There are two essential components to an effective “Green” OA mandate (i.e., one that generates as close to 100% compliance, as soon as possible):
(1) The mandate must uncouple the date of deposit from the date the deposit is made OA, requiring immediatedeposit, with no exemptions or exceptions. How long an OA embargo it allows is a separate matter, but on no account must date of deposit be allowed to be contingent on publisher OA embargoes.
This is exactly what the New HEFCE policy for open access in the post-2014 Research Excellence Framework has done.
(2) Eligibility for research assessment (and funding) must be made conditional on immediate-deposit (date-stamped by the journal acceptance letter). Again, this is in order to ensure that deposits are not made months or years after publication: no retrospective deposit
The deposit requirement for eligibility for research assessment and funding is not itself an OA requirement, it is merely a procedural requirement: For eligibility, papers must be deposited in the institutional repository immediately upon acceptance for publication. Late deposits are not eligible for consideration. 

This engages each university (always extremely anxious to comply fully with REF, HEFCE and RCUK eligibility rules) in ensuring that deposit is timely, with the help of the date-stamped acceptance letter throughout the entire 6-year REF cycle, 2014-2020. 

These two conditions are what have yielded the most effective of all the Green OA mandates to date (well over 80% compliance rate and growing) at University of Liege and FRS-FNRS (the Belgian Francophone research funding council); other mandates are upgrading to this mandate model; Harvard FAS has already adopted immediate-deposit as one of its conditions. And now RCUK— thanks to HEFCE/REF — will reap the benefits of the immediate-deposit condition as well (see ROARMAP)

OA embargoes are another matter, and HEFCE/REF is wisely leaving that to others (RCUK, EU Horizon2020, and university mandates) to stipulate maximal allowable embargo length and any allowable exceptions. What HEFCE/REF is providing is the crucial two components for ensuring that the mandate will succeed: (1) immediate deposit as a (2) condition for REF-eligibility.

But let me add something else that will become increasingly important, once the HEFCE/REF immediate-deposit requirement begins to propagate worldwide (as I am now confident it will: UK is at last back in the lead on OA again, instead of odd-man-out, as it has been since Finch):

The immediate-deposit clause and the contingency on eligibility for research assessment and funding also ensures that the primary locus of deposit will be the institutional repository rather than institution-external repositories. (Deposits can be exported automatically to external repositories, once deposited and once the embargo has elapsed; they can also be imported from extrenal repositories, in the case of the physicists and mathematicians who have already been faithfully depositing in Arxiv for two decades,) 

But besides all that, many of the eprints and dspace institutional repositories already have — and, with the HEFCE mandate model propagating almost all of them will soon have the email-eprint-request Button:

This Button makes it possible for users who reach a closed access deposit to click once to request a copy for research purposes; the repository software emails an automatic eprint request to the author, who can click once to comply with the request; the repository software emails the requestor the eprint. (Researchers have been requesting and sending reprints by mail — and lately by email — for decades, but with immediate-deposit and the Button, this is greatly accelerated and facilitated. So even during any allowable embargo period, the Button will enhance access and usage dramatically. I also predict that immediate-deposit and the Button will greatly hasten the inevitable and well-deserved demise of publisher OA embargoes.)

Sale, A., Couture, M., Rodrigues, E., Carr, L. and Harnad, S. (2014) Open Access Mandates and the "Fair Dealing" Button. In:Dynamic Fair Dealing: Creating Canadian Culture Online (Rosemary J. Coombe & Darren Wershler, Eds.) 

Let me close by noting another important feature of the new HEFCE/REF policy: The allowable exceptions do not apply to the immediate-deposit requirement! They only apply to the allowable open-access embargo. To be eligible for REF2020, a paper must have been deposited immediately upon acceptance for publication (with a 3-month grace period).

(No worries about HEFCE's optional 2 year start-up grace period either: Institutions will almost certainly want their REF procedures safely and systematically in place as early as possible, so everything can go simply and smoothly and there is no risk of papers being ineligible.)



Postscript. Expect the usual complaints from the usual suspects:
(i) "This is a sell-out of OA! It's just Green Gratis OA, not Libre OA: What about the re-use rights? And if it’s embargoed, it isn’t even Green OA!"
Reply: Relax. Patience. A compromise was needed, to break the log-jam between the Finch/Wellcome Fool’s-Gold profligacy and publisher embargoes on Green OA. The HEFCE immediate-deposit compromise is what will break up that log jam, and it’s not only the fastest and surest (and cheapest) way to get to 100% Green Gratis OA, but also the fastest, surest and cheapest way to get from Green Gratis OA to Libre Fair-Gold OA.
(ii) "This is a sell-out to publishers and their embargoes."
Reply: Quite the opposite. It will immediately detoxify embargoes (thanks to the Button) and at the same time plant the seeds for their speedy extinction, by depriving publishers of the power to delay access-provision with their embargoes. It is also moots the worries of the most timorous or pedantic IP lawyer.

It thereby provides a mandate model that any funder or institution can adopt, irrespective of how it elects to deal with publisher OA embargoes.

And a mandate that can be simply and effectively implemented and monitored by institutions to ensure compliance.

***********************************************
* You are subscribed to the SERIALST listserv (Serials in Libraries discussion forum)
* To unsubscribe, send an email to the server address: LISTSERV@LIST.UVM.EDU . Do NOT include a subject line. Type as an email message these two words: SIGNOFF SERIALST
* For additional information, see SERIALST Scope, Purpose and Usage Guidelines.
***********************************************