Filesystems Aren't a Feature

blog
41

Great discussion. There is an answer to all those problems of not finding back documents, laziness of users, etc.

  1. Hierachical file structure as a tree is outadated. Nowadays, all the files should be put end to end on the disk, with metadata, and what’s the name of this : D-A-T-A-B-A-S-E :slight_smile:
  2. this “new FS” will be good only if the user is OBLIGED to enter informations. Otherwise, they’ll do like in the past, but instead of saving in the default location with basic name, it’ll be saving with default content of all the metadata… So we oblige users to fill at least a file name and a few other params. Other ones like date, last revision, keywords, resolution (if image) etc… will be done automagically by the system.
  3. Think of how organized users save their documents. You have main themes (f.ex games, docs, professional…). Then sub-themes (for docs for examples: my mails, my letters, my excels sheets)… Then all sorts of sortings (by author name or albums for MP3s, by genre or year for divx movies). So the new FS have to propose tagging the file with “common” tags (personal or profesional, names of persons the doc is written by or intended to, topics, actors in the movie, etc etc…). each tag will suggest other tags. For example if you add the “movie” tag, it’ll suggest you to add the year, actors…
    On top of those common tags, let’s add “custom” tags for power users… and we’ll get a wonderful DB.
  4. We add a support for history of documents
  5. We use a visual on-the-fly 3D representation of docs. When you enter, you see boxes with words (one per tag, biggest ones are shown). Click on a box, the room is filled with all the main tags of the docs having the 1st tag, and so on…
  6. Downside of this: the sarch engine must be intelligent enough so that if you tagged “music” instead of “mp3” he finds anyway… AI !!

My 222 cents :slight_smile:
Example: MP3, Lagoona, dream = in 3 clicks I got it

42

2 things. There’s a word processor called Yeah Write! (http://www.yeahwrite.com/) which inverts the whole filesystem thing. Instead of requiring you to enter a filename, the program saves the file constantly. You can’t lose a keystroke. The interface is that of a filing cabinet. You click on a drawer and the drawer opens. The open drawers has a series of tabs, each identifying a different kind of document eg. journal, memo, letter etc. You click on a tab and start writing your document. Since you never have to save it, you don’t need to worry about filenames. The first line you write becomes the ‘tag’ or filename internally.

You organise your files by creating new drawers.

I’m not sure if you can create new templates or if you have to buy them.

Yeah Write! isn’t perfect, but it is different and very easy to use. In fact, it’s hard to misuse.

It incorporates some of the ideas discussed here, except for version control and search.

My ideal filesystem would automatically keep versions of all documents. The last one saved would be the current one, of course, but you should be able to pull any version and make it current.

The post above with the ‘find’ command, wasn’t really being facetious. It’s more that sometimes a command can cut through all the bullshit more quickly than a GUI widget.

I agree that the computer should do more work for us. It “knows” its limitations better than us. More work needs to be done on making computers aware of themselves.

I’ve often thought that computers should introduce their capabilities to the user. “This is what I know how to do. Out of that list, what do you want?” Icons help, but icons require the user to understand what the icon means first. The user must also understand the concept of icons. Sometimes a simple menu works much better.

On the AS/400 system, every screen has a complete help system behind it. I know green screens have gone out of fashion nowadays but it was very hard to get lost on an AS/400. It was so customizable.

I always want a computer to learn how I work, and set things up for me. I shouldn’t have to do things twice. It should have monitored what I did the first time and automatically created a path or menu or workplace or whatever to make that available to me the second time.

So it should know if I write a letter and then run spellcheck, the next thing I want to do, is to print. If I do it five times in the same sequence, it should know that I’ll do it again. Why not? What else is it doing while it’s waiting for me to finish the letter?

If a guy pulls up the same spreadsheet every morning, it should just pull up the spreadsheet the first time the mouse is moved. If he does email first, bang! email first.

43

File saving and searching has changed dramatically because of Google Search and Windows Desktop Search. Simply put the file name has less and less relevance when the file content itself is indexable and fully searchable. I no longer care what my file name is or even what type of file it is.
I only care about the contents whether that happens to be data or the metadata … although someone still needs to solve the issue of video and picture tagging.

Great topic!

44

One more point … FileSystems are not just storage locations. They also serve as metadata for related items. That is, one could have a C# project, several supporting documents, visio files etc, all with data that is related but the relationship may not include relatable keywords. Well in this scenario I would still like to associate these file with each other.

I could do it through tags within the documents but why would I when I could put them in a folder structure that implies that they are related.

Folders and files are good, I think our problem is searching, and searching by file name is dead!

45

Very good discussion.

I recently started using a writing project management tool called Scrivener (LiteratureAndLatte.com). It auto-saves when the keyboard is idle for more than 2 seconds, a great feature which stopped me from losing data. This is only a great feature because Scrivener ALSO does many levels of undo. (Writers often make changes and then decide the original version was better. Auto-save without multiple undo is worse than no auto-save.) Scrivener can also snapshot files, so you can save a version, do days of changes, and then grab content from the original if you need it.

Scrivener eliminates the necessity of having filenames by keeping documents for a project inside a “project binder” (actually a folder). You can thus have an idea, create a new document to write it down, and NOT name the document. I had no idea how onerous creating filenames was until I stopped having to do it. In other programs I will sometimes postpone saving my work because doing so would require breaking my flow of ideas to create a $#% filename. And if I use the default filename, I have to go back and fix it later.

Another great Scrivener feature is the ability to do a search, then chain all the relevant documents together into one long document to skim through the contents.

As a writer, I love the idea of being able to search by content. In real life outside of Scrivener, it hasn’t worked so well. Until recently, I kept incremental backups of projects I was working on in my Documents folder. Since I’ve owned a Mac since 1987, the eventual result was a folder with over 300,000 files in it, most of which were incremental backups of OTHER files. This made content searching virtually useless. (Imagine you googled something and got 300 hits of different versions of the same Wikipedia page. Ack!) Instead, I relied on filenames.

But filenames and folders suck, for a lot of reasons people described above. Even with aliases, it is unnecessarily difficult to have a given document grouped with other documents of project A, and also with other documents of project B, and also with other reference documents on the same subject.

I like the database-with-tags idea a lot. To work for me, it would need an automated way to search ONLY for the most-recent version unless it either doesn’t find the wanted tags in the current version, or I include older versions in the search.

I also do psychological research that involves perceptions and strategies. People think about and interact with technology using widely varied strategies and mental models. Based on their information processing style, strategies, and mental models, people need different interfaces to make sense of information. Think multiple intelligences. For instance, some people need to SEE files to remember them, whereas others can tuck them away in folders and simply remember that the folders or the categories they represent exist.

Think of a physical desk. Some people organize a desk by piling stuff on it, some by filing everything away and having a minimum of stuff on the surface, some by sorting stuff into pigeonholes, and so on. The desk provides a work surface, but doesn’t require the user to use it in a certain way. A desk that REQUIRED piling, or filing, or pigeonholing, or handing everything to a secretary for them to deal with, would suck for every user that didn’t use that exact working style. And also for users that DID use that style most of the time, but occasionally need to do something else.

Computer interfaces need to be like the desk, and provide a variety of interaction options, EACH of which is simple, comprehensible, and easy to use. I think we’re a few years away from that, though… :wink:

46

While i have read “about face” and completely understand that hdd’s are a hack and not part of the best implementation for a computer i tend to disagree with the “hide it from users” paradigm, of course some information should be hidden but the file system is very simple to understand, most 10 year old children have an understanding of how it works. What’s needed until the current file system is replaced is better training, i have found many people who work in an office eight hours a day have no concept of what a file or folder is but a few hours training would explain it adequately and if a user is unable to understand such a simple concept after that they have no business being anywhere near a computer in the first place.

I mean really it’s not rocket science, files go in folders which you can make rename and delete and when you open something you are opening a copy and it is a copy till you save it. is it really worth catering to people who can’t understand this.

47

The problem is not with the file system but with the users. Unfortunately, many users simply do not spend adequate time learning how to navigate and utilize the file system. I don’t think we should compensate for thier shortcoming(s). A file system definately has its place on the personal computer, whether it be the current model (which is somewhat old) or a new design. How much an application wants to abstract the file system is an individual/team choice. Certainly, most users have no concept to what it is. Searchable content has always been a problem which some companies are attempting to solve with products that abstract out the file system.

Most people just surf the web and do email, in those cases, I think an appliance without a file system or a hybrid system would be a better choice (not a PC) for them, but we shove computers packed with bloatware down thier throat at Circuit City, Best Buy, etc. A simpler device would serve the masses in a better capacity.

48

Hierarchical data in many folders is faster to find than just loads of linear data in one folder. This is logical and it is commonly used in search algorithms. One problem I have come across is when I have lots of files, versions, updates and patches, scripts, config-files, scrap folders, work folders and so on for ever. Then I do not remember what was the hierarchical ordering system I was using and how was it applied to folder structures and files. Also I am not sure, if this structure is applied into all parts of my folder structures. From time to time I find better ways to organize files, but there are lots of previous files and their copies. Now, I could write up the organization method. Then when I change it, I would have to manually restructure all my folders and files. Maybe future operating systems could help me in that work so that it keeps record of the organizing algorithm that I have created. The OS could also apply it to the folders on command when I change the algorithm. Also the file system should be able to be locked according to MyFolderPlan v 3.4 so that the folders stay clean and organized.