Some old notes on the role of tabs in web browsers.
What's wrong with tabs?
- Screen overflow. Tabs flow offscreen when a high number are open. This means that all the open tabs cannot be seen or selected. It also makes the display of newly opened tabs difficult.
- Difficult to navigate. Finding a specific tab can often be difficult, particularly when a large number of tabs are open. Only a small amount of information is available for each tab.
- They require planful action and forethought (when being used to mark pages which might be returned to, for instance).
- Users are required to close unwanted tabs - tabs create management tasks for users.
- Their purpose overlaps with existing browser functionality (eg. history and bookmarks) but does not integrate with it.
- Some uses of tabs are not easily discoverable (e.g. opening links as new tabs).
- They add another layer of interface complexity.
- They are hidden from many window-switching mechanisms, particularly GNOME Shell's overview.
The uses of tabs
- Page history - a way of marking pages for revisitation
- Organising and grouping pages
- Queuing pages
- Comparing pages
- Page persistence
Tabless design experiment
(These are comments on the interface outlined in web-overview.png and web-webpage.png)
It might be interesting to have an interface that is similar to the Documents app, but the nature of the Web is very different from the user's personal documents. The Web it is a huge collection, largely unorganised; browsing it is partly a matter of making sense of that information and adapting it to one's needs and interests.
All other desktop browsers are using tabs and they seem to be doing quite fine, so I think that we would need a good and grounded rationale to move away from that. I am not saying that we can not do it, just that we would need to have a good understanding of why that would be a good idea. Switching between open tabs is the second or third most common action in the web (the most common is following links) and this design proposal would make it harder to do so.
One of the benefits of tabs is that they act as reminders. If they are not immediately visible, they would need to be kept in the user's working memory. This would be made worse if we sorted the open-pages list by access time: the list would change every time, so the user would always need to scan it to find the desired page. Episodic memory would not work ("I opened this page and then that one") and spatial memory would not work ("work-related pages are towards the left side"). With many open tabs, this design has a very real risk that some of them would simply be forgotten over time.
It is true that tabs fail when the user has opened too many, but it is not quite clear how this design would improve this corner case. Nevertheless, there is certainly room for improvement on the current status of browser tabs. There seem to be marked personal differences among users: some people rarely switch tabs, whereas others use them profusely to organise their browsing. It is true that tabs-on-top fail these power-users; maybe we could provide a second UI for them, for instance by allowing to place tabs on the side and auto-organising them in trees.
Having a Home page/tab/... to organise browsing, like Firefox and others do, is a good idea. It allows us to provide a custom place from where to begin browsing the web. It could have a central URL+Search bar, plus views for favourites, recent, read queue, bookmarks, history... It is open to debate if this view should let you organise your open tabs as well.