Harvard's OA Policy and the Hybrid Copyright Retention and Deposit Mandate

---

SUMMARY: My OA comrade-at-arms, Peter Suber, in an OA News posting that I quote and comment upon below, compares Harvard's (CMo) Copyright-Retention Mandate with opt-out and (DMn) an Immediate Deposit Mandate with no-opt-out.

The amendment of the Harvard draft policy that I am urging, however, is not to substitute DMn for CMo but to upgrade CMo to CMo+DMn so as to jointly mandate BOTH CMo AND DMn (the former with an opt-out option, the latter without).

That way the Harvard policy will not only capture (1) all the deposits that would have been made by authors who did not opt out of the CMo clause but also (2) all the deposits that would otherwise not have been made at all, because the authors had opted out of the CMo-only mandate.

The amended mandate loses nothing at all of what would have been generated by the original Harvard mandate alone, it only adds further gains, while still allowing opt-out from copyright retention.

---

Peter Suber: I'm not saying that the Harvard-style mandate will generate a higher level of compliance and OA than a dual deposit-release style mandate (or what Stevan calls immediate deposit/optional access). I think this remains to be seen.

I agree that whether it would be CMo or DMn that generated a higher level of compliance and of OA would remain to be seen (after Harvard's proposed CMo trial period of three years -- although we do already know from Arthur Sale's data that DMn approaches 80-100% within two years).

But that is not what I propose. I propose adding the Deposit Mandate (with no opt-out) (DMn) to Harvard's current Copyright Retention Mandate (CMo) (with opt-out, exactly as it already is now):

Not CMo vs. DMn, but CMo vs. CMo+DMn.

That will insure the Harvard mandate against the possibility of another needless loss of three years of research usage and impact (as happened with the unsuccessful first version of the NIH OA policy, until it was upgraded, three years later). And as the Harvard policy is likely to have many emulators worldwide during the three-year trial period (as the NIH policy did), it is especially important to give them them the right model to follow.

Peter Suber: At Harvard, faculty will give the institution permission to host copies of their articles in the institutional repository, when they don't opt out, and Harvard will take responsibility for making the actual deposits.

There are two components here: (1) the CMo mandate itself, with its all-important question of how great its opt-out rate will be, plus (2) an implementational detail: Harvard engages to do the deposit on the author's behalf (this is elsewhere called mediated or "proxy" deposit).

There are things to be said for and against the implementational detail (mediated deposit): Yes, it's easier for an author if all he needs to do is pass his digital paper to someone else to deposit it for him. Many repositories have tried that, with a certain amount of success.

But authors depositing for themselves (or their students or assistants doing it for them) is a more direct way of doing it then putting the paper in a central queue, and it has some perks of its own: potentially faster turnaround time, more direct author control, and, perhaps most important, assimilation into the author's standard work-flow for drafting and publishing a paper.

In reality, the deposit only takes a few extra minutes' worth of keystrokes per paper, over and above all the keystrokes already performed in writing it and submitting it for publication. Direct author deposit moots the risk of delay in waiting for indirect deposit mediated by a third party (neither the author nor the journal).

The jury is still out on which of the two ways of implementing deposit is the more reliable and efficient procedure (but my guess is that it will prove to be direct author deposit, just as it did with word-processing, email, and online searching).

Peter Suber: If Harvard is fleet and efficient in making these deposits, then it could provide immediate OA to all the articles not subject to opt-outs (which, for the sake of argument, let's peg at 95% of the total).

The implementational detail (of direct vs mediated deposit) can be set aside as an open empirical question.

It does not seem to me to be sufficient, however, to assume, for the sake of argument, a 95% non-opt-out rate -- especially in weighing CMo vs. CMo+DMn.

If, three years hence, the non-opt-out rate turned out to have been 75%, 50% or 25% (let alone 4%, as in the case of the failed first policy of NIH), the implications concerning whether it would have been better to adopt CMo or CMo+DMn from the outset would be very different.

So everything rides on what the opt-out rate for CMo would actually turn out to be. And no one has any idea, as this would be the first CMo mandate.

But since the comparison is CMo vs. CMo+DMn (rather than CMo vs. DMn), it is not clear that we even need to know how the deposit rate and the OA rate for CMo alone would have compared with the OA rate for DMn alone.

We do already know from Arthur Sale's empirical analyses as well as the growth data on the other mandated IRs, that the compliance rate for DMn mandates approaches 100% within 2 years.

So with CMo+DMn we know we would have at least 80-100% deposits -- plus whatever deposits CMo alone would have generated, over and above that. (There is no need to know or assume how many that would have been: CNo+DMn would inherit them all!)

Peter Suber: If Harvard is slow to make these deposits, then it will miss a beautiful opportunity arising from the permissions it will have in hand, and may as well defer to publisher embargoes.

Peter is expressing concern here about the possible delay in mediated deposit (I share that concern) and comparing that delay to the delay arising from publisher embargoes.

But a far bigger worry is the opt-out rate, hence the size of the loss in deposits (both the OA deposits and Closed Access, almost-OA deposits) with CMo alone.

(As it happens, my proposed amendment, CMo+DMn, did include an implementational suggestion too, one that would help remedy this: The Harvard Mandate does not currently stipulate how the paper is to be provided to the Provost's office for mediated deposit. One potential way is by email. But another is by the author depositing it directly in a Harvard Institutional Repository and then just emailing the Provost's office the URL plus the permission. That would cover all those cases where the author does not opt-out of the CMo component of CMo+DMn -- as well as all the direct deposits that opt out of the CMo clause but comply with the DMn clause.)

So even in cases where the author elects to opt out of the CMo clause of CMo+DMn, there would still be the binding DMn clause, requiring the deposit in the Institutional Repository; and hence there would be the resultant OA and almost-OA that that in turn generates. (The CMo+DMn's no-opt-out DMn clause alone would contribute 80-100% deposits within two years.)

Peter Suber: Likewise, if opt-outs are rare, the level of OA could approach 100%; if they are common, it would be much lower.

We indeed have no idea what the opt-out rates of a CMo mandate would be, with or without mediated deposit. If opt-out is low, it's low, if it's high, it's high. Unlike with DMn, we do not know in advance what it will be.

(But we do know that another prominent policy with an opt-out -- the initial NIH non-mandate -- failed dramatically until the opt-out was removed by upgrading it to a mandate -- but only after the attendant loss of another three years of research access, usage and impact. Hence the pre-emptive strategy for Harvard's proposed CMo mandate, to avoid the needless extra risk during the three-year trial period, would be to upgrade to CMo+DMn from the outset.)

Peter Suber: By contrast, under a dual deposit-release strategy, Harvard could have 100% of the articles on deposit. But a good percentage of them would be subject to publisher embargoes and about one-third of them would be subject to flat publisher prohibitions on OA archiving.

First, we must again recall that we are talking here about adding the no-opt-out DMn clause to CMo, not about replacing CMo with DMn.

The current publisher policy figures are the following: 62% of journals endorse making postprint deposits OA immediately (and an additional 29% for preprints); about 38% only endorse making postprint deposits OA after an embargo period -- which ranges from 6 months to forever.

(Let's round off the immediate/embargo ratio at 60% of deposits set as immediate OA and the remaining 40% of deposits set as "Closed Access," but with almost-OA provided for that 40% during the embargo by means of the semi-automatic "email eprint request" Button.)

Now here is the crucial point: Peter is here comparing the deposit rate (100%) and OA rate (60%) for a no-opt-out DMn-only mandate with the corresponding deposit and OA rate for a CMo-only mandate allowing opt-out (in the latter case both the deposit and OA percentages being, identically, the unknown CMo non-opt-out percentage: XX%).

But the relevant comparison for my proposed amendment would be this deposit percentage and OA percentage (both identically XX%) for CMo-alone versus the total deposit and OA percentages for CMo+DMn:

Those total CMo+DMn percentages are, respectively, XX% + 100% = 100% and XX% + 60% (which is at least 60%, plus whatever the non-opt-out copyright retention rate is for the articles that are not already part of that 60%).

In other words, compared to CMo alone, there would be no loss in copyright retention and only potential gain in OA with CMo+DMn.

And let's not forget that, with 100% deposit, there will also be systematic almost-immediate, almost-OA for all the non-OA deposits (thanks to the semi-automatic "email eprint request" Button). That too is a pure gain, with no accompanying loss, compared to the current CMo-only policy.

Peter Suber: Hence, each type of policy risks delayed OA (one through institutional sluggishness, one through publisher embargoes), and each risks incomplete OA (one through faculty opt outs and one through deference to publisher policies).

Peter is here comparing (1) delay because of mediated deposit arising from CMo alone with (2) delay because of publisher embargoes arising DMn alone.

But for the proposed amendment, the pertinent comparison would be between CMo alone and CMo+DMn. And that comparison shows that there would be no losses and only gains, if CMo were upgraded to CMo+DMn, whether we reckon in terms of opt-outs, deposits, OA, almost-OA or delays.

Here's another way to put this same point. The comparison between CMo and CMo+DMn must take into consideration:

(1) Green publishers that endorse immediate author postprint self-archiving but do not agree to rights-retention (i.e., they insist on exclusive rights transfer despite the pressure from CMo mandates like Harvard's): Authors who would not want to renounce those journals, would opt out of CMo. Hence no deposit and no OA, even for those Green journals.

(2) Embargoed or Gray publishers that don't agree to rights retention (despite the pressure from CMo mandates like Harvard's): Here again, authors who don't want to renounce those journals opt out of CMo. Hence no deposit and no almost-OA.

Hence it is essential to include in one's reckoning the (1) unknown Green opt-outs and their resulting non-deposits and OA loss, plus the (2) unknown Pale-Green and Gray opt-outs and their resulting non-deposits and loss of almost-OA. Both of these would arise from opt-out because of unwillingness or failure to renegotiate copyright as required by CMo.

(Harvard's speed in doing mediated deposit, though salient too, is not, I think, nearly as critical as the risk of loss from (1) and (2).)

Peter Suber: If the two key variables -- speed of deposit and level of opt-outs-- work in Harvard's favor, then its policy could be better than a dual deposit-release strategy. But if not, not. Because this is contingent, I can't recommend one type of policy over the other without knowing more about the probabilities.

Again, this empirical uncertainty is only pertinent to a comparison between the CMo policy and the DMn policy.

But the comparison I am recommending is between the CMo policy and the CMo+DMn policy. There we can already calculate apriori that CMo+DMn can only do better, not worse, than CMo alone, on all points of comparson: deposits, delays, opt-outs, OA, and almost-OA.

Peter Suber: Because Harvard's is the first university-level mandate to focus on permissions rather than deposits, it deserves a chance to show how well it can work.

Yes it does. But adopting CMo+DMn instead of CMo alone will not diminish that chance one bit: It will just add additional, independent likelihood of success to it.

Or, to put it another way, CMo+DMn makes it possible to experiment with a Copyright Mandate without any need to sacrifice the already demonstrated success and benefits of a Deposit Mandate in exchange.

Peter Suber: Can the two types of policy be blended, so that Harvard faculty give permissions (subject to an opt-out) and make deposits (not subject an opt-out)? Yes. But if Harvard is fleet and efficient in making the deposits, that won't be necessary.

The degree to which CMo+DMn would generate (1) more deposits, (2) more OA and (3) more almost-OA than CMo alone depends not on Harvard's speed of mediating deposit but on what the (unknown) opt-out rate for CMo alone would prove to be.

CMo+DMn is accordingly an insurance policy. Whether an insurance policy was indeed necessary is only known after the insured period has elapsed. But in this case, since the insurance policy would cost nothing but a consensus from Harvard on upgrading to CMo+DMn, I think Harvard has nothing to lose and everything to gain if it upgrades now, from the outset, rather than waiting for the three year trial period to elapse, and another three years of research usage and impact to be needlessly lost, as NIH did.

---

The following is not from Peter's blog posting but from an off-line email exchange (reproduced with permission):

Peter Suber (email): "I'd have no objection to the addition of [a DMn] mandate (indeed, I'd be among the first to call for the addition of such a mandate) if it would improve upon the procedures Harvard puts in place to make the deposits itself."

CMo+DMn would not only improve on the implementation of the submission and deposit process, but -- incomparably more important -- it would improve on the deposit rate, the OA rate, and the almost-OA rate! All the gains would be from the (unknown) percentage of authors who would opt out of the present CMo-only mandate. And CMo+DMn would not lose anything from what the present CMo-only mandate would generate from the authors (XX%) who would not opt out.

Peter Suber (email): However, it's certainly possible that if the original resolution had included a deposit mandate, it would not have been adopted by faculty. We'll never know."

Peter, when you give your talk at Harvard's Berkman Center on March 17 (assuming that the CMo policy has not been miraculously upgraded to CMo+DMn in the meanwhile!), would you consider explicitly raising the following three questions:

(i) Was Harvard's successful consensus on adopting CMo reached because of the "C" (Copyright Retention) or the "o" (the option to opt out)?

(ii) Would consensus not have been reached with CMo+DMn (or DMn alone)?

(iii) Having reached consensus on CMo, is it not now possible to reach consensus on upgrading CMo to CMo+DMn?

Because if CMo+DMn was not even considered in the Harvard FAS deliberations -- or if it was not adopted because of concern that a policy without an opt-out would not be complied with -- then it might be helpful to point out to your hosts (1) Alma's Swan's international, cross-discipline author survey results, which found that 95% of researchers report they would comply with an OA self-archiving mandate from their universities or their funders (81% of them willingly); and (2) Arthur Sale's comparative studies on actual author behaviour, which confirmed that the OA self-archiving rates for universities with an OA self-archiving mandate (with no opt-out!) approach 80-100% within 2 years of adoption, whereas universities where deposit is not mandatory languish at deposit rates of c. 15%.

With CMo+DMn Harvard has nothing to lose and everything to gain. Perhaps if this is made sufficiently clear now, an immediate upgrade is still possible, rather than having to wait for a 3-year trial period to decide (as with the first, failed NIH policy).

Stevan Harnad
American Scientist Open Access Forum