Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?
The Internet

Redesigning The "Back" Button 356

TheMatt writes "Nature Science Update is reporting today about research by New Zealand scientists on redesigning how the "Back" button works in your browser. They point to the fact that the current "Back" is more of an "Up" in a stack of pages. They propose a system that records all pages visited. A good summary page of their efforts in web navigation (including a interesting thumbnail-style "Back" menu) can be found on their page."
This discussion has been archived. No new comments can be posted.

Redesigning The "Back" Button

Comments Filter:
  • by craenor ( 623901 ) on Monday December 30, 2002 @03:48PM (#4983226) Homepage
    come to the U.S. and apply for a gov't grant to study this...probably get $5 million a year at least.

    Important research!
  • by deathcow ( 455995 ) on Monday December 30, 2002 @03:48PM (#4983227)

    Doesnt Amazon have a patent on this??

  • WHY? (Score:5, Insightful)

    by Computer! ( 412422 ) on Monday December 30, 2002 @03:49PM (#4983235) Homepage Journal
    The average web browser's "back" feature is almost the only software feature in existence that is universally understood, and works as advertised. If it aint broke...

    • Re:WHY? (Score:3, Informative)

      by Malnathor ( 588724 )
      Actually, the way I understand it is as follows: Visit page 1, click a link to page 2, hit the back button, and then go to page 3. Now, if you use the back button from page 3, page 2 is not on the list. The back button they propose would include page 2.
      • Re:WHY? (Score:4, Insightful)

        by kawika ( 87069 ) on Monday December 30, 2002 @04:17PM (#4983448)
        I've seen people expect the back button works that way, and they've been confused when they click Back multiple tims and it doesn't show them all the pages they have been to. However, I don't see that the "new" approach offers that many benefits. The pattern of previous page visits is a tree. Any approach that tries to flatten out a tree is going to surprise (or annoy) someone. Most browsers have a History feature that lets you see where you've been and that works a lot like the proposed Back design.
        • Yea, but the problem with History (after browsing through 80 sites) is that you have to know the name of the host with the content you're interested in. Was that interesting news item on Windows on,,,,, ... Without context or clues, you have to search through a lot of history, remember or write down URLs, or give up...
    • Re:WHY? (Score:4, Interesting)

      by ez76 ( 322080 ) <> on Monday December 30, 2002 @04:12PM (#4983409) Homepage
      The average web browser's "back" feature is almost the only software feature in existence that is universally understood, and works as advertised
      This ceases to be true once you throw cache-controlling headers (which force a refresh on many browsers) into the mix. For example, on banking sites (and other sites that want transactional semantics), the Back button will often force a reload if your browser honors "Pragma: no-cache" or "Cache-control" headers.

      Also, as a previous poster pointed out, the back button also works unintuitively (compared to, say, the standard edit menu Undo function) when you browse to a new page from a page to which you've clicked back (works more like a tree than a chain in that case).
      • ez76 comments:
        Also, as a previous poster pointed out, the back button also works unintuitively (compared to, say, the standard edit menu Undo function) when you browse to a new page from a page to which you've clicked back (works more like a tree than a chain in that case).
        On the contrary, its exactly the same. Thae the following example:
        1. User types "important stuff"
        2. User hits UNDO
        3. User types "stupid stuff"
        4. User wants to go back to "important stuff"
        Too bad for the user. UNDO works pretty much the same way that the BACK button does (or vice versa).
        • Re:WHY? (Score:3, Insightful)

          by dan g ( 30777 )
          Well I guess you're both overgeneralizing. A quick test shows emacs's undo will redo previous undos, but word will not.
      • For example, on banking sites (and other sites that want transactional semantics), the Back button will often force a reload if your browser honors "Pragma: no-cache" or "Cache-control" headers.

        This is not correct behaviour. The HTTP 1.1 specification [] specifically states (section 13.13) that cache control should not prevent the viewing of stale pages in a browser history.

    • Re: WHY? (Score:2, Funny)

      by Black Parrot ( 19622 )

      > The average web browser's "back" feature is almost the only software feature in existence that is universally understood, and works as advertised. If it aint broke...

      The problem is that "back" is the wrong way when you're on the other side of the equator.

    • Surf to a page, click back a couple times and then click a link on the page you're looking at. Now, you can't get 'back' to the pages you looked at before you hit the back button.

      It's really irritating. In order to get around it, you need to open a new window, which clutters up your desktop.

      Perhaps a better back button would mean a lot less windows open for experianced surfers.
  • Sorry but, (Score:5, Insightful)

    by llamalicious ( 448215 ) on Monday December 30, 2002 @03:50PM (#4983243) Journal
    For the labelling:
    I prefer to think of my "back" button as working like a paper book. I generally don't flip pages "up" when going to a previous page, so the "back" terminology is friendly to me.

    As for the idea:
    All I really need the back button to do, for better efficiency, is to skip posted forms, that's all I want. What did I miss in that article that really make their system stand out from stacking? I like my stacks dammit.

    • Re:Sorry but, (Score:3, Insightful)

      by SanLouBlues ( 245548 )
      Then think of their design as a back button for "choose your own adventure books". So if you want to easily flip between having jumped through the dimensional portal and having gone to look for your missing friend you don't have to reread the original forking page.
  • Bah! (Score:2, Funny)

    by oGMo ( 379 )

    Bah, I thought of this years ago. Back buttons suck. So does the whole linear web browser model. I mean, it's the web, right? Why is it always back and forward? Why don't we see a web (graph) view?

    I always wanted a web browser called "Sting" that displayed stuff like this and let you "cut through" the web. ;-)

    • Re:Bah! (Score:2, Interesting)

      by Milo Fungus ( 232863 )
      Exactly. Read Weaving the Web by Tim Berners-Lee [] for a good explanation of the origins of the WWW. A lot of the Web as we know it is a hacked version of Berners-Lee's original vision and intent. People weren't willing to make client programs text editors, so the Web became a publishing medium viewed with a browser instead of the interactive medium it was intended to be. It would be easy to display your browsing history as a hierarchy or as a link web, but it would probably take up more space on your screen. Display space is at a premium. (I wish Mozilla would let me interactively resize the tab buttons, for instance. They take up too much space.)
      • by oGMo ( 379 )
        It would be easy to display your browsing history as a hierarchy or as a link web, but it would probably take up more space on your screen. Display space is at a premium.

        Yeah, this is why i see having a "full screen" (or however big your window is) graph view, and when you click on a node, the node replaces it. If you click on a link, it would "iconify" the content back to its node on the graph view, add it to the graph, load the new content, and then show it to you.

        It's interesting about the interactivity bit. I've seen various projects that do similar things, but most are toys/proofs-of-concept or not very developed (like the wiki stuff, although I haven't looked at that much so maybe it's more advanced than I realize).

        Thanks for the book reference, I'll check it out sometime.

  • Not good (Score:3, Insightful)

    by teslatug ( 543527 ) on Monday December 30, 2002 @03:52PM (#4983267)
    On average, the two systems worked about equally well.
    Then what's the point of changing it. It seems to me it would just add more confusion and frustration.
  • no.. it is "Back" (Score:3, Insightful)

    by trefoil ( 153310 ) <brents@easystreet . c om> on Monday December 30, 2002 @03:53PM (#4983268)
    when you start a browser.. it begins on your "home page" from there you may jump from site to site.. not necessarily deeper into a website, but more often than not, it is. So to me, the "Back" button has to do more with "Back Tracking" as in taking a hike, and back tracking towards "home".
  • by Bonker ( 243350 ) on Monday December 30, 2002 @03:54PM (#4983274)
    Is what I think they're looking for.

    Lave the back button alone. It does what it's supposed to perfectly well. As long as it's not applied to file-systems or any other PC arcana, it's perfect for the task.

    If you want to make something that works for both file-systems or GUI shell browsing and web browsing, design a new tool. Don't overload the existing tools and make them useless for both tasks.

  • Does anybody remember the OS/2 Warp (3.0) system web browser? I vaguely remember a really nifty tree display for page history that would show everywhere you were at one time and everywhere you went from there.

    • by RetroGeek ( 206522 ) on Monday December 30, 2002 @05:07PM (#4983773) Homepage
      Does anybody remember the OS/2 Warp (3.0) system web browser?

      I do.

      It had a complete Web page showing all the links and hyper links that you visited.

      The first page was at the root level, then each link from that page was nested, with each subsequent link nested in turn. Each link was shown with the page title and was a link so you could re-visit that page.

      After a few hours it was interesting to see your browsing process. First you were here, then you went there, and there, and ...

      I miss that feature. It showed Web browsing in a non-linear fashion.
      • Ditto; it was a tree structure stored as an html page.

        One could save an entire session just by bringing up the tree in a browser window and saving it. That meant that, for instance, one could work through search results or links on a page one-by-one, hitting "back" one each page, and end up with a storable record of everything visited.

        Those were the days, when web browsing was considered something more than a Markov process.
  • by Gorm the DBA ( 581373 ) on Monday December 30, 2002 @03:55PM (#4983292) Journal after reading the article, I see that they're trying to make it so that users can get back to the main page easily so they can get to the information they want.

    Why not make it *really* easy and develop a "forward" button that would actually take you to the piece of the Mega-pagecount-poorly-indexed-searchbuttonless web portal of doom that you're really interested in? They could call it the Psychic Fast Forward or some such.

    Base it off of all of the Total Information Awareness data that the government wants to gather about us, so it predicts what you want.

    And then place locks on your browser so that you really only want to go to the major sites.

    Then eugenically engineer society so that you don't even know that you ever wanted to go somewhere else.

    NOW we're making the web useful!!!!!!!!!!!

    • You would think with all the annoying things you can do with Javascript, they could incorporate a way to pre-populate "forward" history info, so that could be done.

      Or has it been done?
  • by teamhasnoi ( 554944 ) <.teamhasnoi. .at.> on Monday December 30, 2002 @03:57PM (#4983302) Journal
    IE has an annoying habit of clearing the text boxes of a page when I get a timed out page and hit the 'back' button, say when posting to /. (slower than ever!?)

    Chimera and Phoenix keep that information in the box, saving me from having to copy the text, just in case.

    A feature I would like similar to 'back' would be to reopen the last page I was on when I last closed the browser. Often, I close the window and find that I still need some info that was on that last page. I hate browser history ie: I have that turned off, so I can't hunt through the history to quickly find the page.

    That feature would be nifty. Or something to make me less of a spaz.

  • If it is not broken don't fix it!
  • What about Forward? (Score:3, Interesting)

    by Remik ( 412425 ) on Monday December 30, 2002 @03:59PM (#4983319)
    I'm actually more interested in the possibility of redesigning the functionality of the forward button.

    In the current implementations, the forward button loses it's registry once you go back/up and then click a link. It's kind of like creating a new time line in your lose all the pages you had been to in the previous line...before you went back. Why should it be that way?

    • Why should it be that way?

      Given that the forward button can only take you to one place (unless you want it to open a bunch of new browser windows when you click it), how else would it work, logically?
      • Exactly. There are 3 cases with no obvious alternative behaviors besides mind reading... would someone like to explain what else it could do at these points?

        A) You click the back button. A back operation is performed. The forward button enables, allowing you to undo the back operation.

        B) You click the forward button. A back operation is undone. The forward button is enabled only if there are more back operations to undo.

        C) You click a link on a page, which navigates you to a new page. The forward button disables itself, because it doesn't know what you might click next. The browser could possibly preload a predicted page, however that would be different functionality than in A and probably should be a different button altogether... the same functionality could be done (and is already being done in some browsers) w/o a button at all.
        • No mind reading required.

          A) Assuming forward is active because we have just done one or more backs, it would activate with a list all of the routes previously taken from the current point.

          B) It is also active if a previously followed link path from the current page exists.

          C) Correct. Forward cannot be active after following a fresh link, but back is inactive before you start browsing. How is this relevant?

          I don't presume to have it all worked out, but I think my post to the parent presents a start.

    • I'm not so sure about the retooled back button either, but I've also thought about / wished for an improved forward button. The current forward button only provides the most-recently-used forward line, e.g. if you start at A, link to B, link to C, link to D, back to C, back to B, link to D, back to B, then only D will show as a foward option. I would like a forward button to become a hierarchical menu of previous browse lines. You would have to select which browse line and how far along it to proceed. For example, from my previous example, upon the third visit to B, the forward button would present a menu showing the two paths previously taken from B:

      -> C -> D
      -> D

      I think you could do something similar with "back".

      Of course, if I was really motivated, I'd jump into one of the plethora of OSS browsers and implement this, but instead I'll just make like a typical /.er and complain. ;-)

    • Indeed. I've had a forward button in every browser[1] I've used since
      1994, and I've yet to discover a use for it. Is it in case you hit
      back by mistake, or what? If I were redesigning the web browser
      toolbars to conserve space, the forward button is the first thing I
      would drop. (Well, that's assuming you've already turned off the
      things Navigator's prefs dialog lets you turn off easily, such as
      print and home.) I'd probably remove stop too and use the extra
      space to make the history button a first-class citizen. (Oh, that's
      another thing about the forward button: the history list gives you
      all of its functionality plus a great deal more. This is also true
      of back, but back you use so constantly that it needs to be easy to
      hit quickly.)

      [1] Except for non-GUI/non-mouse browsers, which have an equivalent
      keystroke that does exactly the same thing. Oh, and telnet to
      port 80 doesn't have it either.
  • By using script to change it to a "Stay-Here" button. Those are the sites you make a point of never bookmarking, or ever intentionally visiting again.
    • If you find a site where that happens in Mozilla, please file a bug. Mozilla's policy is that fast redirects should not add an item to session history, where "fast" means something like "under 10 seconds and not in response to user input in the intermediate page". It's already fixed for most cases in Mozilla. One case that was fixed recently was a redirect in an onload handler (bug 124245).

      Btw, in most cases, sites did not break the back button intentionally. They were just trying to redirect from one URL to another, and didn't know that Netscape 4 required you to use a specific redirect method in order to avoid leaving an entry in session history.
  • by tps12 ( 105590 )
    Yeah, the Back button is "up" for a down-growing vertical stack, but it's also "left" for a right-growing horizontal one. They're each equally intuitive and consistent, and the "left" model seems to be a pretty well-entrenched standard. I don't see any reason to mess with what is probably the easiest to use UI element in any modern web browser.
  • by Lxy ( 80823 ) on Monday December 30, 2002 @04:00PM (#4983330) Journal
    The back button is fine the way it is.

    If the Back button takes me to where I've been, why doesn't the Forward button take me where I haven't been yet? I want a button that takes me to where I'm going to go before I ask it. Is that too much to ask?
    • Check out the fast forward button [] in Opera 7. If it detects a "next" button on the page your looking at (ex. Google search results) it detects that and allows you to jump to it. I'm not sure if it pre-fetches or not. Kind of neat.
    • ...Back to the Future? ;)


    • Actually... (Score:2, Interesting)

      Though this is supposed to be a "funny" post, what you're suggesting is exactly the problem. The back button functions perfectly fine. It's the forward button that needs work.

      Programmers simply need to rethink the history of page clicks as a tree instead of a stack. Navigation back on a tree always takes you to a root. It is at that point when the user should have the option of selecting different branches that have occured. For example:

      1. Start at Yahoo
      2. Read a news article
      3. Go back to main page
      4. Go to Slashdot

      Now, at this point, if you hit the BACK button, it should take you to Yahoo. When there, however, the FORWARD button should offer you the choice of jumping to the article you read, or going to slashdot. That would solve the problem nicely. Except, if you do a lot of browsing, that dynamic tree could get awfully big in memory.
  • This is news?

    First, the back button in IE and Mozilla has a drop-down that will show the previous 9 or so pages.

    Second, there is the History button/menu, which will display a full listing broken down by site and date.

    Maybe some of these "academics" should actually pull their heads out for a look at the real world now and again.
  • Maybe I haven't had enough caffeine today, but I'm not understanding what this team has changed, exactly.

    Every browser I've ever used has a back button that takes you through the history of every page you've visited, not just index pages.

    What am I missing here?
    • > I'm not understanding what this team has changed, exactly.

      Indeed. In particular:
      > They have replaced the current stacking system, which only records
      > index pages, with one that records every page in the order it was
      > visited.

      Huh? What current stacking system are they talking about? I have
      NEVER seen that behavior. Nor would it make any sense whatsoever;
      the web is designed around the principle that all pages are
      horizontal from one another (which is why it's an href, not a vref).

      Some experimentation has been done with the concept of "Up", either
      by using the link tags or by s/[^\/]+\/?// URI trimming, but while
      both have theoretical merit, such a small percentage of the web is
      structured in the expected way that these features turn out to be
      you never use or something you use very rarely (respectively).

      As far as recording only index pages (does that mean only index.*,
      or does it mean something else?), I thought about that for almost
      four seconds, but in the end I concluded that it's imbecilic.
  • When you hit your back button and select another link, you're really branching the tree of visited sites.

    It makes more sense to have an explorer-style tree view than a history. That way you can navigate a site and still have an idea where you have been.

    • I was going to post a similar comment.

      You're totally right about the functionality required. The problem I see here is user interface. While a tree diagram is quite easy to follow, a tree similar to the file browser one does take up rather a lot of space.

      Thinking about it, I tend to use Mozilla's tabs as a means to launch several links from the same page, which allows me to flick thhough them, and return. This allows most of the functionality, but it does get confusing remembering where the pages were linked from. Perhaps what we need is a nested tabs view or something.
    • Ok, shoot me for replying to myself, but I thought about it more and it occurred to me that the best UI tool isn't a explorer-style tree, but a multiple-level menu--similar to existing back-menus, but branched:

      Page3 (last visited) | page4 (linked from page3)
      | page5 (linked from page3)
      | page6 (linked from page3)
      Page2 (linked from page1)
      Page1 (started here)

      You are currently on page7, linked from page3. As you can see, it only branches when the back button is hit on page4,5 or 6, and you choose another link on page3. So back-button behavior is preserved, but enhanced to prevent information (clicked link) loss.

  • OS X panel view (Score:2, Interesting)

    by jpsst34 ( 582349 )
    I'm not an OS X user. But I wish I were. I really like the panel view (or whatever it's called) in the file browser. With every click, it shifts the panels to the left, and adds another at the current location. This gives a great visual view of history and allows you to sort of back up to the last wrong turn and go in another direction. It kicks the ass of the MS tree view. Something like this would be great in a web browser.
  • Big Deal. They can change it all they want, but my dad still won't be able to figure it out.
  • after you read the article...

    make sure you go "up" to slashdot, rather than "back" to slashdot to post a comment...


    they need to backup and rethink their verbiage (pun intended). psychologically, the way the human mind thinks of time and travelling, the back button just makes too much sense.
  • This sounds an awful lot like the Mozilla site navigation bar [] to me. This was removed right before 1.0 was released, which is why I'm still using a pre-1.0 version of Mozilla myself (well, that and the fact that Mozilla was already rock solid before 1.0).
  • So install the Google toolbar [] if you're using IE, or the Mozilla variant [] if you're using Mozilla, and use the "up" button provided there. Whee.
  • Of course they were. The subjects are always extremely enthusiastic about new things they've never seen before. I would have a better chance of believing the paper if they would have had a few "extremely UNenthusiastic"s thrown in. As the paper stands, everyone liked pretty much everything that was thrown at them. That is BOGUS.
  • Instead of a back button, create a belly button.
  • by Randolpho ( 628485 ) on Monday December 30, 2002 @04:16PM (#4983441) Homepage Journal
    The first thing I thought as I was reading the article was, as everyone else has commented, "how is this different?"

    It really *is* different; the problem is that the article explains things very poorly. Here's the difference:

    With normal browsers, when you click the back button to a previous page, and then follow a new link on the previous page, the page you were on before you clicked the back button and followed a new link, is removed from the list. This is what they mean by "stack" behavior.

    What these guys are proposing is that every time you visit a page, it goes into the back list. Thus if you are on, say, page 2, and click the back button to page 1, then follow a link to page 3, the list stored in the back button is 1 - 2 - 3, and you will go back to page 2. In the current system, the list stored would be 1 - 3; page 2 is gone from the list and no longer available via the back button.

    So now you know. Regardless, this behavior is already available in I.E. 5.x and above via the History explorer bar. A simple sort by Order Visited Today gets the list exactly as proposed by the article. Except for the thumbnails, however, which is a very good touch.

    Personally, I think it would be best to have *two* such buttons; one that has stack behavior (current "back" button), and another that has the proposed temporal behavior; perhaps as a history pull-down menu.
    • See, here's some programmers clearly overthinking the problem, and not understanding how the stuff is employed in the real world. This is why this stuff is best left to UI designers.

      They're viewing the forward/back as popping and pulling off of a stack. Your average nontechie has no grasp of what that means, to them, forward/back is analagous to the path you took to get there.

      This morning, I left my home and drove on the highway (1), and half asleep took the wrong exit (2). I went back to (1) and continued to work (3). Later when I reverse my route (by going 'back'), I dont want to go to 2 again. The path I took is (1)-(3), the reverse of that is (3)-(1)

      The back/forward analogy is perfect as it is.

      What these guys describe is (in english) a previous/next or earlier/later feature, not significantly different from the history menu/bar.

      And Up/Down is navigating a fixed tree structure (going Up from yields
    • In my experience, the back button is almost always used for "no, that isn't what I want, go back and try something else." The previous pages you went back from are very rarely relevant. Tabs or windows can cover other uses, assuming the user is aware of them and they're easy to use/understand.

      Thus if you are on, say, page 2, and click the back button to page 1, then follow a link to page 3, the list stored in the back button is 1 - 2 - 3, and you will go back to page 2.

      This alternate behavior would be a nightmare. (and I'm a programmer, so stfu you UI design hippies ;)

      Let's say you wanted to repeat this, and go to page 4 (also linked off of page 1). You'd have to go back to page 2, then back to page 1 before you could get to page 4.

      Now suppose you wanted to go to page 5. You go back to page 3, then back to page 2, then back to page 1. As you can imagine, this quickly becomes unusable. You have to keep going back through everything you didn't want.
    • That would, to put it politely, suck. I don't want the browser to forget that I was on page 2 at one point. I might want to get back to it again. I assume what you meant isn't that page 2 is GONE from the history, just that it isn't stored in the history multiple times, and is just there the first time. (So if you read pages 3,4,5 from page 2, you normally get a history of 1-2-3-2-4-2-5 and this thing would collapse the redundant 2's so you have instead 1-2-3-4-5. There would still be a '2' in there, but not each time you go back and revisit.)

      Personally, I'd hate this. If I want to get back to '2' to see the next link on it (to page 6 perhaps), I want to just go back to the most recent step where I visited '2' in the history, not all the way back to the very first time I visited it ever, which if '2' has a lot of links on it I've been reading through, could be buried quite deep.)

      The problem is that in reality you browse through the web as a tree of nested links, but the browser only remembers this as a one-dimensional list, not as a tree. It will always be ugly to try to mash what in the real world is a tree into a data structure that is only a list. The only real fix is a user interface that presents you with your browsing history as a tree rather than as a one-dimensional list. This might be implementable through cascading menus when you click-and-hold the back button rather than just a single list.

      So you might see something like this:

      1. Page 1
        1. Page 1.1
        2. Page 1.2
      2. Page 2
        1. Page 3
        2. Page 4
        3. Page 5
      (Imagine the above done as a cascading pulldown menu. Slashdot filtered out my attempts to create ascii-art to show it the right way.)
      What these guys propose is worse than what we have now, in my opinion.
  • Apparently the author of the article has never heard of the history button in IE.

    That reminds me of an article I once read on IE 5 in which the author said "I wish the favorites were available via a drop-down menu like in Netscape." Sheesh...
  • If I'm browsing through a set of pages, and I go back three pages then click on another link, those three pages disappear. The problem is slightly alleviated with browser tabs, but those tend to clutter up my screen during serious surfing time.

    Just because it isn't broke doesn't mean it can't be fixed. Windows is universally understood but that doesn't mean a more powerful solution can be found/hould be used/be optional for those who can handle it.
    • If I'm browsing through a set of pages, and I go back three pages then click on another link, those three pages disappear. The problem is slightly alleviated with browser tabs, but those tend to clutter up my screen during serious surfing time.

      I would argue that anyone who uses the back button *wants* those 3 pages to disappear due to errant navigation. Perhaps a toggleable setting could enable a linear history, but I bet after a while you'd end up using your tab method anyway.

      (Note you can keep tab/window use down to 2 by dragging links from a desireable navigation page onto a second, but preexisting, destination page tab/window. I feel this is way more usable than any other back button functionality or options could be... and it is doable in pretty much any browser, even old ones.)
  • by Zinho ( 17895 ) on Monday December 30, 2002 @04:25PM (#4983498) Journal
    The Mozilla project has had people working on this for almost 3 years now, see bug #21521 on Bugzilla (they deny links from Slashdot, so I won't even try). Unfortunately, there are technical problems that can't be ignored when designing a system like this. One of the stickiest problems is the fact that, as you browse, the history of where you went becomes larger and larger - it starts to act just like a memory leak. Using menu items for this (like the go menu or, I think, the back button's menu) makes the memory problem worse, since menus are memory intensive. There are also cross platform compatibility issues to deal with.

    The article mentions the non-technical issues as well: "Unsurprisingly, it's harder to return to index pages with this system - so it's easier to get lost in big websites. New users tended to solve problems either very efficiently or very inefficiently." I believe that this is one of the bigger problems the developers of more advanced navigation systems face, how to provide controls that afford the user good access to the information.

    I wish them luck. And if you want to see something like it in Mozilla, please vote for bug 21521 on Bugzilla. It's only got 7 votes, which is pathetic.

    On the other hand, if no one cares, perhaps the answer really is to just let it drop. Once again, I wish them luck.
  • For those who still think they're not coming up with anything different...

    Basically they want a two-dimensional navigation button(s), not the current one-dimensional ones (back/forward).
  • I'm confused, all this research shows that people use the back button VERY frequently. I personally almost never use the back button. So what is it that people are doing that require the back button, maybe solving the need for one would be a better idea. Are people trying to visit list of links? How about a more obvious tabbed browsing UI. What are the other uses for back? Can anyone tell me. Personally long before tabbed browsing was available I always opened in new window. So I really am guinuinly confused as I havn't used the back button in years.
  • Have both kinds (Score:4, Informative)

    by iabervon ( 1971 ) on Monday December 30, 2002 @04:34PM (#4983538) Homepage Journal
    Your navigation is actually a tree (or a graph if you consider a page to be the same regardless of where you get there from). The conventional "Back" button goes up the tree, which is the simplest operation for going toward the root, and quite useful.

    The real problem is that the conventional Back and Forward buttons, between them, don't let you traverse the entire tree, but only the right edge. There needs to be some way of getting to the other pages (for example, I'd like to take another look at the article; I can't navigate there in my history, even though it was on my screen two documents ago, nor can I get there from here without either starting from my bookmarks or losing my comment). They use a button which essentially is an "Undo" for following links.

    Their results follow from the ability to access your entire history rather than only the right edge, along with using an operation that is frequently the same as the usual design (if you follow a chain of links down, and then go all the way up); this suggests that an approach which retains the regular Back button and adds an "Undo follow" button to go to the document you were on before. Since Forward is relatively rarely used, it could reverse both of these operations, depending on which you did (i.e., undo history navigation).
  • > They propose a system that records all pages visited.

    "Good lord, man, you've invented the history list!"

    Chris Mattern
  • I tried combining the Back function with an Up function. Every time I pressed the new Back-Up button, though, my computer would just start making beep beep noises.

    Never try that again.

  • Screw the back button, what I want is an intelligent forward button.

    Say you're looking at the page:
    and then you go to the page:
    At that point I want the browser to request
    in the background. If it is availible, the forward button should become clickable and take me there.

    The code could check for predictible changes between pages, and if it thinks it's found a pattern, it requests it. If the page is there, it turns on the forward button.
    It could also be set up to jump to the next anchor in an HTML document, if any exist.
  • The history I thought was kept separate from the back button list because it would be a pain to go through your entire history when you are looking for a page you just recently left, etc. I'm not going to say it is a bad idea though, and maybe they have a neat idea to make it work, I dunno. This reminds me. My manager told me something I'm trying to get the hang of. When I hear someone doing something stupid I'm not supposed to say anything. Then when they screw up miserably, then I tactfully present to them in writing why their idea sucked. It is much harder to do than it sounds.
  • There is an analogous problem with Undo/Redo in editors.

    If you make a change, Undo it, then make another change, the Redo functionality is gone for the first change. The first change you made is irretrievably lost. At least in most editors.

    BTW the article says the Back button "accounts for 40% of all Internet clicks." This might be true for IE users who don't have tabbed browsing (and the article shows a screen shot of IE's back button). I've seen IE users find a bunch of Google matches, click on one, go back, click on the next, go back, etc. I don't see how they can stand it. (Yes they can open new windows but that can be annoying in its own way, and they usually don't.) With Mozilla's tabbed browsing I hardly ever use the back button.

  • All this talk about the back button is interesting, but it seems to ignore one of the biggest failings with todays web browsers, and that is the whole page based metaphor. Now it works great for content that lends itself to it, but it sucks for the ever increasing sites that use the browser as an application front end vs a simple content reader. For anyone whose done any application creation in html/http you know what type of nightmare exists trying to keep track of user sessions and making your app "back button" proof.

    What browsers need is a more robust control mechanism that allows the site to control exactly what happens when the navigation buttons are pressed. Moving around in an ecommerce site is a lot different than a message board. Now I'm not saying make it a free-for-all, but people do expect certain "logical" behaviours and many are smart enough to deal with minor shifts in absolute behaviour depending on context. This combined with other improved navigational aids (e.g. like the article, better history, etc) would make EVERYONES life easier.
  • I really don't see what these guys want to accomplish. How does "back" not really go back? The article seems to imply that "back" understands the idea of a site hierarchy and will literally only go "up" in that hierarchy (whatever that means... Toward the root of a site?):

    Cockburn and his colleagues reprogrammed web browsers so that their back button was based on the order of pages, not their hierarchy

    See? This just makes no sense. When I click "back", it goes to the previous page I visited. In chronological order. No "hierarchy" involved. The linked article seems to imply otherwise.

    Now, if they mean that non-server content, such as the state of a running JS/Java program, or user entered data, will not persist, I can understand their point, but wouldn't *WANT* it to stay around for others to find on my machine later.


    The order-based back button was good at navigating between distant pages.

    Now here, I've definitely missed something. How does an order-based back button make it *easier* to go between distant sites? A hierarchical button (if such a thing existed) would do that better.

    Overall, either I missed something *REALLY* fundamental in what these guys did, or they did nothing and obscured that fact with lots of talk about irrelevant relational concepts. From the words they chose to use, a back button *already* behaves like what they want to change it to, and the supposed benefits of their change fit better with the behavior they claim the back button already has (which it doesn't).

    As the best credit I could give them from the article, I could assume that the author completely misunderstood the research and reversed the two concepts. Thus, the research would actually have the "new" button behavior using a hierarchy of sites, rather than strict chronological behavior.

  • Whether the metaphor is "Back" or "Up" is immaterial if the underlying capability is the same. And that capability hasn't appeared to have changed since Mosaic. Prior to the web's commercialization, when most sites were static, the "Back" button wasn't so annoying. Moving retrospectively through too much useless crap is as much a waste of time as trying to use a browser to find something in the first place.

    The "History" file won't cut it, either, because it, too, forces you to move backwards through everything.

    Often, I want to move back to a specific URL I saw earlier in a session, but I don't want to bokmark it. How about allowing users to build their own breadcrumb trails?
  • The anti-/. effect button.

    Ok, I've got nothing.
  • do I really want a graphical display of all the porn pages I've visited recently sitting out in a tray on my browser?
  • What would REALLY be nice, is an entirely new function/button called "Tree View" that would include all the URL information from all the browser windows used during that session and map them to a branched tree.

    Possible features might include using different colors for urls visited in different browser windows; A zoom in/out for deep detailed tree viewing; hover over a URL and get info like when you were there and for how long, etc.

    I'd really love to see what my tree looks like at the end of a furious browsing session. Aside from being a practical browsing tool it could even help improve technique and shed light on surfing habits.

    Maybe I could even learn not to get sidetracked so darn much.
  • What we really need is for browsers to support a set of standard site-specific navigation buttons scriptable by Javascript. The amount of coding that goes into stateful sites to deal with users who hit the back button at "inappropriate" times is enormous -- anyone here who's coded a shopping cart will know exactly what I mean.

    The benefit for the user would be a clear, standard set of buttons -- as opposed to the often confusing, overly "creative" navigation third-rate designers foist upon us -- and fewer confusing errors.
  • Umm.. no, unless you want to redefine terminology, up would move you up a folder on a web server, like the 'up' feature on konqueror.

    The back button system may have its problems, but it is far from incorrectly named.

Q: How many IBM CPU's does it take to execute a job? A: Four; three to hold it down, and one to rip its head off.