Warn users about publishing to the ADC

[ThomasThelen]

The Arctic Data Center is a highly specialized repository that contains human-curated packages. There's a workflow used within NCEAS to publish packages that involves a manual verification process where a human reviews the package before it gets a DOI and shown to the public.

In order to cut down on any accidental publishings of non-arctic data to this repository, I think we should have some sort of warning/extra popup system. The KNB is a more general purpose repository and more appropriate for Tales so I don't think we need one there.

Option 1: Show text on the screen when the ADC is selected from the dropdown

Option 2: Popup when the user clicks 'Publish' with the ADC selected

Option 3: Both

If we really wanted to, we can add a description of the repository (option 1) and then show the popup if they select the ADC.

I like Option 2 the most because it requires action from the user to confirm that yes, I have high quality arctic data and it belongs here. Any strong feelings on these options, @craig-willis , @bodom0015 ?

[amoeba]

I think a proper warning, as you suggest is a good idea. I wondered what this might look like if it were more of a guide than a warning though. What do you think of something like this?

Some key things I wanted to hit on:

• Best case default: If the user has no idea, they should go to the KNB
• Makes it clear what will happen next.
• Improves branding visibility

[ThomasThelen]

One reason we put the repositories in the dropdown is because we can scale it as the number of repositories increases (in anticipation for adding dataverse nodes). We can of course take your design and turn the repository area into a scrollable area with nodes.

I think that making the Moderation section clear extends to providers other than DataONE, so I think it's a field that we should add.

At this stage I'm reluctant to give the publishing dialog any significant makeover, so I wonder if there's a compromise where we can show the moderation level and keep the current layout. For example, we can put this information in the More Details sections and auto expand it.

[amoeba]

Sounds good. In that case, I'm more a fan of option 3 but I think it'd be great if the text was less of a warning and a bit more helpful. For a phrase like "ensure your data is relevant here", we could pretty much list one or two things about the repository that can tell them whether it is without requiring them to go look.

I think this is an especially good idea because there's no link to the Arctic Data Center in the modal. Maybe it should be easy for a reasonable user to to be able to confidently pick from the options provided because the modal provides enough information.

[craig-willis]

We discussed similar issues at the June AHM as we consider adding more publishing targets to the system.

A few questions:

• Are the curation requirements on ADC enforced in the API -- i.e., we current can't programmatically assign a DOI? If so, this means that the behavior of ACD and KNB are different in WT today?
• Does KNB really accept everything (ala Zenodo)? If so, why was my example Tale deleted?

Building on @amoeba's "moderation" and "DOI assignment" characteristics, here's what we've seen as we look at publishing to Dryad, Dataverse, and Zenodo:

• Zenodo: Anything goes, no curation/moderation, user can mint their own DOIs.
• Dataverse: Optional full curation. Each Dataverse installation can have multiple "Dataverses" each with different policies associated with particular journals/communities. Some have no curation and others have strict curation workflows. For example the AJPS Dataverse allows users to "Submit for review" (dataset is private, DOI is reserved) but only a curator can publish (dataset is public, DOI is assigned). Curation is decentralized.
• Dryad: Full curation. Users can only "deposit" packages (DOI reserved), which go through a centralized curation process (DOI assigned, package made public).

If KNB is really willing to accept everything, then perhaps we should consider disabling ADC today? The number of actual WT users that will want to publish to ADC in the near term seems much lower than the number who will accidentally do so. We can include ADC when we have an approach in place to handle curated targets.

This discussion also emphasizes that we should continue to prioritize Zenodo integration. During the early days WT users can deposit without controversy.

[ThomasThelen]

Right now publishing to the ADC, KNB, and development servers function the same way; the ADC does not enforce curation via the API which leaves the question of how we determine when a dataset has been accepted in a non hacky way (periodically checking for the obsoletedBy field in the system metadata). This might be a great segue into a discussion about supporting workflows from other repositories.

I think the KNB will accept most things as long as it, in some aspect, pertains to ecology. For example, a dataset containing fishing license records should be valid there.

I've shot a message out on the NCEAS slack for input on whether or not we should disable the ADC feature until we have this settled; I'll post an update here