Shared group tools
==================

* In less-well-organized computational groups, it's often the case that
  everyone is developing their own code, and the same things are
  re-implemented over and over.

* Why re-implementing is good:

  * In some sense, researchers do need to do new things, and starting
    from scratch is easier.

  * Also, by creating your own things, you have more opportunities to
    learn.

* But on the other hand...

  * If everyone is re-implementing themselves, how does someone learn
    how to do that well?  Do you see good examples to learn from -
    that's the best way to learn.

* Can you create some baseline tools, such as code libraries,
  datasets, or procedures, that bind people together and provide a
  incentive for people to learn from each other?

* You might need some :doc:`research engineers <research-engineer>` in
  order to take the most advantage of this - without someone watching
  over your tools, it might turn to unusable chaos!