A possible solution to defining "widely reviewed"

All,
There are a number of cases where we do not want to list required actions, but want to allow a number of ways to satisfy a criteria. The "Wide Review" criteria is an example of this. It is up to a Working Group to show that a specification that they want to progress has been "widely reviewed". Traditionally, one method they used was to announce on the TR page that the specification is ready and that if you have not yet reviewed and commented upon it, now is the time to do so (a.k.a. "Last Call"). The WG can show "Wide Review" by showing that a number of comments were received, replies were generated and the commenters accepted the replies. But suppose there is no last call, but a series of Heartbeat Working Drafts that carefully update the status section to identify which sections are stable and should be reviewed. The database of processed comments helps show the document is reviewed, but there are questions as to whether the review was "wide", where "wide" mostly means outside the community producing and implementing the specification. Showing reviews outside this community helps to show "wide". Showing the groups with Dependencies and Liaisons have done review also helps show "wide". But, the Process Document needs a definition of "Wide Review" that is testable and such statements as those above are not sufficiently precise.

That leads me to suggest that we use a technique the US Federal Regulations use; that of the "Safe Haven". A "Safe Haven" is a testable criterion that, if met, guarantees that that broad criterion with which it is associated is also met. But, it is possible to also meet the broad criterion without meeting the "Safe Haven" criteria.

How would this work for "Wide Review"? An example "Safe Haven" would be to say a WG meets the "Wide Review" criterion if it: 1) has issued a Last Call and 2) its comment database shows comments from all dependent groups and groups and individuals outside the specification developers and implementers.

But, there would be other ways to satisfy the "Wide Review"; for example, by doing a distributed sequential review whose comment database shows the same wide participation.

Steve Zilles