Any attribution requirement can unexpectedly burdensome

[heathermeeker]

Regarding this clause: "You include with any Data you redistribute all credit or attribution information that you received with the Data, and your terms require any Downstream Recipient to do the same"

Those who aggregate and redistribute the data will need to include "credit or attribution" information. There are two issue with this:

--Licensees may have a hard time determining what constitutes credit or attribution information. Consider requiring it to be specified in a manner in which it can be identified by automatic scripts. In Apache 2 this is done via a specific text file name. In Polyform (non-commercial) we did this by requiring "any plain-text lines beginning with Required Notice:". Those are both software licenses, of course, but just examples. For Polyform project we thought a lot about how to do this to minimize notice burden.

--Even so, any notice can be unduly burdensome for researchers. For an amusing perspective on this, see the "bullfrog problem" article. http://peterdesmet.com/posts/illegal-bullfrogs.html. Consider whether notices are truly a legal necessity, particularly when anti-plagiarism rules may provide the motivation for attribution without creating legal liability.

[kemitchell]

The relevant Polyform PR on required notices was https://github.com/polyformproject/polyform-licenses/pull/37.

[hannakim123]

Thanks @heathermeeker for your comments on the attribution and credit issue. We do see the value in simplifying attribution but we also see the value in understanding the provenance of the data. We have to balance these competing interests. With data, researchers and the like want to know where the data came from and whether they can trust the data and having attribution or credit information can help with that issue. We also didn’t want to prescribe a specific way of doing the attribution in that the technology can change and there are varying considerations for different types of datasets (images, text, video, or audio). The choice to put attribution or credit information is up to the Data Provider and if no attribution is present, the Downstream Recipients do not need to create one if they are redistributing.

This is a good issue to put into an FAQ to explain attribution and what’s the minimum requirement but for now, we will update the comment section in the draft.