The rules that protect a course students are on

Three rules apply to everything a connection writes. They are not settings and cannot be switched off. Each one exists because of a specific way automated authoring goes wrong.

1. A new course arrives inactive

Any course a connection creates is inactive, so nobody can enroll on it. Making it available is a person’s decision, taken after reading it.

Why. The failure this prevents is a course appearing in your catalog, being found by a learner and being started, before anybody who knows the subject has read a word of it.

What to do. Read it, then activate it yourself — see Who may enroll, and how.

2. A lesson a student has read cannot be changed

Once any student has opened a lesson, a connection cannot alter it. Not the words, not the questions, not the order. This does not expire.

Why. Somebody has studied that material and may be assessed on it. Changing it underneath them means they were taught one thing and tested on another, and nothing on screen would tell them that had happened.

“Read” means opened. Somebody who opened a lesson and closed it again counts the same as somebody who studied it for an hour. That is deliberate: either it has been seen or it has not, and that is a question anybody can check. A rule with a time limit — “frozen for six months” — would mean a lesson is protected on Monday and editable on Tuesday because nothing happened, which is a rule nobody can predict.

What to do. Copy the course. A copy carries all the material and none of the results, so nothing in it is frozen and every lesson can be edited. The original keeps its learners and their work. Teach the new version to the next group. This is the intended route and it is why asking for the copy permission is worth doing — see Copying a course.

You can also still edit the lesson yourself, on screen. The rule applies to connections, not to people: you are expected to know what you are doing to whom.

3. A change to a question is flagged back to you

When a connection creates or changes an assessment question, the reply says so, separately and explicitly, under a heading that means a person needs to look at this.

Why. This is the failure mode with the longest delay between cause and discovery. An assistant writes a plausible question with the wrong answer set as the correct one; it is graded automatically; learners lose points for being right. Nobody finds out until somebody complains, and by then there are results built on it.

This is a flag, not a block. The question is written, and it is live if the course is live. Grapevine tells you clearly; it does not hold the question back. If a new lesson containing questions is added to a course that is already running, those questions can reach learners before you have read them. If that matters for your course, add lessons with questions yourself.

And underneath all three: nothing is deleted

No connection can delete anything. If you ask an assistant to remove a module and it submits an outline that leaves that module out, the module is not removed — it stays where it is, and the reply lists it as something left alone.

This occasionally surprises people who expected the outline they submitted to be the whole truth about the course. It is the right way round: an accidental omission is then a report, not a loss.