As a developers, techleads, teamleads, and architects we often provide different kinds of estimates: for sprint tasks, release planning, new requirements, project ideas, etc. But are they really helpful? Sounds a bit provocative, but that's what video Estimates or NoEstimates is about.
The author starts with the investigation of the real meaning for the term and comes up with a conclusion that estimates are just a guesses about uncertainty:
An estimate is a guess of the amount of time [work time or elapsed time] to create a project, a feature or some bit of work in developing software
An interesting note here is that we should remember that task effort is not the same as task duration:
- Effort (Work time): the amount of time to do the actual work, not including time waiting
- Duration (Elapsed Time): the amount of time that passes from when we first start on some work until it's completed (this time includes interruptions, clarifications, waiting for some resources, etc.)
If estimates are just guesses, why do we need them? The answer is that they help us make decisions:
- Should we do project A or project B?
- Should we do this project or feature?
- Do we need to hire more people?
- Do we need to fire some people?
So estimates give us a felling of control. But is there any control?
The author shares a funny story to illustrate how estimates are really made. When a manager asked him to provide estimates, he asked his more experienced colleagues how to do it. The answers were like "I just make up some numbers really quick because I'm not going to waste my time on that", "I just write down a bunch of of small numbers because he [the manager] seems to love small numbers", "I'll quickly look through each of the requirement and try to think about something I did before and how long it took me and I write that number down". As a result the author "quickly wrote down a bunch of numbers that were high" and gave them to the manager. This type of information is what we base our plans and decisions on 😃
So the main idea is that we cannot rely on estimations to make serious decisions. Moreover, there is no sense to use estimations accuracy as a metric for project success or team effectiveness. Instead of that we should build innovative and creative environment to promote culture for better software creation (where estimates are not needed probably).
The talk is really nice, it makes you rethink some everyday routines, but it's more about problem statement then about particular solution (there is also a hashtag #NoEstimates to follow in X to read more). As for me I don't think we can get rid of estimations in the near future but it's good to remember what estimations really are and not heavily rely on them for important decisions.
Also I revise how I do estimations. Typically I use scale like 1, 3, 5, 10, 15, 20, 30, 50, 100+ to quickly define the size of the task. The more uncertainty in the task, the higher my estimates 😉
#management #engineering