helped that we don't follow the Scrum process per se and our teams don't have a Scrum Master. There is no one person in charge of our teams. We do have a project manager who contributes to the team but they are not in charge. They are also outnumbered: a typical agile team in Applications Division has two developers, a half-time project manager and a half-time business analyst (who may sometimes be the same person). The business partners will contribute someone working half-time as a business lead and tester. (The exact composition of the team will vary depending on the exact nature of the project).
We have found that the project manager and/or business analyst have very important roles managing the customer liaison and testing, and setting up user representation. In fact, from my point of view, one of the biggest advantages of our adoption of agile is that it has given much greater focus on users, along with a structure for keeping the business partners actively engaged. These activities require a range of non-technical skills and are just as vital to the success of the project as the technical skills. They aren't the only activities the project manager and business analyst bring to the party, as their traditional skills remain important, but these engagement tasks are vital.
It would be disastrous to lose any of these skills from the team. If the developers just write spaghetti code, no amount of management wizardry will save the project. Equally, if the customers and users aren't engaged, we will do little better than an old-fashioned waterfall project. Software teams really do have to be teams. I can see a risk if people start to think that agile just means the rituals without the content, so I will take Mike's article as a warning of what can go wrong.
There is another thread to Mike's article, which is a mistrust of estimation. Indeed, he includes a large picture of Dr Evil saying, "We'll ask for estimates and then treat them as deadlines", which certainly rings true. But the requirement to report progress is a genuine need. You wouldn't pay someone to build a house without some idea of when it will be finished. Someone who is paying for software also wants some indication of when they can expect to start using it. This is a hard topic and one I'll leave for another time.
I can see that Mike's scenario would be problematic. I have seen examples in our organisation where some non-development staff may have thought they could control an agile team (e.g. as a business analyst) but the structure of the agile team has (correctly) worked against them. In general, it is certainly easier to explain daily stand-up meetings and two-week iterations to non-technical people than it is to educate them in the ways of continuous integration and refactoring. But that has been OK for us, because the non-technical members of the teams have other responsibilities.
Perhaps it has helped that we don't follow the Scrum process per se and our teams don't have a Scrum Master. There is no one person in charge of our teams. We do have a project manager who contributes to the team but they are not in charge. They are also outnumbered: a typical agile team in Applications Division has two developers, a half-time project manager and a half-time business analyst (who may sometimes be the same person). The business partners will contribute someone working half-time as a business lead and tester. (The exact composition of the team will vary depending on the exact nature of the project).
We have found that the project manager and/or business analyst have very important roles managing the customer liaison and testing, and setting up user representation. In fact, from my point of view, one of the biggest advantages of our adoption of agile is that it has given much greater focus on users, along with a structure for keeping the business partners actively engaged. These activities require a range of non-technical skills and are just as vital to the success of the project as the technical skills. They aren't the only activities the project manager and business analyst bring to the party, as their traditional skills remain important, but these engagement tasks are vital.
It would be disastrous to lose any of these skills from the team. If the developers just write spaghetti code, no amount of management wizardry will save the project. Equally, if the customers and users aren't engaged, we will do little better than an old-fashioned waterfall project. Software teams really do have to be teams. I can see a risk if people start to think that agile just means the rituals without the content, so I will take Mike's article as a warning of what can go wrong.
There is another thread to Mike's article, which is a mistrust of estimation. Indeed, he includes a large picture of Dr Evil saying, "We'll ask for estimates and then treat them as deadlines", which certainly rings true. But the requirement to report progress is a genuine need. You wouldn't pay someone to build a house without some idea of when it will be finished. Someone who is paying for software also wants some indication of when they can expect to start using it. This is a hard topic and one I'll leave for another time.
Comments