Knowledge transfer mapping

Rationale

If knowledge management is at the core of software development, software engineers should understand the knowledge transfer and knowledge storage inside an organization or a team as a first step towards optimizations.

Proposed Process

We propose the following process to be tested:

• take as example a recent user story or feature that was recently developed
• list the questions (as many as you remember) that the team had to ask in order to implement the user story or feature
• for each question, document the path from the question to the answer. Include people, documents, internet searches, etc (as much as you can remember)

Ideally use a visual way to show the map.

What to observe on the map

We expect a few things to become apparent on the knowledge transfer map:

• long chains of knowledge transfer (the telephone game effect)
• outdated or misleading knowledge stored in documents or diagrams
• knowledge stored in an obfuscated manner in code
• missing knowledge, requiring either additional documentation or structured knowledge transfer

Once these issues become apparent on the knowledge transfer map, a few optimizations can be attempted, such as:

• bring the knowledge within the team or within the easy reach of the team
• refactor the code to express the knowledge in a clearer manner
• add automated tests to the code to document by examples the behaviours of the code
• add more documentation artefacts related to design, architecture, configuration, deployment, runtime etc.

Inspiration

This process is adapted from value stream mapping. The simplified visual representation of a value stream map can be used to represent knowledge storage and knowledge artifacts.

Testimonials

This process has not been used since it was proposed on 29 June 2020.

Critique

This process might not provide useful information if the feature you are analyzing was something completely out of the ordinary and that you don't expect to do again.

It may be necessary to build the knowledge transfer map by taking into account not only a team but a full organization. This means that it's not enough to discuss with one team, but instead you should map the knowledge transfer outside the team as well. This raises questions regarding to the return on investment of the knowledge transfer mapping. However, at this point, the only way to find out if it's useful is to attempt it.

The most severe criticism is related to the nature of tacit knowledge and organizational knowledge. Since this type of knowledge cannot be transferred, and since it might happen that most knowledge is stored in such a way for a specific feature, the analysis might not reveal anything of substance.