Commons:WMF support for Commons/Upload Wizard Improvements

From Wikimedia Commons, the free media repository
Jump to navigation Jump to search
The printable version is no longer supported and may have rendering errors. Please update your browser bookmarks and please use the default browser print function instead.

Part of the Commons support from Wikimedia Foundation will be directed towards improving the current user experience with UploadWizard. We started our work in Fiscal Year 2023-2024, as part of last year’s Annual Plan, and we will continue to provide support for Fiscal Year 2024-2025 as part of objective and key result WE2.3 (“Guide contributors to add images and references that comply with project guidelines and increase trust in content, for example, by flagging potential issues during their upload/addition”).

The target audience for this project is primarily focused on new users, but we are welcoming feedback from more experienced users.

Goals of the project

The problem we want to solve is how to reduce the number of bad uploads through UploadWizard, so that we can reduce the number of deletion requests, and in perspective, reduce the burden on moderators and improve the quality of our content.

Our aims are:

  • To improve the Commons upload experience for users;
  • To attempt to decrease the burden for contributors with extended rights, by minimizing the possibility of uploading media that might trigger a deletion request.

Focuses of the project

This project has focused in the first months on UploadWizard design improvements, starting from the user research results we conducted and improving our initial designs with a community consultation. Our goal was to make it clearer to new users how to provide the right information when uploading new media, reducing the possibility of uploading material that would be flagged for deletion.

During these first months, we focused on the “own work” versus “not my own work” workflows, which also included improving option selection and exploring the possibility of adding machine detection “speed bumps.”

We are now focusing on the section of UploadWizard where users get to describe and add captions to the media they are uploading. In parallel, we are investigating how to programmatically detect and flag potentially problematic content in UploadWizard.

Discovery of the problem

In July 2023, we conducted a user research with participants from 12 countries about the common issues users face when uploading images. The report highlighted several points in which the experience of uploading new media on Wikimedia Commons can be improved, giving more context and explanations about licensing and the general rules of the project, explaining better what to do when uploading media that is not the own work of the uploader, and possibly revamping the current UploadWizard to make it more user friendly.

We then conducted interviews with Commons administrators, coming from 10 different linguistic communities, about their take on UploadWizard, and the way it is used to upload media. The interviews revolved around the various levels of effort that deletions imply (from speedy deletion to more complex copyright issues), the problems arising from the usage of UploadWizard, and what are the expected improvements that administrators suggested to us.

We also conducted an analysis in July-August 2023 about the most common reasons for deletion requests on Commons. The most common are “copyright violations” (including violations of freedom of panorama, derivative works and threshold of originality), and media “not being within project scope”.

Design proposals

We also conducted usability tests about the proposed improvements to UploadWizard. The report highlighted that most of the proposed changes, in fact, improved the understanding of the process by new users, even if it didn’t prevent in all cases a user from uploading a media that would be flagged for deletion. The tests provided us with more ways of improving the work already done, and present a better version of the changes.

Success criteria

  • Qualitative feedback from users, user testing of upload process improvements
  • Decrease in the percentage of uploaded media that trigger deletion requests

What’s out of scope?

  • Full UploadWizard redesign (although more improvements might be considered in the next iterations)
  • Optimization of Commons’ deletion queues
  • Creation of a new anti-copyvio tool

Screenshots & Prototypes

“Own work” vs. “Not own work” step

Based on received feedback after the first round of improvements carried on in Fiscal Year 2023-2024, we suggest making some more changes to the step in which users select if their media is an “own work” or “not own work”.

The present design had users choosing the wrong option from the start, leading to an incorrect path from the start (e.g. someone uploading someone else’s media choosing “own work”, because it was part of their job to upload that media), and did not emphasize enough that anyone can re-use the media, and that re-usage is not limited to Wikimedia projects.

From our tests, the above changes improved the possibility for users to select the right option at the start, confirmed by the fact that this right choice introduced them to the right path and the right questions.

Uploading on someone else’s behalf

Our tests suggest that our current approach does not accommodate the case in which users upload photos on behalf of someone else. More specifically, users do not understand the questions due to their lack of understanding of Creative Commons licenses, assuming it’s something “legal” and therefore choosing the “I don’t have this information” option, preventing them from completing the upload. While it is good to not upload things without proper knowledge, we think there is an opportunity to educate users on this step.

The tests about the new message showed that people better understood what was being asked for and why. When proposed with the scenario of uploading someone else’s work, every user in the test correctly chose the right option “I have permission”.

The warning that was shown upon selection of that option was understood by most people, and stopped them from proceeding without getting the consent. Some users who decided to proceed despite the warning, had either chosen to not read the warning or did not understand it clearly due to the UI not in their native language in the test. However, based on the results, the introduction of this option should greatly reduce the number of uploads without VRT verification.

Adding custom PD tags or custom license options

We listened to your feedback in the previous months, and we added a custom field to enter a specialized PD tag or custom license, in case a more specific tag/license is required when appropriate.

Adding other “I don’t know” options

From the use cases we analyzed, we found out that there were cases where people assumed that things like magazines and logos are free to share and confidently chose a public domain option. One of the assumptions we worked on is that “public domain” might be misinterpreted as “anything made available to the public”, as also found in previous tests. We reworked the message to attempt to clarify such misinterpretation.

This is unlikely to completely stop all wrong interpretations, but the new wording will allow flagging these uploads as potentially problematic, if the user has not provided any concrete information (e.g. if they indicate having found a media on the internet without providing any evidence of why it's in public domain).

Adding space to clarify AI prompts

Based on feedback we received during last Fiscal year, we introduced an option for users who are uploading AI-generated media to add the prompt that was used. The relevant field appears when the user selects the “AI option” in question 1 of “own work” flow.

The existing field that compels the user to enter the name of the AI engine used will remain mandatory. As for the prompt field, the design recommendation is to leave the field optional, as the user may or may not have the prompt readily available or may not have saved the original prompt. This would prevent the case in which the user would choose a potentially incorrect action to proceed anyway.

Past improvements

These improvements were carried out during Fiscal Year 2023-2024.

Choosing "Own work" vs. "Not own work" workflow

There is also a prototype of the proposed changes to the "Release rights" workflow, accessible at the following link (some links or functions may not work as expected due to prototyping limitation).

Changes to "Describe" workflow

Project contacts

If you want to know more about the project or to ask a question, you can reach out to Luca Martinelli [Sannita], Movement Communications Specialist