Read on Twitter
There are 3 critical logical fallacies that people make when they think about estimates. And they are quite easy to debunk too! A #NoEstimates Thread...🧵
The First: people think "better" estimates get you more predictability. If this were true, all transportation systems in the world would spend MILLIONS on estimators! Instead, what they do: measure, repeat.
They measure past performance, and assume similar future performance. A great example of this is the drawing of bus/train/air traffic timetables.
In #Agile software development, this can be easily done by measuring cycle time for Epics/Features/Stories, and using that to plan!
In #Agile software development, this can be easily done by measuring cycle time for Epics/Features/Stories, and using that to plan!
The Second: people think that estimation can help remove the problems that come with dependencies, for example, that estimation can help plan better dependencies. Instead, the more you estimate the more errors you build into the planning, leading to catastrophic errors!
The problem is that estimation errors grow exponentially in a network of dependencies, and estimation errors are never "recovered" by being faster with other work. This is obvious because dependencies are one sided (B needs A, not the other way around), and ...
... no matter how much faster you are with C, if you delivered A late, the rest of the deliveries that depend on A, and B will be late, and propagate that error further down the network of dependencies!
The Third: people assume you can plan "everything" (otherwise estimates make no sense whatsoever!). Instead, we never know what "everything" is until we start delivering.
So, in fact, delivering something is more important than knowing what is "everything" that we need to deliver!
In short: there's no logical foundation to the use of estimation! #NoEstimates (read the whole thread)
• • •
Missing some Tweet in this thread? You can try to
force a refresh
