So... the Royal Society wanted to write 2017 Marsden success to ORCID records. What happened next?
Jason Gush
Warning: the below describes both our communications approach to an ORCID-registration campaign and some rather technical information about how we used the Hub to write funding. Feel free to ignore the bits which aren’t a second nature to you.
One of the target features for the NZ ORCID Hub was to be able to use it to write to other sections of the ORCID record, beyond affiliations. From our New Year’s Eve post, you'll have seen that we ended 2017 with the production Hub having a somewhat rough and ready implementations of both the ability to write funding, and a requested feature to customise the messages the Hub sends.
To road test, and as a useful pilot campaign, a Society and Marsden Fund Council approved effort to recognise investigators awarded in the 2017 Marsden Fund funding round was undertaken. This is a report on what we did, how it went, and what’s changed as a consequence.
The first issue: getting Marsden data for the 133 contracts involving 399 distinct investigators written as json ready for importing on the Hub.
|
Hurdle no. 1. Marsden houses the bulk of its data offline in MS Access (yes, I know J), and for baroque reasons we’ve a combination of 32bit Office and 64bit python which just won’t play nicely together. In retrospect, changing the python environment may well have been simpler but instead we went with shifting the relevant table snapshots into a contemporary 64bit MariaDB instance that our python environment could see. |
In this first attempt, we read each contract and its data into a python dictionary (i.e., title, public summary, amount, category, start year, end year, and the contract id). We iterate through this dictionary to retrieve each contract’s team information (i.e., name, email, and role), and write these together with the Fund’s organisation data into a third funding dictionary which was dumped into a json file.
|
A bit of experimentation showed that the order of the contributor list was rather unstable in the Hub: initially it’s a reverse of the order presented in the file; however, as each team member gives permission the order (and thus any subsequent write) will be shuffled. This is obviously undesirable for funding, and unacceptable when we shift to works, so the team were quick to get a fix in place into the current production. |
To let people know what we were doing, on the 21st of
December email was sent to each of the 399 investigators explaining what would
happen (the Hub would be sending them an invitation), when (the 8th
of January) and, importantly, showing them the information that would be
written to their records if they gave us permission.
As well as the seasonal barrage of “out of office” replies, we received twelve enthusiastic acknowledgements along with one correspondence from/about a group concerned about the additional exposure of their project. While this group decides how to proceed, this contract was removed from the upload and thus the invite queue.
As a co-effort by the Society and the Marsden Fund Council, this was an excellent opportunity to use the new customisation feature: the Marsden/Society co-branding logo was used, and as this didn’t work on the black header and footer banners, these were changed to white, i.e.:
On the 8th January, the uploaded file was activated spawning (after the removal of the contract under discussion, and minus bounces) a total of 386 Hub invitations. Of course, with the assistance of the good Murphy we hit a snag: a coincidental reactivation of a config glitch brought the Hub down for two hours.
|
That config error also crashed the Hub before the automated error notification loaded, which it took us to start getting email from users before we realized that not all was well. Once the UOA team were notified, service was quickly resumed and the 27 who had contracted us were mailed with both our thanks and the notice that we were up once again. Rather than spam everyone, only those 27 were mailed as we were comfortable that any silent failures would be caught when we sent the intended reminder. |
A small confusion about the difference between task and invitation expiry (we’d never discussed it explicitly with the dev team) meant that invitations expired on the 22nd of January rather than the 8th February promised. This came to our attention with an email from a user who’d failed in their attempt to use the invitation link, and spurred us to start the reminder process early.
|
There were at least two consequences of that expiry email: i/ proper warnings are now given for expired invitations; and, ii/ task expiry was modified so that it could be reset by the uploaded resetting/reactivating the task. |
That task reset would have one of two effects: if the user had given permission and the access token was long-lasting, the funding item would be updated with all known ORCID iDs; however, if the access token was short-term, had been revoked, or just never given then the user would be sent a new invitation email with a fresh two-week lifespan.
|
The first version of the funding task page lacked a “reset all”; this proved painful for a large task that spread over 10 pages. Creation of a “reset all” funding function was pushed up the priority list, and is now on production |
This time the Hub’s messages were given a little more background, specifically the invitation’s expiry date, i.e.:
At the close today, the final tally was 240 access tokens, i.e., 62% of
invitations were translated into access tokens and funding written.
https://pub.orcid.org/v2.1/search/?q=%22marsden%20fund%22
Thinks we learnt:
Good communications are important
- let your people know they're going to receive a hub invitation and preferably tell them what you'll be writing.
- let us know when you're undertaking a large campaign so we (and the dev team) can be standing by
- try not to undertake a campaign over the summer break :)
Try to keep task size manageable; large tasks will never complete and make finding problems difficult. Fifty–sixty is a good size; 100 is large.
Don't forget invitations expire after two-weeks, and the task itself after a month (athough this is being changed).
As with our experience with other mail campaigns, you can expect reminders to net you another 50% of the initial response.
To do:
Lots of tweaks needed to the UI, especially improving the guidance for general use.
A little bit later we’ll be taking a look to see what differences there were in uptake, and once we’ve understood where we need to put additional effort we’ll start working our way back through the Fund’s (and the Society’s) history.
If you've made it to the end: congratulations! We hope this gives you some insight into what's needed to get a campaign of this kind going. Although not the 95% conversion of my hopes, our internal threshold for success was in the region of 60% and we're happy to have met that.
If you've any questions, feel free to get in touch.
Cheers,
Jason.