Peter, that functionality exists. You just use:
find ~ -name ‘*’ | xargs fgrep -il ‘world war 2’
Very interesting article 
The observation about people not understanding Explorer much less My Documents, etc. is very true. I was trying to explain to my accountant the other day how to move some files from his old computer to new one and ran into the exact same issues.
OS X’s Spotlight feature removes almost all need for using the file system on my mac. If I can remember the name of the file or any of the contents or the date or something similar, I just type it in spotlight and the file is listed.
Disks are kind of a hack, and not having to save file changes would be good. I am thinking that the change should be in the file system and not the applications themselves. We have loads of disk space now, and can easily afford to save a different copy of a file automatically each time it’s window is closed, with the file system keeping track of the changes and which is the most up-to-date file.
Jeff, I think if you watched Mac users, back in the pre-X days, you would have seen quite different behavior. The Finder (Windows Explorer analog) works better; more people “got” it and took advantage of it, even with a 32-character name limit. There were a lot of reasons for this - notably metadata of file type and creator, which eliminated the need to put cryptic extensions on the names, and a Find that behaved in the natural way (by name) by default, and a general sense of, “here is the starting point.” I knew Mac users who had the same kind of blindness to the file system, but they were much rarer.
Raskin was a brilliant man, way ahead of his time, and his early insights make me receptive to his later ideas, but I think he was wrong on this one. Don’t eliminate the file system; make it work better.
find ~ -name ‘*’ | xargs fgrep -il ‘world war 2’
The article is about hiding the details of the filesystem from users.
Do you think that a user who wishes to have the details hidden from them so they could more easily access their files would actually be the type of user to want to key in a command that’s 38 characters before you even count the search term itself, and containing such implementations-specific nonsense words as “xargs” and “fgrep” and “il” which mean absolutely nothing to a casual, nontechnical?
Personally, whether I want the details hidden from me or not, I don’t want a 38-character-long Find command.
Filesystems will probably always be with us, just like command lines. But in the future, we will look on file-systems much like the command line today. Obscure and vestigial remains from a brutal past.
Users have problems with the document/application split as well, although this is less of a problem now thanks to the “Create new” menu.
We now consider undo/redo to be natural. When Ctrl-Z doesn’t work, it’s a bug. There is nothing that says the undo stack cannot be persisted along with the document. The document can safely be saved automatically and continuously.
Think about the Drafts folder in Outlook. You don’t have to create filenames for items you save - they just appear there with all the text/images intact.
You described the Palm OS. On the original Palms, the title of a “Memo” is the beginning of its contents. Notes are saved as you modify them; even if you turn off the machine before clicking “Done”.
I think the original Palm OS is fine… I wonder why they screwed with it and introduced --yep, You guessed it— a hierarchical file system… complete with folders and file names. Totally non-Palmish. Fortunately, the latest OS is so buggy, I returned the thing and got me an old Handspring off Ebay.
And another pet peeve of mine-- is it “directory” or is it “folder”?
A directory is a directory. A folder is a GUI object that represents a directory.
For example, “Desktop” at the top of a Windows Explorer treeview of the file system is a folder, but “C:\Documents and Settings\Administrator\Desktop” is a directory.
First and foremost, millions of people have become accustomed to Windows Explorer and its counterparts for the other OS’s. So we must all accept that the existing filesystem interface cannot be scrapped to invent new data storage/retrieval model.
I agree with Daniel, and feel tagging is the best way to improve existing filesystems. Of course, like SomeGuy said, the problem with tagging is about the standardisation of the tags, and for this the OS developers must have some predefined tags for the user to start with.
The tags themselves would be like metadata for the files, and would need to be embedded with the files for easy sharing.
Also I feel search is going to another key ingredient of any future OS, and a killer filesystem would be one which integrates a fast, efficient and non-obtrusive indexing mechanism.
People who are saying that the metadata is unnecessary, and that the OS should look within the file contents… well, what happens for a password-protected document then ?
Ok I love filesystems. I don’t know why anyone would want to get rid of them. All the problems listed with them seem to be user issues. I don’t use the My Documents or anything like that but that’s because I keep my apps on my system drive and my data on other drives. If you had to search everytime you wanted to open something that would take forever and you would have to have more clutter on your screen. Beagle the search program for my linux box already searches within the files for the context of my search. But even though I use linux I still think the windows file system makes tons more sense which means i think it’s better than the mac filesystem also. It only makes sense to have everything seperated by drive then application folder. Not saying that I prefer windows to mac or linux it’s just a better filesystem. I could do without extensions but then that would confuse even more people. It’s so nice to know that my mmc plugins are in the system32 directory and that I can go straight to them when I need them no pesky searching involved just instant gratification. If you can’t remember what you named your files or where you put them you need to be more organized. If you can’t remember enough about the file to find it with the current search programs then it must have not been that important in the first place. Don’t blame the filesystem the PEBKAC.
Hierarchical filesystems are useful for archiving, not for day to day work. Desktop search apps are a step in the right direction, but unfortunately flatten time in favor of “space”. Outlook’s journal timeline features are sort of useful, but flatten space in favor of time.
When working on a project I may be dealing with hundreds or thousands of files many of which have vastly different metadata. For example, frame 1024 of the animation I just rendered for client approval vs. frame 1024 of the test animation I did for myself that I liked, but knew the client wouldn’t, but which I wanted to reference for a different project…
Even with an intern/production assistant dedicated to asset management, after a week the associated metadata loses sync with the real world.
Even in the early 21st century the best filesystem I have is my desk. While not searchable via keywords, tags or other metadata, I can find material immediately. Can’t help but wonder if the real problem with computer file management is that the display itself is too small - even dual 30" monitors don’t supply nearly enough room, but what if my display were as large as a wall?
Very nice replies above, but since our man CH has already made a new blog post, I guess this one is closed.
Choosing a filesystem is in most cases the choise of the underlying operating system. The choise of the operating system is often compared with religion. What you have grown up with, what you know and what you understand. Probably like several other people in this thread I have used several different operatingsystems and I enjoy using the filesystem in both development and standard cases.
As said, if it works dont mess with it. But I have seen so many cases with amatour computer users where the normal systems just dosent work. Maybe a bigger leap in the differece between home edition and professionel editions of the operating systems could be a good solution. But as someguy said, it is good that the creaters of the operating systems will fix it for us.
LOL. I’ve been wondering the whole time reading this when someone was going to mention Linux. Yes, it (imho) organises the filesystem in a much better way. Sometimes there is confusion with /bin, /usr/bin and /usr/local/bin but if you need to care about those specific ones you are deffinately classified as a system administrator (or power user at the very least).
As far as I can recall WinFS was supposed to address many of the issues mentioned here, and I’ve heard people claim that ReiserFS 4 (not yet in mainline Linux distributions) can already do a lot of what is suggested here. I may well be wrong on this though.
Having thinking about, and covering this topic myself recently, I have to disagree.
Besides the fact that the distinction of where the file is stored is immaterial (that is, a disk or nvram won’t make a difference from a UI perspective in any sense of the word), a file system is open-ended enough to support any form of organization.
In my own writings, I asserted that the stricter requirements on organization provided by modern-day operating systems is to blame, combined with so much abstraction by the OS to keep users from having to comprehend what the filesystem is for to provide themselves with an organization scheme.
The end result is really chaos - users who have crammed 400-500 documents in the “My Documents” directory and don’t understand why that receipt template they wrote in 2000 is so hard to find. If you want to look at an analogue, look at people’s non-virtual filing systems: while their collections of books and magazines, and other media might be horribly disorganized, their information on their taxes has efficient organization applied to it that rivals most large corporations.
Why? Because the IRS and any other organization that needs information on their income is going to smack the crap out of them if they don’t. In these days when what a computer contains is sometimes more important than paper renditions of the same data, why should people be encouraged to take organization any less seriously?
What’s funny is that all these discussions on removing the methods that traditionally make up a filesystem don’t even consider backups, access control, or anything beyond the “please spoon feed me” mentality.
I’m not for punishing users by any sense of the word, but actively encouraging people to think less doesn’t really help anyone get anything done better or faster. What we really need are systems that make it easier to organize data quickly and effectively, not systems that render organization pointless.
Let me know when spotlight or google desktop search can find all files on the disk that have to do with english class, spring term, 2005, without any messing around with the filesystem or metadata for the files beforehand, as those would constitute standard operations on a file system.
the (non-)distinction between shelf and self has been implemented in *nix relational databases for neigh onto 20 years. it’s memory mapped files; generally one for the data, one for the indexes, and one each for before and after images.
that M$, not my friend, wants to implement winFS as database/OS (not really a new technology; AS/400 and predecessors were built just that way) is a recognition that RDBMS really is better XML and its cousins (which are Older and More Primitive Technology).
the IBM mainframe doesn’t have a filesystem as PC/Unix users see it. Multics introduced the concept. both it and unix were designed to a word processing (document) paradigm. and thus we have descended into hell (the assertion/assumption that the world is hierarchical). it’s not linear, either. but that’s for a thread on statistics.
This concept has been considered before in earlier days with simpler file systems - pre windows - and there are inherent problems.
However, a variant approach is the ‘black hole’ method where files are saved to or dragged onto a virtual drive/icon, similar to the trash can.
This Black Hole is a script driven manager that decides where files are to go based on preference, version, content, time/date, template-fitting, job type, replication/backup etc, effectively removing the manual side of naming and storing files.
The eventual location is still going to be a known destination and folder set for the techies. The user can still use his applications to the read from the Black Hole, which responds in kind to the requesting application, delivering information about files according to most recent date/version info etc, also via preference and scripting. His fall-back position will always be the actual source directory if the BH isn’t playing ball for him.
This was drawn up in 1993 but Windows and PCs back then weren’t really lending themselves to the kind of storage, object and processing needs that such a concept needs.
There must be enough Linux geeks out there to implement this kind of thing, if nothing else to show that there is something that Linux can do for the common man that Windows doesn’t.
Maybe its been done, I don’t know.
On the other hand, the geek in me is quite happy that users can be fundamentally clueless about even the simplest of things, as so work is abundant… The man who has worked for 20 years with PCs but still can’t pop up a DOS box and enter just one command - I call him “Boss.”
I agree with Christian Mogensen above, with regards to the hiding of the filesystem. Ever since I started using Linux, I’ve been impressed with how compartmentalized everything is. Log files are here, user applications are over there, system applications are in this other place, personal data right there, and config files up here. Linux has a ways to go before it can be considered really user-friendly (though my Ubuntu install is pretty intuitive), but I can see the home folder being one big flat directory where all the documents are stored together. Then a search/tag/AI interface can be built on top of that.
This is an extremely interesting discussion and I like reading the various viewpoints.
In real-life I am a disorganized mess. The pile around me builds as I move from one task or project to the next; closest items represent the most current contributions to and components of the most current nightmare; older and less-relevant stuff is underneath or at the fringes. One can easily determine ratio of relevance between mess and project by distance from PC or top of pile. Occasionally my mental OS overloads and I must reboot-- stop all work, reduce the mountain of crap down to a bare desktop and start over.
My PCs and server are a different story. I have a hierarchal storage and retrieval system with meaningful names that is better organized than anything the IRS could imagine (nod and smirk to previous post). In a few clicks I can find documents going back several years on any subject. My PC organization is the envy of my more mortal friends (heh).
On the other hand, my wife is very organized in real life. Before she moves to a new project, she cleans up and puts away the detritus from the last one. My side of the room is chaos; hers is order. I am in awe of this. And yet, on the PC she can’t find a document she constantly updates from one day to the next. She can’t even find Microsoft Word unless I stick a prominent link to it in the Start menu (actually a link to said document that launches Word). When I began beta testing Office 12, she had a meltdown upon encountering the revamped (and dramaticlaly improved IMO) interface. She balks at using it. I need to remmeber to change the default application launch back to Office 2003 before she does something irreversible to the PC in a fit of pique…
Anyway, any new file system development should take into account the differences between the ways people work. IMO I should be able to select the file system that suits MY working habits, associate it with MY profile, and take comfort in knowing that every application on the system can’t work that way, too. Everyone else can do as they please with their own profiles. La de da.
OOps, I erred in the post above: “knowing that every application on the system can’t work that way, too” should be “knowing that every application on the system CAN work that way, too”
Interesting idea…but STUPID.
Every time developers try andmake things idiot-proof all they do is create better idiots.
Novice users have already been way too isolated from basic fundamental concepts that they really have no idea what they’re doing. He even mentions that many users already are clueless, so his solution is to make them even more clueless tomorrow?
With everything that files entail nowadays, MP3 files, pics of family, web pages, email attachments and how often people want to copy, print, send, share, edit the ideas he has seem unfathomable.
Inevitably every user eventually ends up having to learn what/where/how files are all about when they want to…find a file they downloaded…print a copy of that picture of grandma…copy those cool mp3’s to their new widget…send something to a buddy. The harder you make it for them to find, understand and access the more apt they are to get frustrated.
To me the worst crime of recent history to keep new users in the dark was when XP stopped placing the My Computer and Windows Explorer Icons right on the desktop. I bet everyone reading this has been asked to help by a newbie friend or family member to perform a very fundamental task like this and has seen the look of desperation and bewilderment on their face becase the newbie has never been introduced to the basic ideas of what files are, where they go and how to manage them.