Nautilus "Changing Backgrounds" help is unnecessarily long and doesn't cross-reference desktop background

[Matthew Paul Thomas]

Binary package hint: ubuntu-docs

Ubuntu 9.04, all updates installed

During Canonical user testing, one test subject followed this path in trying to change the desktop background:
 1. opened "Places" > "Desktop"
 2. chose "File System", then "Documents" in the resulting file manager window
 3. double-clicked on "Desktop"
 4. chose "Edit" > "Backgrounds and Emblems..." > "Add a New Pattern...", "Cancel", "Add a New Pattern...", "Desktop" (in the file picker), "Cancel", "Help"
 5. started reading the help, even confirming that it was helpful ("yeah, I just got to figure it out"), but did not finish reading it or follow its instructions
 6. chose "Add a New Pattern..." again, and tried to drag one of the images in the list to "Desktop" in the filepicker sidebar ("I was thinking you could drag it to the folders")
 7. started reading the help again, but again didn't finish
 8. dragged one of the patterns in the list into the "Documents" window, which changed just the background of that window
 9. chose "Edit" > "Backgrounds and Emblems..." again, but it stayed behind the filepicker ("it should come to the front, but it doesn't")
 10. looked through "Places" ("I always go to Desktop and then 'Change Background Picture', I think it is, but when I click Desktop here it doesn't ... Well, usually I click on My Computer, but...")
 11. went into "Places" > "Desktop" again
 12. tried "Places" > "Computer"
 13. tried "Desktop" in the sidebar again
 14. tried "Computer" in the File Browser toolbar
 15. dragged a pattern from the list to the Nautilus sidebar, which added the image to the sidebar
 16. gave up.

There are many design problems that need fixing here, but it was noticable in steps 5 and 7 that the "Changing Backgrounds" help page (a) was too long to be effective and (b) didn't even cover what the test subject was reasonably wanting to find out.

This could be fixed by heavily reducing the amount of text, and then adding a cross-reference to the "Desktop Background Preferences" page from the end of the page.

Examples of how to shorten the text:
* The entire first paragraph ("The file manager includes background patterns ... and on panels") could be deleted with zero loss of information.
* The sentence
        To change the background of a window, pane, or panel, perform the following steps:
  could become
        To change the background of a folder window, pane, or panel:
* "The Backgrounds and Emblems dialog is displayed. Click the Patterns button or the Colors button..." could become "In the Backgrounds and Emblems window, choose Patterns or Colors..."

[Phil Bull (philbull)]

Thanks for reporting this. The affected documentation is in the GNOME user guide, which is being rewritten over the next 2 release cycles. As such, I'm not sure that it's worth addressing this immediately rather than waiting for the new docs. What's your opinion, are many users struggling with this in particular?

Incidentally, is there any chance that the GNOME and Ubuntu docs teams could get our hands on the user testing data you have?

[Matthew Paul Thomas (mpt)]

Unfortunately we can't release raw videos of the tests, because one of the conditions of the subjects' participation is that we don't publish personally identifying information. However, as with other projects, I will report bugs for all the problems test participants encounter with the Ubuntu Help, when they open it while performing other tasks. We can't really test the help directly, because looking for something in the help is not itself an interesting task, so asking people to do it would produce unrealistic results.

This means that the testing data on Ubuntu Help will be rather incidental and piecemeal, so you'll need to extrapolate issues with one topic to other topics. For example, the Nautilus Backgrounds help is needlessly wordy, and insensitive to possible wrong turns the user has made — but those two problems are common to many of Ubuntu's help pages, not just that one. A systematic approach to fixing those problems might use the help page structure I proposed in 2006. <https://wiki.ubuntu.com/UbuntuHelp/PageStructure>

Now, should that wait until the Gnome user guide is rewritten? I don't see any point in waiting, unless there's going to be a license change that prevents text in the existing guide from being copied into new help pages. Otherwise, the new text could be written now, and copied into the new structure when it's ready.

[Phil Bull (philbull)]

Hi mpt,

For fear of getting further off-topic, I'll reply in a separate email regarding the test videos.

Unfortunately for this bug, there is going to be a license change when the whole guide is rewritten. Certainly, the g-u-g is not fit for purpose at the moment. I doubt it helps very many users, as I've seen with my own user testing. So, is it worth fixing this one issue that happens to have been reported as a bug when there are so many other issues lying unreported? Or should we just wait until the wholesale "fixing" of the docs on the arrival of the new guide?

If you have seen a significant number of users struggling with this particular issue and think that improving this topic will result in a reasonable improvement to their user experiences, then we should fix the bug now. Otherwise, we may as well wait.

[Matthew Paul Thomas (mpt)]

I met Paul Cutler yesterday and he explained the license issue to me. I'm a little concerned, though, that the relicensing shouldn't block any improvements to the g-u-g for an indefinite period.

One way of working around this would be to completely rewrite the Nautilus backgrounds section in place, with a source code comment at the start of the section saying that that section is dual-licensed under the FDL and the CC-whatever 3.0 or any later version. That way the whole guide is still licensed under the FDL, but you can copy and paste the text to the new help when its structure is ready.

What I was trying to explain in my previous comment was that because we can't test the help both directly and realistically, we will rarely, if ever, "see[] a significant number of users struggling with [a] particular issue" in the help. We'll occasionally see one person dive into a particular area of the help, and have a problem, and we'll have to make our best guess as to whether other people would have the same problem.

[Phil Bull (philbull)]

That would be a good solution to the licensing issue I think. If we rewrite the whole section, I'd like to plan it properly first and collect some user testing data. I'll try to recruit someone to work on it upstream. The only problems I see with that is that we'd have to convert it from DocBook to Mallard in a release cycle or two, and it might look out of place for a while.

I understand that you won't see many people looking at the help and experiencing problems, although data on that would be useful if you had it. Rather, I was referring to problems that users experience in general. If lots of them are having the same problem and it's nothing that's going to be fixed in the UI any time soon, then we should provide documentation to help people with the problem.