Teaching Time versus Teaching Times?

[psteinb]

Teaching Time versus Teaching Times?

[gcapes]

Ah, the timing question!

I would say is that it's important to set realistic timings, which can only really be arrived at by teaching the material a few times and recording how long it took. As far as I know there's no mechanism to record this easily. However, unrealistically short timings are prevalent throughout the carpentries' lesson, but this is demotivating for both instructors and learners.

I brought this up for the shell lesson swcarpentry/shell-novice#928 and we had a few timings reported as comments on the issue.

[psteinb]

So I guess, my question is what to do about this? Is this something being discussed among other instructors and lessons?

[psteinb]

Great comment @gcapes which I totally agree with. And I believe you put your finger right where it hurts: if the frontmatter lists a time which is unreasonable, this is a problem for me. So, how about two radical ideas:

• only include those times in the front matter which are based on at least 10 timing measurements (e.g. their mean/median value)

• rename these two fields:

---
title: "Classification by a Neural Network using Keras"
likely_teaching: 30
likely_exercises: 20

We could and should discuss facilities in the lesson template which support people to record their timings (and submit them through a PR). Because only this view of a bunch of numbers (or summary statistics thereof) can provide me with a feeling of how long this module normally takes to teach. And normal here not only refers to the name of a distribution, but also to conveying and conceiving an expectation.