February’s Membership Ticket of the Month comes from Samuel at one of Crossref’s sponsoring organizations, Cultural Hosting.
Samuel noticed that last year we assigned a new sponsored member a DOI prefix in the 10.71000s. However, this year we assigned a different new member a prefix in the 10.63000s. Sponsors and other keen-eyed observers generally see that new prefixes are assigned sequentially, so Samuel wondered why Crossref is suddenly assigning lower DOI prefixes?
To answer this question we need to go behind the scenes. It’s true that prefixes are generally assigned sequentially, beginning with the first Crossref prefix (10.1001) and incrementing up by one each time (10.1002, 10.1003, etc.). This would all be very straightforward if Crossref were the only DOI registration agency (RA); if this were true, any new member joining Crossref might reliably have a prefix that was one digit higher than the last new member’s prefix.
However, Crossref is one of 12 RAs, along with our friends at organizations like ISTIC, KISTI, JaLC, DataCite, mEDRA, and others in the scholarly information space, as well as three in other industries entirely (e.g., film/media, construction). To avoid two RAs simultaneously attempting to assign the same DOI prefix to two completely different organizations, we rely on CNRI, the Corporation for National Research Initiatives. CNRI administers the handle system upon which each RA’s DOI infrastructures are built and also serves as a central coordinating body between the RAs.
What this means is that Crossref and our fellow RAs are allocated batches of DOI prefixes from CNRI, then assign those prefixes sequentially to our members. Those batches are mutually exclusive; a DOI prefix that CNRI assigns to Crossref will not also be assigned to another RA. When we exhaust a batch, we request and are allocated a new batch. In practice, what this often means is that we will assign about 2,000 DOI prefixes in sequence to our members and then there may be a jump of a few hundred to a few thousand as we skip over batches of prefixes that CNRI has subsequently allocated to other RAs. For instance: from 2021 to 2022 we assigned new Crossref members prefixes from 10.55268 to 10.57267 (equalling the approximately 2,000 new members who join Crossref in a year). However, the next member that joined Crossref did not receive prefix 10.57268, as this prefix and subsequent ones had been assigned in batches to fellow RAs JaLC and DataCite. Instead, the next member to join Crossref received the first prefix from the next batch allocated by CNRI to us, 10.58168.
This process explains why you’ll usually see new prefixes incrementing by one, and then occasionally incrementing in jumps upward of a few hundred or a few thousand. It does not explain why Samuel noticed that new prefixes this year were suddenly lower than prefixes we had assigned last year.
In 2023 we were contemplating a project to automate some elements of prefix assignment in our system. To test this, we reserved a batch of 6,000 prefixes from CNRI. As research for this project was ongoing, we decided not to assign any of these prefixes to members. Instead, we requested a second smaller batch of prefixes from CNRI, with values ranging from the 10.69000s through the 10.71000s. We actively began assigning prefixes, temporarily skipping over our large batch of 10.63### through 10.69### prefixes.
Earlier this year, we exhausted our smaller prefix batch ending in the 10.71000s. The 6,000 previously unused prefixes were now available to actively assign which is why new prefixes this year are lower than ones assigned last year. Prefixes will now be assigned in the expected fashion, incrementing by one for each new member until all 6,000 are allocated (which, if current pace holds, will be in 3-4 years). After that, we’ll again get a new batch from CNRI and new members’ prefixes will again be in the 10.71000s or higher.
I want to stress that while all of this is interesting (I hope!), it doesn’t fundamentally change anything for our members. There is no meaningful value to a higher or lower prefix number. DOIs and DOI prefixes are not intended to confer a brand or identity in and of themselves; they are simply tools to identify digital objects. Whether your organization was assigned prefix 10.1001, 10.64001, 10.71001, or, someday, 10.100001, your benefits and obligations as a Crossref member remain the same.
One final note: it is possible to move (or “rehome”) a DOI prefix from one RA to another. If your organization wishes to move from Crossref to another RA, or from another RA to Crossref (for which we do not charge content registration fees to reregister existing DOIs already registered via another RA), we can assist you with the process of rehoming your prefix. This can be slightly laborious and intricate but it’s valuable because it helps preserve the functionality and uniqueness of your DOIs. Please get in touch with us at member@crossref.org if you have questions about this.
Thanks to Samuel from Cultural Hosting for this question! Is there anything else you’d like to know about DOI prefixes? Let me know in the comments.
—Collin