What a packaging standard was actually meant to solve

E-learning standards exist because organisations discovered they had bought content they could not move. Everything awkward about them follows from that original problem.

In the late nineties an organisation that bought training content discovered an unpleasant fact. The content ran inside the vendor's delivery system and could not be moved out of it. Change the delivery system and you lost the library. Buy from two vendors and you ran two systems. The content was not really yours in any practical sense, because the thing that made it work was the coupling between the material and the software that presented it, and that coupling belonged to somebody else.

The response was a set of agreements about how a package of learning content should be laid out, what file describes its contents, and how the content talks to whatever system is showing it. If everybody follows the agreement, a package built in one authoring tool can be dropped into any conforming delivery system and will run, report completion, and record a score. That is the whole ambition. It is a modest one, and it was and remains genuinely valuable, because it converts content from a rental into an asset.

It is worth being precise about what the agreement does not cover, because most disappointment comes from expecting more. It does not standardise how the content looks, how it is built, or how good it is. It does not make two courses comparable. It does not describe the learning design. It is closer to a shipping container than to a curriculum: a fixed set of dimensions and fittings that lets an arbitrary cargo be moved by an arbitrary carrier, with no opinion whatsoever about what is inside.

The container analogy also explains the standard's characteristic failure mode. A container that is a centimetre too wide does not fit, and it does not half fit. Packaging standards are similarly unforgiving: a manifest with a malformed reference does not import partially, it fails. This is frustrating in the moment and it is the point. A tolerant standard that accepts near-enough packages produces systems that behave differently from each other, which destroys the only guarantee the standard was created to provide.

The other thing worth understanding is the data model. The standard defines a small vocabulary that content can use to tell the delivery system about the learner's progress: whether they completed, what score they achieved, how long they spent, where they were when they left. It is a deliberately narrow vocabulary, and its narrowness is why so much effort in this field goes into working around it. If the question you want to answer is not expressible in those terms, the standard will not help you, and a great deal of ingenuity has been spent encoding richer meaning into fields that were designed for something else.

That pressure is what produced the newer specifications, which abandon the idea of a fixed vocabulary in favour of recording statements about what somebody did. The trade is straightforward and often misrepresented. The older approach guarantees that any conforming system understands the data, at the cost of the data being thin. The newer approach allows arbitrarily rich data, at the cost of nobody being obliged to understand it. Neither is a successor to the other in the way vendors sometimes imply. They answer different questions and a great many organisations sensibly run both.

For the person whose job is to make a package work on a Tuesday afternoon, the history matters less than one conclusion drawn from it. These specifications were written to protect the buyer from the vendor, and they only do that if the buyer checks. A package that has never been imported into anything except the tool that produced it has not been tested against the agreement, and the guarantee it was supposed to provide has not actually been obtained.

One further point about the shipping container comparison, since it clarifies what these agreements are worth in practice. Containers did not make cargo better, cheaper to produce, or more useful. They made it movable, and the movability turned out to be worth more than almost any improvement to the cargo itself, because it changed who could buy from whom. Packaging standards have done the same modest and consequential thing for learning content. An organisation that can move its library between suppliers negotiates differently from one that cannot, and that leverage is the return on every hour spent on conformance, even in the years when nothing moves.