[dragonwriter]

> The reality is that the sunk costs of your existing codebase will almost always outweigh any advantage of trying a new language.

Sunk costs are an irrelevancy, relative future costs are what matters. Transition costs, especially for a mature product, may sometimes be an issue, especially if for some reason you are blocked from doing an incremental change.

And obviously neither irrelevant sunk costs nor relevant transition costs apply to a greenfield project.

[jacquesm]

This is a very deep insight but it does rely on people knowing what the relative future costs are and this is a much harder problem than looking back and so the part of the equation with the most evidence behind it tends to be given the greater weight.

[jqpabc123]

What was described is standard economic practice --- but people violate it all the time. They retain an emotional attachment to money that is long gone.

In any case, the future is a judgment but the past is history --- and throwing more money at it won't change it or bring it back.