Questions for @RHo:

• For the "add an image" design, the rejection step is a fullscreen overlay, whereas for "add a link" it's a modal (see screenshot below). While I know we tested these options against each other with Concepts A and B, I want to check whether we want consistency here across the two task types.
• "Add a link" is radio buttons, whereas this is checkboxes. Do we want consistency across them? Or should we think about this as a test to see whether checkboxes give us data that we prefer?
• In "add a link", when the user gives a rejection reason, they can go back to the suggestion, see that the "No" is still selected, and tap an ellipse to edit their rejection reason. Perhaps that's not needed here because we're only doing one suggestion per task (as opposed to many links per task). That's how it is in the "add an image" prototypes: if the user goes back from selecting a rejection reason, the "No" button is not selected -- they can just re-select it to go back to rejection reason. Same with the "yes" button -- if the user goes back from the caption stage, the "Yes" button is not selected. How do you think this should work?

Checkbox (multiselect) support will be added via https://gerrit.wikimedia.org/r/c/mediawiki/extensions/GrowthExperiments/+/719373

Seems like this would be more easily done after T290045: Add an image: image inspector

Change 720937 had a related patch set uploaded (by Gergő Tisza; author: Gergő Tisza):

[mediawiki/extensions/GrowthExperiments@master] [WIP] Add rejection dialog

https://gerrit.wikimedia.org/r/720937

It takes them to the post-edit dialog. See T290789: Add an image: post-edit dialog (PLACEHOLDER).

Just to clarify this – for Add-Link, pressing no on the last/only suggestion takes you to the edit summary screen, where you then press publish. For Image-Suggestions, there will be no edit summary screen for rejection responses?

Change 720937 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add rejection dialog

https://gerrit.wikimedia.org/r/720937

In T290046#7376317, @kostajh wrote:

It takes them to the post-edit dialog. See T290789: Add an image: post-edit dialog (PLACEHOLDER).

Just to clarify this – for Add-Link, pressing no on the last/only suggestion takes you to the edit summary screen, where you then press publish. For Image-Suggestions, there will be no edit summary screen for rejection responses?

Yes, that's correct for Iteration 1 where there is only one image suggested.

@RHo -- could you please help with these clarifying questions?

• For the "add an image" design, the rejection step is a fullscreen overlay, whereas for "add a link" it's a modal (see screenshot below). While I know we tested these options against each other with Concepts A and B, I want to check whether we want consistency here across the two task types.
• "Add a link" is radio buttons, whereas this is checkboxes. Do we want consistency across them? Should "add a link" become checkboxes? Looking at the options in "add a link", I don't think that more than one of them should really apply to the same link, so that lends itself to radio buttons. But I am concerned about consistency.
• In "add a link", when the user gives a rejection reason, they can go back to the suggestion, see that the "No" is still selected, and tap an ellipse to edit their rejection reason. Perhaps that's not needed here because we're only doing one suggestion per task (as opposed to many links per task). That's how it is in the "add an image" prototypes: if the user goes back from selecting a rejection reason, the "No" button is not selected -- they can just re-select it to go back to rejection reason. Same with the "yes" button -- if the user goes back from the caption stage, the "Yes" button is not selected. How do you think this should work?

@Tgr -- I just added a requirement here that only certain rejection reasons cause images to be removed from the pool. Could you please tell me whether that's no problem, or if it is substantial work? If substantial work, we can cut it out of scope for Iteration 1.

In T290046#7396000, @MMiller_WMF wrote:

@RHo -- could you please help with these clarifying questions?

• For the "add an image" design, the rejection step is a fullscreen overlay, whereas for "add a link" it's a modal (see screenshot below). While I know we tested these options against each other with Concepts A and B, I want to check whether we want consistency here across the two task types.

Yes my bad, I've updated to use the same consistency with both tasks to not use a full-screen modal.

• "Add a link" is radio buttons, whereas this is checkboxes. Do we want consistency across them? Should "add a link" become checkboxes? Looking at the options in "add a link", I don't think that more than one of them should really apply to the same link, so that lends itself to radio buttons. But I am concerned about consistency.

• In "add a link", when the user gives a rejection reason, they can go back to the suggestion, see that the "No" is still selected, and tap an ellipse to edit their rejection reason. Perhaps that's not needed here because we're only doing one suggestion per task (as opposed to many links per task). That's how it is in the "add an image" prototypes: if the user goes back from selecting a rejection reason, the "No" button is not selected -- they can just re-select it to go back to rejection reason. Same with the "yes" button -- if the user goes back from the caption stage, the "Yes" button is not selected. How do you think this should work?

Yes, confirming that for this single add image task, the behaviour in the prototypes you described is how I think this should work.

@Tgr -- I want to point out that a few of the specifications changed here:

• It went from being an overlay to a dialog (like "add a link" has), which alters the buttons and actions.
• We would like to selectively remove images from the queue based on the rejection reasons. Let us know if this substantially increases scope.

Moving accordingly. I think those are both easy changes.

Change 730678 had a related patch set uploaded (by Gergő Tisza; author: Gergő Tisza):

[mediawiki/extensions/GrowthExperiments@master] Add Image: Do not discard recommendation for some rejection reasons

https://gerrit.wikimedia.org/r/730678

Change 730687 had a related patch set uploaded (by Gergő Tisza; author: Gergő Tisza):

[mediawiki/extensions/GrowthExperiments@master] Add Image: Change rejection dialog style

https://gerrit.wikimedia.org/r/730687

Change 730687 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add Image: Change rejection dialog style

https://gerrit.wikimedia.org/r/730687

Change 730678 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add Image: Do not discard recommendation for some rejection reasons

https://gerrit.wikimedia.org/r/730678

For @RHo review
(1) minor

• It has a "Cancel" button, which returns the user to the image inspector with neither the "Yes" or "No" button selected.

After a user clicks on "Cancel" - the button "No" is still displayed as selected - the button has a blue outline.

(2)

• The user may select the "Submit" button whether or not they choose a response. The "Submit" button is the only way to proceed beyond this step after the user has selected "No". It takes them to the post-edit dialog. See T290789: Add an image: post-edit dialog.

The current workflow now is as follows:

• a user clicks "No"
• "Why is this image not right..." is displayed
• a user doesn't select any reason and clicks on "Submit"
• "Save your changes" overlay is displayed

• Does "Save your changes" is a correct header here?
• The "x" button looks confusing - it looks like the image could be removed from the Summary.

Note: I'll be doing more testing on images being removed from the pool.

Thanks @Etonkovidova - I've moved back to fix the minor issue (1) mentioned, as well as since the copy has since been finalised and updated on this task.

In T290046#7483850, @Etonkovidova wrote:

For @RHo review
(1) minor

• It has a "Cancel" button, which returns the user to the image inspector with neither the "Yes" or "No" button selected.

After a user clicks on "Cancel" - the button "No" is still displayed as selected - the button has a blue outline.

Good catch, the blue outline should be removed on the "No" button.

(2)

• The user may select the "Submit" button whether or not they choose a response. The "Submit" button is the only way to proceed beyond this step after the user has selected "No". It takes them to the post-edit dialog. See T290789: Add an image: post-edit dialog.

The current workflow now is as follows:

• a user clicks "No"
• "Why is this image not right..." is displayed
• a user doesn't select any reason and clicks on "Submit"
• "Save your changes" overlay is displayed

This is correct for Iteration 1 (reject will take users to the review summary screen first) and I've updated the task description accordingly to reflect that taken users to Post-edit dialog is for a future iteration.

Questions:

• Does "Save your changes" is a correct header here?
• The "x" button looks confusing - it looks like the image could be removed from the Summary.

I would refer to specs on T290784 for review but the button in the header is definitely off. the "x" and image icons should be a lighter base20 colour which should help with making them look informational rather than actionable.

Note: I'll be doing more testing on images being removed from the pool.

Change 737784 had a related patch set uploaded (by MewOphaswongse; author: MewOphaswongse):

[mediawiki/extensions/GrowthExperiments@master] Add an image: update copy & QQQs for image inspector, rejection dialog, edit summary, caption UI

https://gerrit.wikimedia.org/r/737784

Change 738051 had a related patch set uploaded (by Gergő Tisza; author: Gergő Tisza):

[mediawiki/extensions/GrowthExperiments@master] Add Image: Save dialog icon color

https://gerrit.wikimedia.org/r/738051

In T290046#7489906, @RHo wrote:

Good catch, the blue outline should be removed on the "No" button.

Note that the outline does not mean "selected" or "pressed", it means "focused" (ie. it signifies where we are in tab navigation, although I guess that's not very significant on mobile). We can move the focus elsewhere if needed, though.

Change 738054 had a related patch set uploaded (by Gergő Tisza; author: Gergő Tisza):

[mediawiki/extensions/GrowthExperiments@master] Add Image: Suppress outline when canceling rejection dialog

https://gerrit.wikimedia.org/r/738054

Change 738051 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add Image: Save dialog icon color

https://gerrit.wikimedia.org/r/738051

Change 738054 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add Image: Suppress outline when canceling rejection dialog

https://gerrit.wikimedia.org/r/738054

Change 737784 merged by jenkins-bot:

[mediawiki/extensions/GrowthExperiments@master] Add an image: update copy & QQQs for image inspector, rejection dialog, edit summary, caption UI

https://gerrit.wikimedia.org/r/737784