Protocols/Protocols Proposal 2022-01

This is a proposal, originally written in first person by Michael Cooper, for how protocols could fit into the WCAG 3 conformance ecosystem.

Background and Proposal

My understanding of the core requirements for protocols are:

1. Protocols participate in the WCAG 3 conformance model,
2. Protocols are used to achieve outcomes when methods are not available. [JF: Protocols are used to define outcomes when methods are not available. (?)]

Details in those requirements are still in discussion, but I understand those to be the core objective.

The word "outcomes" in the second that provides a place for protocols. If a protocol is intended to achieve an accessibility outcome, then it is intended to meet the requirements of one or more WCAG 3 outcome statements. Therefore, I think Protocols support Outcomes in the same way that Methods do, meaning Protocols would be a sibling category alongside Methods.

Thinking of it this way, we see that the Protocol services the Outcome in the same way as a Method does. As I understand the model, a method describes:

1. an action to take,
2. a procedure to test that the action was successfully taken,
3. a way that the test results of the method feed into the scoring algorithm for the Outcome.

These are all attributes we've already agreed Protocols need to have, so I think it is viable to position Protocols alongside Methods. This would allow rating algorithms to be agnostic about what methods or protocols were followed, they just need to receive test results from the method or protocol that are appropriate for the rating algorithm.

Implications

There are several implications of this approach:

Conformance

Because Protocols are treated the same as Methods, they participate in Conformance in the same way. Therefore, it is possible to use Protocols to conform at any conformance level (if a protocol can be defined that is viewed as sufficient for that conformance level). I believe this corresponds to the "single currency" approach that seems to have more support in the group than the "multiple currency" approach.

Multiple Outcomes

It would not be reasonable to require that protocols be separately defined for different Outcomes, as many process-oriented protocols will be aimed at meeting multiple outcomes, and it would be impractical to require tracking multiple similar protocols for different outcomes. Therefore, I think it is a requirement that Protocols can apply to multiple Outcomes. The specifics of how that works will certainly involve debate, but I think we can define a clear scoring mechanism for that situation.

Because I have drawn a parallel between Protocols and Methods, that raises the same question for Methods. My understanding is that we currently are trying to have methods support only single outcomes. While that makes the scoring tree simpler, I question whether we can divide methods up in such a granular manner. Some, yes, all, I'm not sure. So I think we would be raising that question, and need to make a decision either to make Methods like protocols in that respect as well, or to make a deliberate distinction.

Evaluating Inputs

I think we now agree that a protocol can be evaluated by evaluating the inputs, which are the actions that are taken to follow the protocol. Certainly it is possible to evaluate adherence to the protocol, and I think there may be ways to evaluate quality of adherence to the protocol as well (for some, anyways). That gives us a way to have different conformance levels for the protocol.

Realistically, for most Methods I think this is what happens anyways. A method to use a given code pattern can be tested for following that pattern, but the actual accessibility outcome is usually presumed, not directly evaluated (until the Gold level or something).

Choosing Protocols vs Methods

I believe the driving reason to create Protocols is to address accessibility outcomes for which we do not know how to define testable methods. Initially, we will probably look at creating protocols for Outcomes for which we don't have Methods. But inevitably, I believe, there will be Outcomes for which there are both Protocols and Methods available. Sometimes this is because Protocols can support multiple Outcomes, a side effect of which many would want to take advantage. But I think there will be times we decide we want to do both for some Outcomes, maybe because available methods are super esoteric or something.

In this situation, I think many will want a Protocol should have a lower weight in the rating algorithm than a Method would have. While I can see reasons to argue against that, I think we shouldn't fight that battle, and so we should define a satisfactory mechanism for weighting Protocols vs. Methods in scoring.

Everything Else

There are many other questions about how protocols should be defined, evaluated, rated, etc. This proposal does not address any of those questions. This proposal is for how Protocols can fit into the WCAG 3 structure and conformance model. If we can obtain agreement on that, then the group can focus on the harder work of defining what actual protocols might look like.