142

In our digital world we have previously been forced to know the location of digital content. The obvious one being the local hard drive or a shared file server where sometimes extreme structures have emerged over time. Sometimes in a controlled manner, and sometimes not so controlled. We find the same pattern in navigation of a web site where Information Architects (like myself) spend hours and hours finding the correct ontology and structure of the site for our customer. All this effort is made to make a general, easy to understand navigation of the site with the obvious goal to let users find the content they need effectively.

Since 2007 with the increasing use of Smartphones, our users have learned other ways to find content – through the use of apps. There is an app for images (often not in folders) where users find what they need by metadata navigation and not digital location navigation. Users may find images based on when the photo was taken or where (physical location displayed on a map). They don’t know where (on the smartphone) the exact location, and can’t use the folder structure of the smartphone to find what they need. The only access is through apps who can display image files. This is true for the majority of users.

As our preferred search engines get better and better, users find content through search queries rather than navigation in a folder structure. This is valid not only on public and internal web sites, but also internally on a smartphone or desk top device where users find apps and content through searching (i.e. typing the name and given suggestions instantly).

At the same time we see a lot of users navigating through large structures on file shares, on web sites and in applications (ERP’s and other apps) to find what they need just because they’re used to this analogy. So it’s probably not a valid call to rip the folder structure thinking just yet, but will it ever disappear? Or do we want user to categorize content based on folders?

It all boils down to the question: Is the folder structure dying?

Benny Skogberg
  • 54,996
  • 22
  • 140
  • 241
  • 2
    I consider this question closely related (in a complementing way, not an overlapping way) to Can tags have hierarchy? – O. R. Mapper Jun 30 '15 at 18:03
  • 2
    @O.R.Mapper or this one? :-) http://ux.stackexchange.com/questions/20217/has-there-been-an-implementation-of-a-tag-hierarchy-with-graph-edge-relations – Benny Skogberg Jun 30 '15 at 18:21
  • 5
    Since 2007? Really? Use of search/app-specific locations instead of a full user-visible folder structure has been growing since long before that. – Nathan Tuggy Jun 30 '15 at 18:42
  • Folders are still needed for backend things (that picture app might just read the /DCIM folder, for example). Users just don't see that. – Jon Jun 30 '15 at 19:30
  • 1
    @Chipperyman: Not necessarily. The PalmOS did not have folder structure. Nor did it have a filesystem (though it did eventually support reading filesystems from SD cards). Instead it used a built-in database to store everything. Even pictures. – slebetman Jul 01 '15 at 04:04
  • 4
    I'd like to point out that Palm wasn't successful in their quest of getting rid of files and folders (they preferred relational databases). By around 2001 or 2002 I remember they had to implement a fake file/folder system on top of their database because users were confused by not having a familiar user-level access to their mp3s and jpegs (databases of course are fully accessible to apps but is confusing to regular users). The irony is that iOS today actually implement a full filesystem underneath but hides it from the user and present a databaase-like UI to access data via apps. – slebetman Jul 01 '15 at 04:10
  • 1
    I think it might be simply because most users don't have enough data to warrant organizing it in folders. Apps that do encourage users to create lots of data either organize it chronologically (social media, emails) or by folders (Google Docs/Drive). – user253751 Jul 01 '15 at 08:06
  • 2
    Files were once the most common user interface to data. But since every new/big web company is now focusing on platform development, having a common interface to the data available is rendering their platform obsolete. So for them it's better to have their data locked in and only be compatible with those other applications that they deem useful/profitable for them. – SpaceTrucker Jul 01 '15 at 08:36
  • 1
    David Weinberger's excellent book "Everything Is Miscellaneous" addresses this head-on. It's worth a read if you organize data at all. For more details: http://www.everythingismiscellaneous.com/ – Alex Feinman Jul 01 '15 at 16:27
  • 50
    You can have my directories when you pry them from my cold, dead hands! – Josh Jul 02 '15 at 01:07
  • 6
    I sure hope not! My main complaint when using GMail is that it doesn't have folders. You can kludge around it using "Labels" but it's not the same. Even if the folder (directory) hierarchy is just another view, it is a view that I find essential. – TecBrat Jul 02 '15 at 02:55
  • 2
    The users may stop using this structure, but the devices will still need it to organise their data internally. Put simply, filesystems are ridiculously slow when everything's just in one big folder. – micheal65536 Jul 02 '15 at 15:28
  • @MichealJohnson Folders are not the only way to organise data internally. Think about persistent object graphs, like in Smalltalk. – user253751 Jul 03 '15 at 12:10
  • 1
    I really hope not. I like my folders :) – Navin Jul 05 '15 at 11:51
  • 2
    Smartphones massively predate 2007. 2007 is just when the first iPhone came out. Would it blow your mind if I told you that Apple very rarely actually innovates? You just think they do because their marketing is so well-funded. Your Apple fanboyism betrays you. – Lightness Races in Orbit Jul 05 '15 at 18:36
  • 1
    I'm surprised nobody's mentioned things like Emacs' M-x yet. A hierarchy and a flat structure can coexist nicely -- a flat structure is easier when you know what you're looking for while a menu structure is easier when you don't. The two don't have to be at odds. – Sean Allred Jul 06 '15 at 14:35
  • 1
    @LightnessRacesinOrbit I'm well aware of that Apple is best at bringing new technology to the market and not innovate. But the point here is to give example of when it started to get noticed by the general public (which is what Apple did with its iPhone). That's why I mention the year 2007 since it's relevant for the question with the new (to a lot of users) application centric storage. I'm sorry to disappoint you, since I'm a Microsoft guy by profession and not an Apple fan buy. ;-) – Benny Skogberg Jul 06 '15 at 15:55
  • 2
    @Benny: Then don't write "Since 2007 (the first Smartphone)"... – Lightness Races in Orbit Jul 06 '15 at 16:29

17 Answers17

125

Rumors of its death are greatly exaggerated...

Classic hierarchical folder views aren't dying. But they are being complemented by other ways of viewing and interacting with files.

The key trend here is the decoupling of views from the underlying file system.

The old world...One truth ⇨ One view

  • Historically, file UX was heavily tied to the underlying representation of files and folders. Operating systems stored files hierarchically in a single physical location, and the easiest way to design a safe UX was to represent that hierarchical file system:

image of old world: classic hierarchical file system. "One truth, one view"

  • Note: by "underlying representation" I don't mean the organization of data on a hard drive platter, SSD or other mass storage. Instead I'm referring to the intentionally abstracted model of files, paths and directories which OS's use to represent files. This isn't just a UX matter, as path-file model is a fundamental model for most OS's, so programming languages use that same model to provide their most commonly used interactions with the file system.

The new world...One truth ⇨ Many views

  • The rise of the cloud, multimedia, terabyte storage, mobile/tablet, and multi-mode computing has changed the way files systems are used and implemented.

  • The modern approach is to decouple views from the file system. This allows users to interact with file systems in many different ways (search, tags, folders, etc) without compromising the underlying integrity of the data too much.

image of one truth, many views

  • Data is sometimes stored in one location, sometimes synced, sometimes encrypted or distributed... the decoupling of views allows designers to present a unified and optimal interface for users without having to worry too much (or at all) about the underlying file representations or locations.

Folders aren't dead in the new world...

  • A hierarchical view for files/documents is still fundamentally very useful for all sorts of tasks. The decoupling of views and the availability of other interfaces doesn't change the fundamental usefulness of a hierarchical representation for many tasks.

  • Therefore, the folder system continues to live even in modern systems like Dropbox and Google Drive.

  • What has changed is, folders are sometimes just another view into the file system. Cloud files may or may not be organized into folders in actuality... it doesn't really matter.

  • This is great news for designers because it means we can use the best view for the job... and without the hard constraint of coupling views into folders, we can use hierarchical folders when they make sense for the user experience, not just because we are forced to.

Laurel
  • 635
  • 5
  • 17
tohster
  • 41,070
  • 14
  • 107
  • 139
  • 11
    +1 for a "Hell No" answer where everyone else agree. I disagree with you, but appreciate the way you argue that folders really are views in itself as if the corps still lives :-) – Benny Skogberg Jun 30 '15 at 18:14
  • 18
    Benny =) in a world of notebooks and tablets, I still see a lot of pens and notepads brought into meetings at work....suitability to task here really dictates the user choice so the future of folders depends largely on that suitability. – tohster Jun 30 '15 at 19:01
  • 6
    I think it is ironic, that the underlying hardware (bits on the harddrive) never stored the "data" hierarchical, but always as a bunch of flat File-fragments on a big plane, with a Meta-Data container, which can store links for the files. Each file can have multiple links pointing to it. So the idea of the data being in truth in folders will come back and bite you, if you edit a file from one hardlink, or try to count all files by counting the files in all folders ;-) – Falco Jul 01 '15 at 10:11
  • 5
    @Falco I fear you are extending to fallacy there. A file system has always been a software construct - the physical layer has always varied. SD cards, HDD, SDD, etc. However, the mapping of file to filesystem is always 1:1 – Gusdor Jul 02 '15 at 10:59
  • 1
    @Gusdor But in all major File-System, a file can at the same time be in more than one folder. The file exists once, but with two hardlinks - and none of the two hardlinks is better than the other one, So the file is not in one directory! If People think "The filesystem is 1:1" they can run into problems comprehending, their same file is in two places at once (because actually it isn't, it is in one place on the physical data-storage, accessible by file-system semantics) – Falco Jul 02 '15 at 11:34
  • @Falco: Given that FAT32 is used by nearly all digital cameras, I'd say it deserves to be called a "major file system" and it has no concept of hard links. I'm not sure if NTFS has hard links which are indistinguishable from an original file reference. Were I designing a file system today, I don't think I'd include Unix-style hard links that were indistinguishable from normal files, though I might include a form of "link to file that has no "normal" rooted reference", and provide a means of "moving" such a file to a normal location. – supercat Jul 02 '15 at 15:21
  • @supercat NTFS does have hard links, as does ext4 and newer systems like BTRFS and ZFS. It's a pretty standard feature on modern file systems. – Ajedi32 Jul 02 '15 at 16:51
  • @Ajedi32 hard links to files is a required feature for Posix conformance. Thus, NTFS had it even though it was not "natural", so it could support their planned Unix subsystem. Any Unix file system will have it. – JDługosz Jul 02 '15 at 17:18
  • @JDługosz: What exactly does POSIX require with regard to hard links? If the system implemented links which were different from file entries, but from the point of view of POSIX operations looked the same (and automatically updated when files moved, and guarded files that would otherwise be deleted), would that be legitimate, or does POSIX require that they be indistinguishable in every way? – supercat Jul 02 '15 at 17:51
  • @supercat I don't know. I suppose it's relative to other posix-described functionality; so, could be distinguished using non-posix functions. – JDługosz Jul 02 '15 at 17:59
  • Yep! I use NTFS hardlinks somewhat often for exactly that purpose, allowing multiple "views" in my mp3 folder, for mp3s that really need to be in two or more locations based on the (rather idiosyncratic) way I sort my music. It's pretty lame that Windows doesn't, by default, have any way of creating them (though it handles them just fine once you use a third-party tool to create them.) – neminem Jul 02 '15 at 23:02
  • @neminem Third party tool? What's wrong with mklink /H (or /J for directories)? – bcrist Jul 04 '15 at 15:59
  • @bcrist Huh, cool. What's wrong with it is a. I didn't know that was a thing until you told me just now, but b. also I'm not usually organizing my mp3s in the command line anyway, so while I have nothing against the CLI for certain types of operations, in this case a drag-drop explorer extension is definitely the way to go. But yes, you are correct, I was wrong that it is required rather than merely more convenient. Nifty. :) – neminem Jul 04 '15 at 21:44
  • 1
    Does anyone else think that second infographic isn't really saying much? – user253751 Jul 06 '15 at 02:49
  • 1
    @gusdor is correct. The technical aspect of this answer is completely incorrect. There never has been any correlation between the UI (folder structure) and the actual files on the storage device. The folder structure was merely one representation of it (historically, the default one). – DA01 Jul 06 '15 at 04:40
  • @DA01 the directory file system predates windows itself and the windowed concept of folders. This answer just states that the hierarchical directory was the operating system's underlying representation, and that the data tended to be stored in one location (rather than in the cloud for example). How magnetic dipoles were serially arranged and blocked on platters is irrelevant to this question (but apparently relevant to tireless pedantry). – tohster Jul 06 '15 at 05:02
  • @tohster what is SE if not pedantic? :) That said, your initial bolded statement is simply incorrect "decoupling of views from the underlying file system". Your latter comment is correct in that it was the default representation. I'd suggest updating the answer to reflect that the folder structure was the default representation of the file system. – DA01 Jul 06 '15 at 05:06
  • That said, I guess I'm being pendantic. The rest of the answer is still confusing things...namely that the 'decoupling from the file system' is what's new. That was how it always was. Even with the original folder structure, that was decoupled from the actual files on the storage device. I think you want to say what is new is that "how to find files is now decoupled from the original folder-only view" (though that's been true for decades as file searching and tagging and such has been around for quite a while.) – DA01 Jul 06 '15 at 05:07
  • @tohster The directory file system was a UX construct. The directory tree does not describe "where the data is stored", it is a bunch of references, even in FAT32. In any modern (as in, last 40-odd years) file system other than FAT, the folder/directory is the card catalog: the file system is the library. (In FAT, each "card" has a string of yarn attaching it to the "book" it refers to (And yes, this makes a mess of a library.) In other file systems, it just has a note saying "see index 732, shelf 7, stack 32, room 42, building 7, planet 3".) Folders where just a UX wrap of a UX wrap. – Yakk Jul 06 '15 at 14:30
  • @yakk no it's not just a UX construct. An of course files are stored in bytes, chunks and blocks all over a platter or in capacitive or transitive elements in a solid state drive. And operating systems ultimately need to read and write at that low level. But the working representation of files in operating systems happens at one level of abstraction higher, where commands like dir delete mkdir and so forth occur. This isn't just a user interface construct....it is the way that files and directories are data modeled by the OS. Eg API calls and not just the UX use this representation. – tohster Jul 06 '15 at 15:02
  • 1
    dir is a program intended to interpret the state of the filesystem for users. It is a user interface program, not a GUI one, but it is UX. Same for mkdir and delete. The command line is a user interface. If you go back to pre-win1.0 days, programs would interact with the file system far more directly than those programs. You might have a library to help, but that was not essential. In windows, low level access APIs are sitting there right beside the "path" style access, up to security restrictions. And far more systems use posix APIs than win32 APIs. UX doesn't mean GUI. – Yakk Jul 06 '15 at 15:14
  • @yakk using that logic, any data model or API can be termed as UX. It isn't just a matter of UX. Programmers use path, directory and file constructs in languages from Javascript to C to access the file system....that is not because of "UX". It's because the OS is intentionally designed to provide that abstracted representation of the file system. – tohster Jul 06 '15 at 15:19
51

It's not dying completely, but it is becoming a power user niche feature.

Everyone has seen or heard stories about the user who stores everything on their desktop or in a single My Documents folder. Humans are terrible at justifying a large upfront cost like creating and managing dozens of folders just for a possible, small benefit in the future like being able to find a file 10 seconds faster or with less scrolling.

Managing your file system is something only power users need to do or care about. Things like tagging, or Spotlight and instant full-system searching mean a typical user doesn't have to care about where things are as long as they can find them quickly.

This article by Toby Coulstock summarizes the concept pretty well: http://members.iinet.net.au/~chris6/100214_the_file_system_is_the_new_command_line.html

Over time the concept of an "operating system" has changed from being only programs that give you access to your hardware, and has become "meta software" tasked with getting you access to the software you want when you need it. It's one more layer of abstraction on top of decades of computer history, and it might be the one that makes file systems obsolete and something only developers have to worry about.

Nathan Rabe
  • 8,888
  • 2
  • 28
  • 38
  • Great angle of the question which I didn't think about. Thanks (and +1) – Benny Skogberg Jun 30 '15 at 15:04
  • 1
    "as long as they can find them quickly" - well, can they reliably find them? – O. R. Mapper Jun 30 '15 at 17:59
  • 5
    You've got it right: the answer depends on the audience. The "information appliance" sector of the market doesn't really use directories now (and sure doesn't know enough not to refer to them by the F-word :-)). People who use computers for serious computing purposes will keep on using them until & unless something better comes along. – jamesqf Jun 30 '15 at 20:07
  • The one thing I would mention is that sometimes you may wish to have two different files with the same name, placing them in different spots to distinguish between them. This could extend to groups of files or even folders. Even a regular, non-technical user may want to do this if, for instance, they're writing a legal term paper or something in law school and want to make full, multi-file backups. – Panzercrisis Jul 01 '15 at 19:08
  • 3
    To expand on my comment, "just for a possible, small benefit in the future like being able to find a file 10 seconds faster or with less scrolling" is totally what it seems like. Of course, the reality is that in many cases, the searched file isn't found 10 seconds later, but that it is not found at all. That doesn't change anything about the conclusion that only "power users" actually do sort files into a directory structure, but I think it is crucial for the reasoning to acknowledge those "power users" aren't the only ones who would actually benefit from a little tidying up. – O. R. Mapper Jul 01 '15 at 19:51
  • 3
    I have to find literally hundreds of files every day. At risk of butchering a cliché, I'd say that "10 seconds here, 10 seconds there, pretty soon it adds up to real time." – Comintern Jul 02 '15 at 01:47
  • Folders are great for productivity software. But productivity software is no longer a killer app. – user253751 Jul 03 '15 at 12:10
15

Categorization of content is still very important, even after the rise of smartphones. Whether that's by date, by location, or by tags. The more and more popular use of smartphones since 2007 has forced designers to come up with simpler user experiences.

Manually putting stuff in folders can usually be considered a bad UX experience, which is why in a lot of apps, content is categorized automatically. For example, in the Photos app, photos are categorized automatically by location or date, i.e. things that the user would've categorized photos anyway but with location sensors and such these can be done automatically. So my point is, folders and categories are still important, it's just the manual categorization and sorting of stuff in folders by users can sometimes be painful and time-consuming.

Mayo
  • 6,641
  • 9
  • 30
  • 37
Savv
  • 899
  • 6
  • 7
  • 6
    Yes and No. Category and Folder isn’t the same thing. I can add multiple categories to one item, but I can only store my item in one folder unless I make a duplicate. And if I duplicate, I don’t know which version is the current one. – Benny Skogberg Jun 30 '15 at 11:55
  • 3
    @BennySkogberg There are no problems with storing one file in more than one folder. It's not commonly used and visible, for whatever reason, but the option is there. – Odalrick Jun 30 '15 at 12:46
  • @Odalrick So the average Windows user can store the same word document in two different folders without duplicating the file? Very interesting - please enlight me with an example. – Benny Skogberg Jun 30 '15 at 13:11
  • 6
    @BennySkogberg Soft and Hard links provide that functionality. Soft links make a file in another folder appear as a shortcut or alias. Hard Links/Junction Points actually make a "file" appear in multiple folders, but they are all the same file on the hard drive. – JPMC Jun 30 '15 at 13:24
  • @JPMC True that, but as said earlier, not used by many. One of my customers saved the same file to four different locations just to make sure he didn't lost the file. No links, the actual file... – Benny Skogberg Jun 30 '15 at 13:27
  • 7
    @BennySkogberg "Average Windows user"; I don't know. mklink /h new original ? The option is there; but like I said it's not very visible. The special user folders, "My Documents" for instance are also links; the same file in two or more places. I could speculate on why it isn't more visible, usable; but wont, that's another question. Just thought of something: I thought the question was about hierachical structures for sorting chunks of data, a "folder structure"; maybe it's more about how much the file system intrudes in userspace? – Odalrick Jun 30 '15 at 13:33
  • @Odalrick Another bit of dark magic to an "average" user: Something I've encountered with some linux editors (and I presume the same is true for some on Windows), they'll work in a temporary file and replace the original on save, breaking the hard link – Izkata Jun 30 '15 at 18:04
  • 1
    Hierarchical file systems were designed for files to be in one location; while you can sometimes put a file in more than one location, it doesn't work very well because most software is not designed for that possibility. (For example, compress the outer folder with 7-Zip and it compresses the file twice) – user253751 Jul 01 '15 at 08:04
  • @immibis: I wonder why utilities like that don't include an option to maintain a list of the hashes of the files it has processed, and--upon finding that two files have the same hash--check if they're bit-wise identical and avoid the duplication if so? – supercat Jul 04 '15 at 21:05
  • @supercat: Think about it...they don't because it could be extremely slow when dealing with large files and/or slow file systems. – martineau Jul 05 '15 at 00:24
  • @martineau: One wouldn't want to use such a feature all the time (which is why I suggested it be an option), but if one is going to be archiving a bunch of folders that have large amounts of duplicated content, there are many cases--especially when the destination device is slow--where identifying duplicates and sending "duplicate of X" references rather than links could save a huge amount of time. – supercat Jul 06 '15 at 15:17
  • Categorization is indeed very important as you say images are tagged by date, location, face detection. Tag search is cross folder but the item in my opinion can only live in one folder, yes it can have a date tag and a location tag, so with tag search it will show up twice but the original only rests in one place. On G. Drive, Onedrive, Dropbox the likes this is the logic, one item can only be in one folder, on the other hand on smartphones galleries you can save an image/video in two different folders. So you end up creating two copies. Why is that? Thanks – Giulio Sep 30 '21 at 07:14
15

Most people tend to think that certain aspects of technology die away, but I always get the feeling that it is far more common for it to "sediment".

Whenever I see a new trend, it usually never ends like "this is the new best way". Normally it is more like "we solved this problem with one global solution, but now it seems there are different solutions for different cases".

The folder structure is a good example. We have been using folders for a long time for everything. Now we are finding out that certain assets like pictures or music are better dealt with a specialized app.

You mention better search engines, this is great to find exactly what you want, but not to explore/visualize what you have. From a UX perspective think about how music apps let you explore by artists or albums. Also a photo collection tends to be visualized by events, locations or people... Users still want a meaningful way to visualize their music/photo collection with specialized apps.

However not all media types fit nicely into a unique app. A folder layout can convey meaning in a shared environment and combine a wild array of assets for which there is no specialized app.

As software evolves we will see more cases where a specialized app can manage a set of files, but I doubt we will reach a point where our folder layouts will disappear. It is more likely that we will stop overusing folders and just stick only with the cases where they are useful.

SystematicFrank
  • 301
  • 1
  • 5
  • 2
    +1 for truly amazing reasoning. Love the way you lay the words and I'm bound to give credit for the red line argument. Thanks! – Benny Skogberg Jun 30 '15 at 20:26
  • +1, though I still prefer to listen to music by double-clicking files out of A Big Folder of MP3s, because of the other baggage that comes along with app-like music managers, so I think you can add user preference and conscious tradeoffs as another layer at work here. Especially so along user persona lines (as a dev and a Linux user the command line is essential to me, that hasn't "died" yet.) "Sediment" is a great way of putting it; I'm definitely using that. – user1454265 Jul 08 '15 at 13:56
11

Probably. But it's a slow painful death.

It essentially boils down to the need to put something somewhere where we can find it again, or where we can direct someone else to finding it.

We are naturally predisposed to putting things in containers or compartmentalizing in such a way that even if it's a long time before we come back then we can still have a mechanism that mentally helps us find things. A place for everything and everything in its place. Keeping things tidy.

Due to the large number of objects we might have, some hierarchical mechanism has always helped in this process because the brain is great at chunking things to make them more manageable.

However, this way of thinking has a problem - a thing is stored according to one chosen attribute or a specific ordered hierarchy of chosen attributes. Say I have a Word Document, so I store it in a Word folder inside a Documents folder. But what if I have a red square - do I store it in a Red/Square folder or a Square/Red folder - what's more important? And will the same attributes be equally important when I come back?

In the digital world, it can be made easy to find something using one or more attributes in any order, so that these adjectives have no priority. I may not even need all the attributes if there only happens to be one red thing. I may not even use the same attributes that I thought of at the time the thing was stored. My red square was maybe a vector drawing not a raster image, so just find me vector drawings I made last year.

So yes, increasingly it makes less and less sense to organize in a way that suits us as individuals, and we just let software do the work for us, but we have to trust that software to store it, secure it, index it, retrieve it and return it.

Unfortunately I don't (can't) yet trust all software on all devices to do that securely and all via a common interface yet, so I still like to keep control and put stuff where I know it's safe and I can get it back using my own rules thanks very much.

Benny Skogberg
  • 54,996
  • 22
  • 140
  • 241
Roger Attrill
  • 71,049
  • 15
  • 151
  • 246
8

Search helps if you know what you are looking for (obvious). If you don't know what you are looking for, a folder structure can help you find it.

For example, if you are looking for a recipe for chicken Florentine, search will help you find it, or if you have chicken and want a recipe that uses it, search may help there too. But if you just want to cook something, folders can help you find the recipe you didn't know you were looking for. If the folder structure is [main ingredient]/[flavor profile]/[cooking style]/[recipe], you could open the chicken folder and see the options of simple, sweet, savory, and spicy, then open one of those and see folders like grill, roast, fry, sauté...

Dan Henderson
  • 157
  • 7
Fred
  • 81
  • 2
  • 1
    Welcome to UX.stackexchange. If you can please elaborate on your answer. Why is a folder structure a valid way of finding something if you don't know how it's categorized? Finding NYC is easy if you know it's inside NY state and inside USA and inside North America. But if you're looking for Dakar would everyone know that it's inside Senegal and inside Africa? – Mayo Jun 30 '15 at 15:32
  • 1
    I don't understand how a folder structure is any more/less help than search if you don't know what you are looking for--unless the implication is that folders can act as categories in certain situations, which can aid in browsing? – DA01 Jun 30 '15 at 17:18
  • Say you are looking for a town outside of Boston, but you forget the name. You could browse a folder structure containing states, then counties, then from a list of cities and towns, you recognize the town you were looking for. It would be difficult to find this item if you didn't have some hierarchy to browse. – Fred Jun 30 '15 at 20:33
  • 2
    Fred, would be nice if you could add this analogy to your post. Another analogy is a library with many shelves. You are bound to learn something new and unexpected by simply browsing along the shelf, while search reinforces your existing biases and stereotypes. – Deer Hunter Jul 01 '15 at 00:35
  • @DeerHunter - exactly. I think eBay pulls this off fairly well. You can search for something, click on an item, then from the navigation breadcrumbs, discover other items in this or parent categories. – Fred Jul 01 '15 at 12:22
  • @Fred that's true, though is entirely dependent on there being a hierarchical folder structure that a) is organized and b) organized for the specific content a person is seeking. As such, I don't think one can universally say 'folders make it easier to find something you don't know what you are looking for'. A proper taxonomy and navigation system can certainly assist in browsing, but rarely have I ever found folder hierarchies in an OS set up by the user that well. (In otherwords, OS Folder structures can act like an organized tree view of content, but rarely are they organized) – DA01 Jul 06 '15 at 04:43
5

As tools (software and hardware) increase in speed, the value of search begins to eclipse the value of a folder structure as a way to find a file.

On my PC, my MacBook, and my phone, it's simply faster to search for items by terminology rather than seek it out visually.

Most file systems still need a folder structure, so it's not dying, it's just that finding elements within is now augmented and in many cases superseded by search.

DA01
  • 41,799
  • 5
  • 81
  • 142
  • 1
    I have removed the comments here. This is not the place for extended discussions. Please use the [chat] if you wish to continue. – JonW Jul 06 '15 at 19:14
5

Folder structure is not dying

However, its importance and prevalence in the average user's interface is.

The truth is that a user will find the path of least resistance to accomplish their desired task. With this in mind, maintaining folder structure would have to become their desired task. For the average Joe, nope!

The average person has become accustomed to instant gratification, primarily carved by Google themselves.

So if I am Joe and I am getting photos from BookFace and my picture viewer/downloader app conveniently adds my friends' names, photo year, location, etc... as meta data then I have severely minuscule reason to monkey around with the folder structure so that I can easily find pics with Jerry from 2009.

For this we have modern databases to be thankful for.

Mayo
  • 6,641
  • 9
  • 30
  • 37
MonkeyZeus
  • 5,658
  • 17
  • 31
5

Tree view is often complicated and counter-intuitive to the users, but it has several strong advantages:

  • full names which are unique identifiers

I mean, if I access a file named /etc/passwd I know I access a particular file. There is no way I get /home/backup/passwd instead. Tags don't give you this garantee: if you have found exactly one object tagged password, you still cannot be sure it's the one you want.

  • recursive structure which enables encapsulation

Suppose I store my e-mails attachments classified by year and month, and my photos by year and location. In a tree structure, these different classifications can easily coexist withing the same tree without any interference at all, e.g. /attachments/2010/January/bar.jpg and /photo/2010/London/baz.jpg. In a tagged system, January and London will end up in the same category, which may or may not be what you want.

So the tree structure is going away for good where its complexity is not needed, but in cases where its properties are desirable, it won't go anywhere.

Dmitry Grigoryev
  • 2,256
  • 12
  • 19
  • I agree with the rest of your answer (+1 for that), but could you briefly elaborate on "Tree view is often complicated and counter-intuitive to the users", please? It always seemed to me like tree views are among the least problematic controls to users, especially when the tree views have little + and - signs for expanding and retracting, even when users did not cope well with the theoretical concept of a hierarchical folder structure. – O. R. Mapper Jul 02 '15 at 09:57
  • The question itself speaks about how hard it is to create a tree where users easily find what they need. Consider this: when you plug an SD card in a smartphone, you get new apps in the menu, not a nested menu with apps from the SD card. – Dmitry Grigoryev Jul 02 '15 at 12:46
  • Well, I'd argue that behaviour is terribly annoying, because it's harder to find the new apps that way rather than in a nested menu. But that notwithstanding, I see your point here - "a tree where users easily find ..." refers to the tree structure, the choice of what directories exist and which directories are nested into which other directories. Your answer, on the other hand, referred to "tree view", which I associated specifically with the UI control displaying an extensible/rectactable tree as a list of indented nodes connected with lines to indicate their nesting relationships. – O. R. Mapper Jul 02 '15 at 12:48
  • 1
    BTW, it's funny to see how history repeats itself: Program Manager was replaced by a Start menu, which is being replaced by tiles, similar to smarphone apps. I guess the next step is to group tiles to folders, coming back to where it all started. – Dmitry Grigoryev Jul 02 '15 at 13:00
  • Both the program manager and the start menu are trees, just shown in a different way. The Android home screen works a lot like the program manager, as well (regrettably, with nesting restricted to only one level so far). I am not sure how tiles work, as I have not tried any system that uses them yet, though also some users who did not bother to create directories on e.g. the Windows desktop for their icons would sometimes geometrically group them to form visually separate clusters depending on what the applications do. – O. R. Mapper Jul 02 '15 at 13:04
4

Maybe more true thing to say would be that concept of folder structure as file organizing strategy is dying. Or maybe more true: organizing files/documents is becoming more and more automated.

If we have concept of categories and one file can be in two categories then I would really dislike doing manual organization. Therefore all these new technologies use meta-data (date/time, geotags, tags, descriptions) to automate sorting of files into.

One should take a look at example of applications on desktop computers. We have package managers that take care of which applications are installed, which libraries and versions of libraries are installed (dll hell, etc...). And operating system does this for us (developers). Yes, it is more primitive then some metrics used for files/photos (e.g. we use only PATH variable to look for executables) but it's one side of a same coin. And application developer doesn't care where library, or application is installed (if it's available in PATH) because underlying tools automate process for us.

Next stop: documents and data files. Folder structure as it exists will not die, but new layer of abstraction is in development.

edin-m
  • 171
  • 2
  • +1 for thinking of automated categorization. There have been a few more or less successful trials on the topic, and the tools evolve. I think you're on to something big here. Great! – Benny Skogberg Jun 30 '15 at 18:41
2

They aren't dying, just finally revealing their limitations.

In the age of limited processing power and os capabilities, folders were the easiest way to structure data. But not the best, as there's just one dimension to it - that is space.

Tagging and taxonomies (word searching being just another one of those) came and added any number of dimensions, to the point that simply-space may become almost irrelevant inside the same system. That's why folders lost their undeserved superpowers and are now only being used when they're truly needed.

That being said, there will probably always be some kind of folders - your phone or your computer are a sort of folders themselves, even clouds are (synchronized) folders etc. I'd say that as long as there is some kind of space delimitation between data, there will be "folders".

lucian
  • 212
  • 1
  • 5
2

In the past many users were often confronted with an implementation model. The digital structure of operating systems - based on files and folders - was directly passed through to the UI. People were and still are willing to „container“ the information they find or create in to this structure. But only to a certain extent. Since we are human beings whose meaty brains have a mesh network structure we don’t categorize everything to its full extent. We also associate items with many different things meaning that information fragments can fit in to more than only one folder which is why tagging has become quite popular. We start enhancing the documents we want to file away with meta information and we begin to like searching for that information rather than scanning through deep folder structures. On the other hand we need some sort of predefined structure we can come back to. That way we don’t need to recall the method we used to find what we were looking for the first time.

To answer your question: I don’t think folders will die out completely but I do think we are in some sort of transition at the moment. We appreciate the power of search engines and predictive technology either offering us things we presumably like or telling us where things go. We also appreciate some sort of predefined structure where we feel at home and exactly know, where we find things and also know where they belong.

Mayo
  • 6,641
  • 9
  • 30
  • 37
uxfelix
  • 1,880
  • 11
  • 19
1

Search and grouping by automatically extracted aspects are powerful tools but they don't solve the same problems as a folder structure may do:

  • Search is good for finding a file when you know what you are looking for; and you can find similar candidates in the results, which is sometimes helpful, too.
  • Automatic grouping (by location, by time, etc.) is helpful if you are interested in those aspects, or it can be helpful if you struggle to remember what exactly you are looking for - people remember locations and times much better than many other features: "I know it must have been last summer somewhere around Helsinki"

But none of them provides a way to freely group together related content across types and apps.

As an example you may want to group together some candidate images for a flyer, together with the Scribus file and a PDF preview of the last draft version.

Or you group together all documents, notes, and conversations that belong to the same project.

Given that the folders are structured deliberately they can provide speed to their users when they need to jump back and forth between related files.

A good implementation of tagging can provide the same though and even outperform a classic folder hierarchy, e.g. by allowing multiple tags per file.

chiccodoro
  • 171
  • 5
1

It's hard to say if the folder structure will ever truly die, but we can safely say it's been evolving for a number of years.

As far back as 2007 in Windows Vista, the user saw their folder structure separated by carets:

      Windows Vista 2007 - Windows folder location in caret format

Clicking to the right of this area revealed the actual folder structure in its proper back-slash format:

      Windows Vista 2007 - Windows folder location in backslash format

The caret approach is easier to read and more descriptive. Also, each folder description is clickable, allowing quick access to that folder.

Google appear to have adopted a similar approach. In certain search results, the site URL is dropped in favour of chevron-separated (currently non-clickable) breadcrumbs:

      Google search result showing breadcrumbs

The above directs to a URL (ironically, not SEO-friendly) whose "folder structure" doesn't even match the breadcrumb layout:
https://support.google.com/webmasters/answer/47334?hl=en

Are these signs that the folder structure is dying? Perhaps. But one thing's for certain: the chances of a user seeing actual folder structures or URLs are certainly less than they used to be.

rybo111
  • 1,059
  • 1
  • 7
  • 11
0

I do not know what folders are.

But hierarchical representation will remain with humans.

When humans want to model their thinking, they invent something called a mind map.

The basis of the mind map is based on hierarchical representation.

We hierarchicalize everything we want to organize, even our thoughts.

Computer operating systems did not leave out the folder, not even the slightest update. (For example, adding tag to folders.)

Next important point: you are talking in the representation layer. But in the underlying layers in some file systems folders are left out (Object storages [like:min.io])

Next important point: The structure of folders is not purely hierarchical. It becomes a graph using shortcuts

PersianMan
  • 101
  • 2
0

Databases have better performance and they can be indexed by many criterias.

Filesystems usually waste a lot of space and are slow and the indexing is delegated to the user... so the users tend to search the filesystem, which is slow, resource consuming, frozes the machine, etc.

Bill Schwarz
  • 1
-1

Folders are Dead, long live Folders.

Folders died on Windows computers in 2004 with Google Desktop, on Mac in 2005 with Smart Folders, and in mail in 2004 with Gmail. However, all of these look to the user much like folders.

Social media is another example of the use of labels instead of folders. Just because you've put your post in the #yolo folder, it doesn't stop you from also keeping it in the #swag folder.

About the only piece of kit that is not going in this direction are smartphones, but I very much suspect they will catch up soon.

Search, don't sort.

Folders are dead, but they will not be replaced with an app style structure.

Scott
  • 389
  • 2
  • 3