kilabit.info
Build | sr.ht | GitHub | Twitter

Before I am going to rant about Story Points (SP), I am going to tell a story how we build software before we know Scrum or Agile as a consultant.

Once we have a project, we meet the client. Most of the software that we build usually translate the real world system (using form and manual input) into an application. The documentation is usually about how to use the software or the design behind it (database schemas, and so on).

Most of discussion usually around how to fill the form, what kind of values is allowed in this field, who can use fill the form, what did the client expect from the application, and so on. If we are lucky, client know what the input and output should look likes, these make us easy to map and separate the process into modules. Once we can draw the line between modules we can calculate how much the cost to build it. We propose the cost and time to client, of course with additional unpredictable cost and time in our mind, client bargaining and then we make a deal.

Later, we finish one module, go back to client, do the test, gather feedbacks, and goes back to home. One person usually handling the feedbacks, enhancements and bug fixes; while others continue developing the next module.

Repeat this until we finish our project.

I don’t know what kind of methodology that we use back then, but none of us use story points. What the client care, and we too, is how much they pay and when it could be finished; both our goals is to finish this project, a software, hopefully, in the time we have agreed upon. If we can complete it before the time, we can move it into another project, or take the free time, both are happy. If we cannot complete it before the time, both of us will looks bad. From our perspective, we looks like incompetent, from client perspective they may have to delay the integration or report to higher command and explain why why their project is still not finished, which looks bad on them and us.

Around 2014 I go back to works in the office, a start up. I may have heard about Agile before, but this is my first experience to works with sprint, user story, story points, epic, Jira, with all the bells and whistles. This is also the time where I experiences cloud, containers, new programming language, and all the shiny things in software development, so I did not pay attention to the workflow. In fact I can not remember if we use story points or not. I just buried with new experiences and excitements, so everything that I confronts seems like new to me and seems alright. The only things that I can remember and reflect today is that we may have over-engineered at some points (rewrite and refactor), at least we stick to some goals; but that is just another story.

At the next company, is the place where I have bad experiences. I will focus on how we use story points, beside other bad practices.

First, we associate a story point to one day. If you read any journal or articles about scrum or agile, this is a no-no. I let it go at that times.

Second, we waste the times about not filling and updating the story points over empty story description. We have three fields for story points, one for "Original estimated points", one for "Story points", and another one for "completed points current sprint". There is one field for QA called "Completed QA Points". I am not kidding you.

Third, they report the progress as a completed story points in one sprints. What could go wrong?

                   sprint ended
                        |
                        V
o-----------------------o
^                   ^
|                   |
sprint          implementation
started         finished

Since each developer must have X points = Y days to fill in one sprint, at the end of sprints we may have some un-finished story (due to other bugs or unrelated tasks) or finished implementation but not yet tested. So, at the end of the sprint we never reached 100%, obviously, and this makes us looks bad.

And somehow, the "product owner" always questioning, why?

Fourth, they use this points to base our performance. Not based on outcomes, but points.

People over process, yes?

What is the point of SP?

The problem on how to estimate the software project, I believe, has been researched long enough. If my memory served right, there is some formula that we taught in school on how to estimate based on number of functions. Some, estimates it based on number of tables in database. Estimation is small part of the whole picture, which is, how to breaking down a system so we can design and built an application.

In fact, since couple last years, there is a movement called #NoEstimates. I hope that movement ripple through here, like how we got scrum.

I have a bug. From the error logs, I believe it will take story points 0.5 (half a day). The fix is around 16 lines. To replicate the bugs, I need to refactor the code to make it testable, which cost 6-7 independent commits and takes 2 days.

The way I do estimation is by breaking down it into small pieces that I can work in three day or less. If a story takes more than three day, or require more than two peoples, it may need to move into an epic.

If I am CEO of software system or user of application, I did not care about story points. What I really care is a working application, satisfying my needs, and run without any issues; a usable application.