Bug? All changes made in in-document list of references not saved

  1. I format a bibliography.
  2. There are a number of errors in the data.
  3. I share the document with my research assistant who spends 3-4 hours correcting the metadata for about 100 citations.
  4. He “saves” each item after making the changes, but these are not saved back to my Paperpile account or to his. Where are they saved?
  5. I go to reformat, and it pulls the old, incorrect metadata from my Paperpile account loosing all his data.
  6. I rollback Google to before he made the updates and his work is restored. But it is of no use to me unless I can reformat the citation based on these changes.

Thanks for your help.

(PS; my paper is due tomorrow!!!)

PS: I also tried going in and clicking “save” but when I open it in my Paperpile account I don’t see the saved changes. Even making a minor change of my own here and saving again still doesn’t save it to my own PP account.

You will need to edit the data in your Paperpile account and they will propagate to the Google Doc. That does not work in the other direction. If you collaborate with 20 authors on a Google doc and somebody edits a reference it should not affect your library. So that’s not a bug that’s by design. Once a citation has been changed outside of Paperpile, i.e. in the reference view, it will not be updated from the data from your library.

It depends on where your assistant has made the changes. If he changed the data in the Google Doc and you reformat it they should not be overwritten by the data from your library. The data is saved independently in the database.

We thought very carefully about this because it’s not trivial how to deal with that in a collaborative environment. We have not heard a single complaint about that in 18 months. So if there is a bug it would surprise me.

I double checked and in our current implementation you actually can overwrite the data in a Google Document when you change the data in your library also after someone has made changes in the Google Docs version of the item.

But that would not explain why the data would revert in your case to the old version in your library. That can only happen if you change some data in the library yourself.

In any case, no changes will ever be propagated form the Google Docs version to the library.

It really seems that this check to avoid the overrides is not implemented. I was 100% positive that we had that.

We need to think about this again. Probably we can get it in the release for this weekend, probably too late for your paper.

But if you don’t touch anything with your data in your library the version your assistant has made in the Google Doc should be the version that your bibliography gets when you reformat.

But the most robust solution would be to correct the data again right in your library.

Sorry for all the trouble. I’ve refunded your subscription. That’s not the experience we want for our customers.

Ok, now I also see why it got reverted in your case. The update from Library > Google Doc actually also happens if you search your library in the Google Docs plugin. So when you search a generic term and 10 items are returned the changes of your assistant in the Google Doc will be overwritten by the old version in your library.

I’m still stumped that this has been unnoticed. Paperpile has been used for 1000s of collaborative papers but It really seems nobody edits someone else’s references right in the Google Doc.

Thanks for the response, refund, and understanding. I’m also glad you figured out the bug. A couple of notes.

There needs to be a way to save collaborative work on a bibliography back to one’s paperpile. I see why you don’t have this on by default, but there should be an option to save changes in one’s account as well. Otherwise anything fixed in one collaborative paper will still be wrong in the next paper.

My assistant was able to save items not in his paperpile to his account, so he has the proper version there now. He was able to export these to me as an RIS file as well as to share them with me on Paperpile. I would like to now import these to my library, but I’m a little concerned about how my library will handle what are now duplicate entries (still not happy with the way Paperpile handles duplicate since it has overly strict rules for identifying things as duplicates). I could delete all the items I have already and then import - but I will loose all my metadata (tags) and attachments (PDFs). However, at least this way I will be able to format the bibliography properly. Really not sure what the best workflow would be here…

1 Like

Here is my workflow for fixing the problem:

  1. Go through and select all 98 citations in my PP account and put them all in a folder. (See my feature request about automatically creating a folder like this when generating a bibliography.)

  2. Go through all 37 of the items with PDF attachments and save them to the desktop.

  3. Delete all 98 items.

  4. Re-import from a shared folder created by my assistant. (See feature request for more user control over handling duplicates and merging items.)

  5. Reformat the bibliography.

This seems to have worked. I still have to re-attach those PDFs which can’t be easily downloaded from the web and re-apply all my tags and folders, but at least I can submit my paper. (Although actually there are still some errors in the metadata that need to be fixed…)

Oh, I also exported the original RIS of the folder with the uncorrected metadata so that when I’m all done I could send you the old RIS file and the corrected one so you could compare the two (I remember in another thread one of you had asked for this kind of information). Please let me know how to send that to you when it is ready.

We’ve fixed now the bug and changes in your personal library will not overwrite changes you have made in the Google Docs specific copy. We also added a hint to the edit dialog what the behaviour is.

This is a quick fix enforcing the behaviour we had in mind originally: If you edit your copy in your library the item in the Google Doc will be updated. If someone edits the Google doc copy it will not propagate back to your library and it will prevent any further changes of the copy in your library to overwrite changes in the Google Doc.

We are now evaluating if and how the collaborative workflow can be improved. But for now you should at least not be surprised by the behaviour.

As always, thanks for bringing up the issue!

1 Like

This is certainly an improvement. Thanks.

If someone edits the Google doc copy it will not propagate back to your library and it will prevent any further changes of the copy in your library to overwrite changes in the Google Doc.

My suggestion would be to have a “sync” option that allows the changes made to the Google Doc copy to propagate back to the library, and resets the behavior so that future changes to the library update the Doc (until such a time as it gets out of sync because someone edits the Google Doc copy again, at which point one could re-sync manually).

1 Like

Yes that was also our own intuition. We would like to have a simpler option because that adds definitely to the complexity of the whole thing but I can’t think of anything else.

1 Like

While it does add some complexity, there such a sync mechanism could also be useful for shared folders, allowing any party on a shared folder to make updates which the others can then sync back to their own libraries manually if they like. (Right now I’ve been using a work-around of marking items “incomplete” by adding an X to the date, which is the only way that other members of a shared folder can make updates to an item.)

I agree with this line of thinking. The issue is really that since there are multiple copies of a reference in the Google Doc and collaborator’s personal folders, one change has to be made several times as it is. I get that the change should not be automatic as other users nay not agree.

The use case here is that a collaborator makes a citation, and the information from her or his PaperPile is then copied to the Google Doc, but there is an error that needs to be fixed, which another user does to the Google Doc citation. It would be great to use that data again and copy it back to fix the personal PaperPile copy. An indicator that it was updated, at least for the user who inserted it, would also be quite useful so that the user who inserted the incorrect citation is aware and can correct her / his own copy so that next time it will be correct.

7 posts were split to a new topic: Changes to citations are overwritten