The Texas Sharpshooter Fallacy && Retrospective coherence

Do pay attention to The Texas Sharpshooter Fallacy when looking for Retrospective coherence to avoid placing artificial order over natural random chance.

When you do the Retrospective meeting, when you do the root cause analysis, when you try to recognize patterns and search for meanings while looking at data collected from your teams feedback loops, when you try to generalize what worked in your team and you try to replicate it in other teams.

Additional remarks and random thoughts here.

It is not an easy matter because Agile uses and is based on the empirical approach applied in a specific context (the customer, his market, the project, the team, the organization, the technology, ...). This is what it means when you read that Agile is not Silver-bullet, but rather a framework.

What works and what does not in an Agile team is strongly related and very sensitive of the specific context. So it can be very hard to prove empirical evidences, emerged within a single Agile team during a project, with controlled experiments and theoretical studies.  It is more pratical to prove them with natural experiments, observational studies, and field studies that on the other hand are more subject to the The Texas Sharpshooter Fallacy.

Alan Turing says that conjectures are of great importance since they suggest useful lines of research. Our brain is built to work with uncertainty and incomplete informations and follows first fit strategy.

So evaluating empirical evidences emerged within a single Agile team is exactly the context where our best insights and intuitions and creativity together with many small frequent iterations and inspections can in first place be more effective then controlled experiments, theoretical studies and analytical methods (i.e. think of evolutionary and genetic algorithms).
When we do this is important to be aware of The Texas Sharpshooter Fallacy and like a scientist to always be very sceptical about hypothesis and like scientists try to disprove them in order to test them.

Print | posted @ domenica 12 settembre 2010 23:37

Comments have been closed on this topic.