782698 - Make an awesome error page for Reader Mode

Status: NEW

(Toolkit :: Reader Mode, defect, P3)

Description

[Ian Barlow (:ibarlow)]

Attachment: screenshot of Pocket

I was using Pocket today, and I followed a link that wouldn't load properly in their article mode. I was presented with a very smart error page (attached), explaining that either a - the link might not be an article and to try loading it in the standard web view, or b - if users were sure that indeed it *should* be an article, they were provided a way to send the link to Pocket to help them better tune their heuristics. 

We can certainly do a good job of detecting what pages to show in reader mode on our own, but we could potentially do a *great* job if we empowered our users to help us by reporting broken pages.

I'll post some more Firefoxy UX for this soon, but wanted to report this before I forgot about it.

Comment 1

[Lucas Rocha (:lucasr)]

In practice, we'll never be getting error pages in Reader Mode because we check readability before offering reader mode on pages. We might want to add a feedback mechanism to report when reader mode doesn't show correct content though.

Comment 2

[:Margaret Leibovic]

Right now we fall back to loading the original URL if there is an error, so this isn't exactly valid anymore (we used to have an error message), but maybe in the future we want to show an error if we can't parse a page properly.

Comment 3

[Peter deHaan [:pdehaan]]

Ha, I just came to bugzilla to file this (using Pocket as an example).
I definitely think it'd be great to borrow their example when we display a "Failed to load article from page" message (see https://github.com/mozilla/readability/issues/214).

The one use case it wouldn't cover is when we aren't showing a reader mode icon in the address bar, but the user thinks that we should present a reader mode (see http://blog.chromium.org/2015/04/reaching-and-re-engaging-users-on.html). It'd be nice if there was SOME easy way to present a "report this page" UI. Currently for users to give feedback, they'd have to know about our GitHub and/or Bugzilla repos and file a bug.

Comment 4

[Micah [:mtigley] (she/her)]

It would be easier if we could change the messaging from "Failed to load article from page" to something more helpful. That or we could try refetching content (or provide an option to refresh), but that requires UX input.

Meridel, would you be able to provide some input on this?

Comment 5

[Meridel [:meridel]]

Hi Micah, a few questions:

1. Is comment 3 accurate that the use case here is: enable users to file a request to surface reader mode for a particular webpage, because in fact reader mode should be or is available and there is some issue on the part of Firefox that it is not being presented via the Reader Mode icon?
2. Does any messaging around this ^ use case already exist? Per comments 1 and 2 from many years ago, that there is no error message about this currently, right? And there is no existing messaging, error or otherwise, in the product today?
3. Why DON'T we surface the reader mode icon in these cases?
4. How big of an issue is there? What percentage of web pages should have the reader mode icon presented and don't?
5. What is the preferred mechanism to let Firefox know about this issue?
6. Who would be PM on this? Romain?

Comment 6

[Micah [:mtigley] (she/her)]

Sorry I missed this! Let me add this to my list.

Comment 7

[Micah [:mtigley] (she/her)]

I apologize for the very late response here, Meridel!

(In reply to Meridel [:meridel] from comment #5)

1. Is comment 3 accurate that the use case here is: enable users to file a request to surface reader mode for a particular webpage, because in fact reader mode should be or is available and there is some issue on the part of Firefox that it is not being presented via the Reader Mode icon?

I believe comment 3 is suggesting that an error page would enable users to report when a website they've already opened with reader mode has failed to display the article. This would be something this bug can address. However as you mentioned:

There are cases when a site should be viewable in Reader Mode, but we're failing to detect this. If this happens, the reader mode icon doesn't appear in the URL bar and there's no way for the user to report this from the Firefox UI. I'm not sure if this something we should address in this bug

2. Does any messaging around this ^ use case already exist? Per comments 1 and 2 from many years ago, that there is no error message about this currently, right? And there is no existing messaging, error or otherwise, in the product today?

No. We don't have any messaging in Firefox around not showing the reader mode icon on sites that should be viewable from Reader Mode.

3. Why DON'T we surface the reader mode icon in these cases?

There's a number of reasons:

• We've specified a site shouldn't be viewable in reader mode (eg: github, amazon, etc...)
• The article doesn't meet a minimum character count (by default this is 140 characters, I think)
• We also track a "score" that's determined by a number of things: which elements on the page can be considered, content length is too short, etc...

4. How big of an issue is there? What percentage of web pages should have the reader mode icon presented and don't?

It's hard to say. Right now we try tracking what sites are failing to show the reader mode icon from community reports on Github.

5. What is the preferred mechanism to let Firefox know about this issue?

Right now it looks like a lot reports come in through filing issues on either the GitHub repo or bugzilla.

6. Who would be PM on this? Romain?

There isn't a designated PM for this right now.

Comment 8

[Meridel [:meridel]]

Thanks, Micah. Romain, can you please advise on prioritization and timing for this request?

Comment 9

[Romain Testard [:RT]]

(In reply to Meridel [:meridel] from comment #8)

Thanks, Micah. Romain, can you please advise on prioritization and timing for this request?

Reader mode is not currently a priority given the low engagement we see on the feature - low user count (0.2% of DAU) and most users who do engage end-up spending less than 5 seconds in that mode which signals that they either hit the button by mistake or have issues with the product experience https://sql.telemetry.mozilla.org/dashboard/reader-mode-engagement
Overall I'd say that priority of this specific bug is low but we need to consider the reader mode priority overall - i.e there seems to be wider issues with reader mode given the low engagement we see.