Slashdot Mirror

← Back to Stories (view on slashdot.org)

How To Deal With The Spatial Paradigm

Posted by timothy on from the tradedja-for-a-paratwenties dept.

PostThis writes that there's been "a lot of talk about Gnome's spatial Nautilus lately and so Christian Paratschek puts everything into perspective weighing in the pros and cons of this particular user interface paradigm. In any case, there are always alternatives."

24 of 38 comments (clear)

  1. How to avoid the debate alltogether... by Anonymous Coward · · Score: 1, Insightful

    Direct-access user interfaces, like Apple's Spotlight, Black Tree Software's QuickSilver, ObjectiveDevelopment's LaunchBar (all for Mac OS X) and Candy Labs' AppRocket (for Windows), are the future of file management interfaces.

    The spacial vs. browser-style debate isn't worth winning, because either way you're sticking to metadata-ignorant heirarchies that humans just aren't very good at dealing with beyond a certain point.

  2. Even better than the real thing by Anonymous Coward · · Score: 1, Interesting

    As far as I am concerned file managers like Nautilus, Windows, and Mac OSX can put as much effert into their work as they please, but they can never get the functionability of a "two plane" file manager like Gentoo:

    http://www.obsession.se/gento o

    There are many of this sort, but Gentoo is by far the best. Sadly doesn't the author like GTK2, so he won't port it=(

  3. It's all about BeOS... by curious.corn · · Score: 1

    ... and it's sweet desktop (notice the unix screenshots?) metaphor. Spatial is good, but it's counterpart is metadata driven virtual folders; extended attributes aren't only for acls... A userland daemon, fam monitoring and a berkleydb hanging around. It's not that the pieces aren't there...

    --
    Mi domando chi à il mandante di tutte le cazzate che faccio - Altan
  4. What I dont get... by Kusanagi · · Score: 1

    ..is all the hooplah over this issue. It's not like this is a new paradigm. Even windows used to be spatial - win95 before they came up with the whole browser integration thing.

    In Linux, I still prefer to use the command line. I'm just a keyboard whore. :) To me, "mv * ../foo" requires less effort than the GUI equivalent.

    My first real computer (trs-80's don't count) was an Amiga and it used a spatial interface - tho I preferred to use Directory Opus, an ol' midnight commander style file manager, for my hardcore file management.

    Even on my dad's DOS machine, I primarily used Stereo Shell (another of the MC style).

    So, whatever happened to that paradigm? :) Side-by-side lists of files, each navigational, but a lot more useful that the web-browser-like navigation. People are forgetting that we used to use file managers to manage the files and only sometimes open them - most of the time files were opened from the program that we were wanting to use them in. When was the last time you launched an application and THEN opend the file with the 'Open' option in the File menu?

    --
    -Major Kusanagi, Section 9
  5. The only mistake by Cycon · · Score: 3, Insightful
    ...was requiring the user to use gconf to put it back.

    It's bad enough to change default behavior on a user (at least it was during a major release) but all they had to do was add a preference to "open new folders in the same window" .. the same way windows does, and a lot less people would be upset.

    Personally, I think the spatial idea is pretty useful when you have multiple monitors, and lots of space to spread out the "remembered" window locations. On a single screen the benefit just isn't as big.

    --Cycon

    --
    Your Brain + EEG + LEGO Robots = Brainstorms
    1. Re:The only mistake by Anonymous Coward · · Score: 1, Informative
      You know, Mac OS9 (and earlier) was like this. But they had a real simple solution... if you held down 'option' when opening a folder, the new folder opened and the old folder closed itself. This was remarkably easy to do, since you tend to use a mac with one hand on the mouse and one on the keyboard anyway (control-click for context menus, if you have the single button mouse).

      With OSX they just changed the behavior, which you could change in a preference.

      With Windows I think it's what... hold down control when you open a folder to have it open in a new window?

  6. Newbs by Laxitive · · Score: 4, Insightful

    So, I read the article. And the author makes a reasonable point about newbies being more accustomed to the spatial metaphor. I'm not going to dispute that. It might or might not be a valid claim.

    The question I want to ask is: what about those of us that are NOT newbies? The author states early on that he tries to avoid anything that would expose the filesystem tree abstraction to the end user. Maybe you could argue that it is good for newbie users, maybe not. But it DEFINITELY isn't good for non-newbie users.

    Look, the filesystem is a TREE. That's what it IS. Any metaphor that you try to make the filesystem fit some other pattern will only take you so far. A tree is a very nice, clean structure. A lot of its expressive power is lost when you try to impose some strange alternate metaphor on it. When you deal with the filesystem as a tree, any operation that maps well onto trees, you can map well onto filesystems. It's a powerful abstraction.

    And quite frankly, don't we WANT newbies to be learning the actual behavioural properties of the tools they use, rather than an artificially constructed interface which we deem them more able to use? Won't this lead to more intelligent users?

    Personally, I think it's insulting to people to say that they can't "deal" with basic abstract structures. It's not THAT complicated guys. And we're not that smart for knowing how filesystems work. MOST people in the world can grok the concept perfectly fine, you just have to teach them. Perhaps some people feel threatened by that?

    I was an avid gnome user. I stopped using it once I noticed the clear trend for gnome to assume that I'm dumb. That I can't deal with certain choices - which are better made by the developers than by me. Limiting excess in choice is fine.. but there is a fine line between reasonable limits, and top-down control. I think gnome crossed the line a ways back.

    Keep your spatial browser. I'll keep my trees.

    -Laxitive

    1. Re:Newbs by harves · · Score: 1
      Look, the filesystem is a TREE. That's what it IS. Any metaphor that you try to make the filesystem fit some other pattern will only take you so far. A tree is a very nice, clean structure.
      And quite frankly, don't we WANT newbies to be learning the actual behavioural properties of the tools they use, rather than an artificially constructed interface which we deem them more able to use?

      A computer is a tool, and should be approached as such. A user (newbie or otherwise) should not be forced to adapt their mind to structures used for efficient electronic computation. The spatial metaphor is founded on the idea that the human mind is not a computer, and so the most suitable structure for electronic computers is not the most suitable structure for humans.

      A similar approach is taken with Epiphany's bookmark system where users categorise their bookmarks by selecting (possibly multiple) topics to associate a bookmark with. Many people have the same response (it's not a tree! give me my tree!), but have ignored the principle here. We want to make the computer human-oriented, not the human computer-oriented. Consider it for a while.

    2. Re:Newbs by belroth · · Score: 1
      A computer is a tool, and should be approached as such. A user (newbie or otherwise) should not be forced to adapt their mind to structures used for efficient electronic computation. The spatial metaphor is founded on the idea that the human mind is not a computer, and so the most suitable structure for electronic computers is not the most suitable structure for humans.
      Rubbish. Most of the time the file structure is irrelevant to the computer. The tree structure is so that people can find things easily. You can put all your documents in one folder if you like, mp3, documents, spreadsheets, grpahics - your computer couldn't care less but it'd probably be a mess for you to deal with.
      A similar approach is taken with Epiphany's bookmark system where users categorise their bookmarks by selecting (possibly multiple) topics to associate a bookmark with.
      That's a different thing - I readily agree that the tree structure has limitations and a multi-dimensional approach would probably be better.
      --
      I hereby inform you that I have NOT been required to provide any decryption keys.
    3. Re:Newbs by Too+Much+Noise · · Score: 1
      A computer is a tool, and should be approached as such. A user (newbie or otherwise) should not be forced to adapt their mind to structures used for efficient electronic computation.


      wow, nice contradiction here.

      How do you approach a tool you're not familiar with? You learn to use it. Nobody is 'born' with the spatial paradigm deeply rooted in one's brain. Heck, it's not even all that natural to boot with - you don't naturally remember objects by position, but by their relationships with each other. That's neither spatial nor tree-like, although it's used by both. And the tree-like structure is not computer-optimized, but that was already pointed out.

      The better question would probably be why did the browsing paradigm survive for so long while the spatial one gets reinvented every few years, then dropped? I don't have an answer to that, but venturing a guess I'd say users outgrow 'newbie interfaces' - at some point the spatial paradigm simply stops being useful enough. No user stays newbie forever, as others pointed out, organizing your files into moderately deep trees isn't rocket science; on the other hand, sorting through several thousend files in one folder would end up annoying anyone.

      And to close the argument, let's rehash the most 'mom-and-pop' counter-example for spatial pardigm: digital photo. With all the digicams out there, this is bound to be one of the major ways of filling a hard drive. Now, all those cameras give the files some silly counter names that overflow quite rapidly. So users have 2 choices: either rename every picture to something unique in order to dump them all in the same 'spatially-organized' folder, or put them in unique folders for each shooting session. Now the first choice, aside from being, shall I say, tiresome, has the disadvantage that the one 'my pictures' folder will start opening up rather slowly once you have thousands of pictures in it (especially if one uses thumbnail previews) and finding files will become rather tedious[*]. How long do you think a newbie user will stick to this method? and when he/she moves on and organizes stuff, how soon until the spatial paradigm breaks?

      [*] please don't bring up metadata searches in this case. Metadata is not something that gets automagically embedded into an image file, you still have to type it in. I am yet to see someone who would add metadata to several hundreds of pictures - and even if that someone exists, it won't be a 'newbie user'.
    4. Re:Newbs by harves · · Score: 1

      If the tree structure is for human use, why is it such a common structure for computers? They're used in the file system because they're efficient for computers. The suggestion that a human could place all their files into one directory and the computer "couldn't care less" is wrong - there is a bound on the number of files which can be contained in a directory. Why is that?

      Imagine if the file system was unstructured (a flat list), but with the filenames given standardised names like :etc:X11:XF8Config. You would find that, although unique names are provided for each file, the computer quickly bogs down. Every time it needs to access a file it needs to load and search a massive list to find it's location on the disk.

      Anyway, my point was that users have been prevented from constructing wider tree structures which they can use spatial information to navigate. Computer users tend to create logically clean (but deep) tree structures, as a *result* of computer science creating tools to manage trees efficiently. But a user (relying on spatial information) doesn't *need* logically clean tree structures, and (it is argued) is faster with wider trees.

    5. Re:Newbs by harves · · Score: 1

      I'm not sure where it was pointed out that the tree structure is not computer optimised. Can you point me to that?

      Anyway, spatial Nautilus is designed to utilise what spatial memory abilities you have to allow for flatter directory tree structures. This is what was demonstrated in the article. Other Gnome projects are doing the same (notice that the menus in Gnome projects are becoming smaller, with much less menu nesting?). Deep hierarchies are being avoided in all aspects of Gnome, and Nautilus demonstrates that. Many files + shallow tree = wider tree, and to navigate that you will need to use what non-computer-based skills you have. Whether or not spatial memory abilities are built into the brain is not a real concern - you have them, so you might as well use them.

      I won't spend too much time addressing the second half of your response as it deals with arguments I didn't bring up. I will say that I have thousands of photos, on my hard drive, and use spatial Nautilus. An example path is /share/Photos/2004/Mountaintop. I have no problems whatsoever navigating to that location with spatial Nautilus. The directory /share/Photos/2004 contains around 10 sub-directories. /share/Photos/2003 around 20 sub-directories. Spatial Nautilus helps me navigate these kinds of wider trees.

    6. Re:Newbs by belroth · · Score: 1
      If the tree structure is for human use, why is it such a common structure for computers? They're used in the file system because they're efficient for computers.
      They're used in computers because computers (especially OS) are designed by humans, and humands find it easier to work with. There's really no objective reason to have /bin /sbin /opt etc. separate.
      The suggestion that a human could place all their files into one directory and the computer "couldn't care less" is wrong - there is a bound on the number of files which can be contained in a directory. Why is that?
      Because the filesystem ihasn't been updated to cope with a larger number of files. When OSs are modified to cope with 64bit addressing this shouldn't be an issue.
      Imagine if the file system was unstructured (a flat list), but with the filenames given standardised names like :etc:X11:XF8Config. You would find that, although unique names are provided for each file, the computer quickly bogs down. Every time it needs to access a file it needs to load and search a massive list to find it's location on the disk.
      Or it could use a hash to reduce this search through the index of file locations considerably, and I'd be very surprised if there weren't better ways I haven't thought of in the 2 secs I gave it.
      Anyway, my point was that users have been prevented from constructing wider tree structures which they can use spatial information to navigate.
      Prevented how? They may have not been encouraged but that's a different thing. A nice analogy would be that you can write a word processor in Fortran if you want to, but it (Fortran) doesn't make it easy.
      Computer users tend to create logically clean (but deep) tree structures, as a *result* of computer science creating tools to manage trees efficiently. But a user (relying on spatial information) doesn't *need* logically clean tree structures, and (it is argued) is faster with wider trees.
      Maybe both the design of the OS and the way people use them are a reflection of how most people think. Most library classification systems are similar too.
      I'm not saying that we can't do better with an intuitive leap and some clever design, but I'm far from convinced that trees are as evil as some people are suggesting. Databases use trees a lot because they give fast access and there's been a lot of research on this. OS are moving towards a database oriented approach now too.

      With a really good UI design the user should be able to use whatever method works for them with no regard to how the data is stored at the filesystem level, it's called abstraction, and that would be far more useful than forcing people to use either a tree or spatial paradigm.

      --
      I hereby inform you that I have NOT been required to provide any decryption keys.
  7. it's great.... by sohp · · Score: 1

    ...if you don't mind changing the way you have been doing everything effectively for the past decade or two. David Gelernter, in Machine Beauty , derides this situation, reminding us how bad it is to deal with "a complex or weak program that forces you to bend to its worldview instead of accomodating yours."

  8. Dead American programmers tell no tales! by Blackheart2 · · Score: 1

    From the article: First, I wanna tell you what I usually do to make a computer easy and consistent to use for a newbie...

    The Grim Reaper replies: Shut up, you American! You Americans, all you do is talk and talk, and say "let me tell you something," and "I just wanna say..." Well, you're dead now, so shut up!

    Ahhh, I feel better now. :)

    --

    BH
    Fools! They laughed at me at the Sorbonne...!

    1. Re:Dead American programmers tell no tales! by Blackheart2 · · Score: 1

      Oops, I see now Christian is a fellow German; pardon my knee-jerk reaction.

      He still writes like a stereotypical American, though. ;)

      --

      BH
      Fools! They laughed at me at the Sorbonne...!

  9. Just break Gnome by pabtro · · Score: 1
    It was clear, even back in 1995, that this file navigation paradigm didn't work. I was astonished to see it implemented in Gnome 2.6. For me Gnome has been rendered unusable. The author of the article lengthly explains how he has to re-organize his world just to accommodate the new changes. This is disgraceful; even a child is able to navigate files a la explorer way. Nice and convenient way to break a very viable platform for Linux desktop applications. The issue for me is that I won't use Trolltech's billboard (KDE).

    There are not many alternatives left.

    1. Re:Just break Gnome by Puggs · · Score: 2, Informative

      Not many alternatives left?????

      How many do you need? - Fluxbox (or blackbox or *box), Windowmaker, IceWM, Enlightenment, XPde, None, ION, Ratpoison, or the several more that I cant be bothered to find for you...

      Alternativley, you could use google to find out how to turn the spatial interface off & use the old nautilius

      Or perhaps you'd be happer with the huge range of choices on Windows - explorer.exe, explorer.exe or possibly explorer.exe ;)

  10. Tree stucture are natural and easy to comprehend by Gnulix · · Score: 2, Insightful

    If there really is a need for all these articles that explain what spatial file browsing is, how it works, and how it should be used -- then there is something seriously wrong with spatial browsing.

    The ordinary tree structure has worked for millions of users, most were of them newbies when they got introduces to file browsing. I wager that there are no one (or very close to no one) who'll complain about the idea of a tree structure.

    If people can use an ordinary index in a book, they can understand a tree structure. Most people will think in a manner that is similar to a tree structure; Let's see, pictures, vacation, 2003..

    People like having multiple views of the same information. Why are virtual folders in e-mail applications so popular? Because they provide simple, logical views of the same date, used in different context!

  11. Re:Why I dislike spatial by Too+Much+Noise · · Score: 2, Informative

    I dislike spatial for the same reason I hate websites that pop up new windows for every link.

    Use Firefox with Tabbrowser Extensions, and you can arrange it so you'll never have a link open in a new window again. Neat huh?


    Neat. Also, completely, utterly, un-spatial. But a beautiful clue for all the spatial zealots saying their way is the one true way. Thank you.
  12. Re:The better question.... by Creepy+Crawler · · Score: 1

    Hey, I was trying to karma-whore.

    Gimmee a break, Ive 3 trolls to feed at home, and a sick wife.... ;-P

    --
  13. Spatial Paradigm is completely retarded by p3d0 · · Score: 3, Funny

    Just my personal opinion, but the whole reason I use a computer instead of a hundred Post-It notes lying on and around my desk is because the computer supposedly organizes things better. Now they want to have the computer deliberately and faithfully mimic desk clutter? That's so damn stupid, I want to scream.

    --
    Patrick Doyle
    I mod down every jackass who puts his moderation policy in his sig. Oh, wait a sec....
  14. NEWSFLASH: he actually likes spatial browsing! by khanyisa · · Score: 1

    It's regarded as NEWS that somebody actually likes spacial browing and can find a few reasons to justify it! This article proves that spatial browing has not convinced the majority of people.
    What we really need is better file metadata and indexing tools...
    I have 16000 files in my documents tree... how exactly am I supposed to arrange those in a two-level directory hierarchy?

  15. Re:Tree stucture are natural and easy to comprehen by wjwlsn · · Score: 1

    I am more comfortable with a tree-oriented system as the default, but there are times I want a new window. Both metaphors are natural and easy to comprehend. They both have their uses. I find navigation to be much easier in a tree-structure, but drag-and-drop seems to work much better in a spatial structure. Why not make it easy for me to do both? Another thread here mentions old MacOS 9 behavior of allowing either way... why shouldn't we have something like that? If I single-click, keep me in the same window. If I double-click, open up a new window. Best of both worlds, easy to use, easy to understand.

    --
    Getting tired of Slashdot... moving to Usenet comp.misc for a while.