Your One Stop Keyur Shop!

yellow 6

Yea I know I just got this website going again, but bear with me, this entry is near and dear to me.

John Siracusa at ArsTechnica is at it again with another critical look at the OS X interface. Unlike a lot of reviewers he clearly states at the very beginning that he does not like the Finder and goes into the review with that assertion. So Siracusa is arguing that the OS X Finder has a huge shortcoming due to the fact that it loses spatial orientation thus causing major UI woes. I'll take his argument one major point at a time...

1. Goodness of spatial interfaces
I cannot argue with the goodness of spatial interfaces. Even under tremendous clutter the user who created that mess will find what they are looking for because it was where they put it. However Siracusa needs to address the major shortcoming of spatial interfaces which is the lack of screen real estate. The light switch analogy to illustrate how muscle-memory is formed was great, but you never need to have light switches that control more than 2 or 3 sets of lights together. Whereas on computers, users often have to manage large sets of files. Screen real estate becomes a major issue when you have a large number of folder windows open, especially considering that there never has been a method in the MacOS to immediately switch to an open window a la the Alt-Tab and Taskbar in Windows. MacOS only allows users to cycle through each running application not each open window. Evolved human skills at managing visual information doesn't help much when you've run out of screen space. You can make excuses about this point, but a busy screen does not facilitate ease of operation especially a task like file management, which is one of the core functionalities of the MacOS Finder.

2. Coherency and Stability
The "one-to-one relationship between folders and windows" is a great relationship, however it does not work well in the Unix environment. Unix filesystems and HFS+ support hard links which allows the same file to exist in multiple locations at once. To some extent this frustrates the one-to-one relationship, because now the same folder or a file can be two places at one. Also with the web being as pervasive as it is today, users are familiar with the concept of a text-based address. Whether that address be http://www.mywebpage.com/ or /home/myfile/mymusic, a filesystem path doesn't necessarily confuse the user. I'm not advocating that users be unnecessarily exposed to text-based path names, but rather pointing out that users are familiar with text based paths and use them regularly. Siracusa cannot be denied though regarding his points on the stability and consistency of appearances. Changing how the same folder(s) appears from one double click to the next does not work with or without the one-to-one issue, its really that simple.

3. Live Searches
Awesome concept, but poorly suggested. Live searches don't make much sense based on what type of file data is offered by the OS today. Those data being: creation/access/modification date and time, file size, permissions, location, and type. Its hard to think of the usefulness of a live search based on this dataset that a user would find much use with on a regular enough basis to make a folder to watch regularly. Also this list is bound to be littered with random files such as web browser cache files and other files that have no real use for the user. However think about the concept of Live Search with a query like: all text based documents older than 6 months that contain a given string or all pictures larger than 50K of my dog. That would be far more powerful not to mention useful. More on this concept tomorrow. Another take on the whole concept of Live Search would be to have search bookmarks so to speak with a search can be bookmarked like a website, but I love the idea of a specially badged folder that updates a specific search query in realtime. I don't think bookmarked offer as much functionality as the Live Search folders.

4. Proxies
It seems to me that the very concept of a "proxy" icon that is not an alias kills the whole spatial orientation idea. This "proxy" icon gives user two ways to get to the same folder/file. One is the folder/file itself and then you have the "proxy" icon in the live search. This confuses me because this seems like its negating the spatial orientation and outright breaking the one-to-one relationship requirement.

5. Finder Browser
Why deliberately obfuscate the concept of file management? Windows does this and it doesn't offer a clear benefit. Also the concept of Navigating is another idealogy which is an option to spatial orientation. I simply don't understand why mix and match like this. Siracusa throws in the address bar when all this time we've been on the topic of spatial orientation precisely so that users don't have to worry about text-based path names, even if it is just for expert users. Ambiguity is the best way to ruin an otherwise good interface. Windows has a thousand ways of proving this. Design one file management concept and run with it don't patch up a poor design with another concept thrown in for good measure. Ambiguity in an user interface is not the way to go. The Shelf that Siracusa discusses is another concept I disagree with. Although the desktop was not designed with this intention in mind it serves the exact purpose of the shelf perfectly without the need to add more UI. I feel its not good to add more UI when there is already an established concept that also serves this purpose without creating any extra problems or UI.

6. Customized Browser Windows
This is a can of worms just waiting to be opened. A lot of people have asked me and I've thought a lot about whats wrong with blurring the boundaries between the OS and applications. Its sort of like creating a car thats also an airplane, both of these machines have opposite goals. One needs to stick to the ground the other one needs to lift off the ground. The task of managing files and the task of editing/creating files are different in their very nature. Microsoft's UI/UE team would argue with me about why a user should even have to worry about files and where they are and such. Thats fine if it could be done perfectly, however there is no such thing as "perfect" technology. Data that users go through the trouble of creating or acquiring is important, no questions asked, data loss is simply unacceptable. However for the forseeable future user's should have more or less direct control over where their data is stored. That doesn't mean they should be forced to manage their files, but managing files should not be made harder or abstracted confusingly. Also when you have a special folder window that becomes a psuedo <insert simple editor type> things quickly become more complicated than necessary. It is not worth while for the user or the designer to have a system, "that is a jack of all trades but a master of none." Another issue that arises when very specific application level functions (image editing, document editing, etc.) are blended into the OS it becomes questionable where the line should be drawn. At which point does the OS stop being the OS and instead become every imaginable application. Microsoft for one is definitely going towards that vision where the OS is the end all and be all. I cannot agree with that vision because its not realistically possible for one interface to work for everything. On second though, forget doing everything. Its not reasonable for one interface (based on the WIMP model) to do more than a few similar tasks well, let alone a lot of dissimilar tasks.

7. Finder Plug-ins
This is a gray area much like skins. In most instances I'm not a big fan of skinning because its messing with a good thing. Yes its nice to have the option but in the long run its just not worth it. Its also prudent to remember that when you have an open API that adds functionality to one of the primary OS interfaces you are bound to run into some major headaches like poorly designed UI, poorly written code causing stability issues, etc. Let the OS do what its best at doing and allow applications to do what they are best at doing.

and in closing...

In Defense of the Finder
A lot of people have <insert list of standard OS X Finder compliants>. However most people forget that the Finder is the computer even for a lot of power users who don't want to get dirty with the command line. The stability of the Finder is priority #1. The Finder between OS 9 and OS X is completely different except for general apperances. For all intents and purposes the Finder is being recreated and a lot of the <insert list of things OS X Finder is missing> will take a while to reproduce because stability cannot be compromised. Apple is too smart to throw away what they've learned over the nearly 18 years that the classic Finder was maturing. The OS X Finder will get there.

Quoteses:
"So you didn't mean to say 'naan-sequitor'?" - Ram. K.
"I'm already convinced that Mozilla is the way to go. I haven't used IE since I installed it." - Gautam P.
"Good luck kid, come back if you ever need anything." - Erik O.

Linkses:
Bush's human rights hypocrisy
To the victor go the spoils
What more could I ask for?
Awesome print adverts we never saw.
Great preso, cute penguins.
Theres a reason why Altavista sucks now.
I purchased a push lawn mower and only used it once when the engine blew up.

Posted by Mr. Keyur at April 2, 2003 11:43 PM | TrackBack