Publication Status shows only one deaccessioned dataset when logged in

[cmbz]

When does this issue occur? When logged in, a link to only one deaccessioned dataset is visible in the Publication Status facet.

Which page(s) does it occurs on? Main page of the (Harvard Dataverse) repository

To whom does it occur (all users, curators, superusers)? I have the following roles for this repository (Harvard Dataverse):

• Admin
• File downloader
• Contributor
• Curator
• Member

What did you expect to happen? I expected to see a link to thousands of deaccessioned datasets rather than a link to only one dataset.

Which version of Dataverse are you using? 5.11.1

Any related open or closed issues to this bug report? I didn't see any related bugs.

Screenshots:

[jggautier]

I've always thought that the Publication Status facet shows only the deaccessioned datasets that you have edit permissions on. I thought this was by design to reduce the discoverability of deaccessioned datasets.

I'm not saying this is how it should be, just what I think is going on.

Only superuser accounts would see a count of all deaccessioned datasets, since those accounts have edit access to all datasets in the installation, and the one account I can find of yours on the Harvard repository isn't a superuser account.

@cmbz, is that one deaccessioned dataset one you published or otherwise have edit permissions on?

[jggautier]

I wonder if there might also be some confusion about permissions and their effects, too. There are several GitHub issues related to confusion about permissions. Could you share where you see that you have those five permissions in the repository?

[cmbz]

@jggautier I didn't publish the dataset or have any specific permissions associated with the one deaccessioned dataset that I can see. That's one reason why the display seemed odd/wrong to me.

[cmbz]

The range of roles I have show up under the Roles section on the My Data tab.

[cmbz]

I've always thought that the Publication Status facet shows only the deaccessioned datasets that you have edit permissions on. I thought this was by design to reduce the discoverability of deaccessioned datasets.

Got it. That makes sense. In that case, I would expect to see zero deaccessioned datasets. To my knowledge, I have no permissions for the one deaccessioned dataset (or its dataverse) that I am able to see.

[jggautier]

Ah, is that deaccessioned dataset at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/SCPL2Q? I see it in the facet too when logged into an account that isn't a superuser account.

I think it might technically be because in that dataset's permissions table, anyone with an account has been given the "Member" role:

I'm not sure why that was done for this dataset or if and how often that's done for other datasets in the repository. The "Member" role lets other accounts "view both unpublished dataverses and datasets".

[jggautier]

In the short term, I'm going to remove that Member role so the dataset doesn't show up anymore for anyone except admins of the Dataverse collection, the dataset creator and superuser accounts. And I think the curation team will eventually just delete this deaccessioned dataset.

But if or when anyone is able to do a review of permissions in general, which is what I think the issue at https://github.com/IQSS/dataverse/issues/4389 is about, they could talk to the admins of that collection about why they added the Member role to that dataset.

What do you think @cmbz?

[cmbz]

Sounds good to me. However, I think that there is still the potential for confusion about what a user might expect to see/not see for deaccessioned datasets in the Publication Status facet in the event another collection chooses to allow access to all Members....which is admittedly an edge case.

[pdurbin]

Without digging in deep, yes, I think Julian is right that only superusers should see lots of deaccessioned datasets. I also agree with both him and Ceilyn that this should be better documented to set expectations.

My memory of how this work is that we only show a deaccessioned dataset "card" (which also appears in the facet list) when it's the only surviving version of a dataset. That is, if there's a draft or a published version, we don't show a deaccessioned dataset card. And of course, such a card is only shown to users with permission, such as the Member role mentioned above.