This bit is self-evident with simple math, corroborated by financial results for companies with mentorship programs (see footnotes in OP).
The simple math goes like this:
More mentorship = Mentored devs share knowledge and cultural influence with the team, multiplying over-all team productivity & code quality.
Once learned and put into regular practice, code quality benefits have compounding effects — rather than a one-time, temporary improvement in code quality, the team reaps ongoing rewards over the lifetime of the mentee’s employment + lifetime of the team’s code + lifetime of employment of team members also lifted by that mentee’s mentorship.
In other words, the existence of high-quality code in the code base breeds more high quality code in the codebase. Mentorship creates a virtuous cycle.
This is like sticking your money in the market and reinvesting earnings. The return compounds over time. Interest from code quality improvements may continue to accrue years after the mentee moves on.
Returns grow in more than one way:
- Mentees teach others and multiply benefits (benefit is multiplied by team size, often 2–7x knowledge capital return)
- Team produces higher quality code with lower maintenance costs and higher productivity during extension (quality breeds quality)
- Higher quality code contains fewer bugs & is easier to extend without introducing new bugs = fewer bugs in prod. Bugs in prod cost 15x more to fix. Mentorship involves more frequent and effective code reviews. Every hour of code review saves 33 hours of maintenance. (data citations in “The Outrageous Cost of Skipping TDD & Code Review”)