The Space Toast Pages

Matthew Rasmussen's journal of journals on various topics of interest, published here, there or somewhere since 1999.

Mac OS X Interface Criticisms

For the statement of purpose, skip to the end. Let's get into this...

• Slow interface elements. When you click on a menu in Windows XP, there's often a "meh" moment where it hesitates before opening. As far back as System 6, clicking on a Mac menu meant, menu: now. Two Mac interface elements have always been painfully slow, though: Dashboard and Spotlight.
• Dashboard. Great idea. Bad execution. I always found Dashboard unusably slow on my G5, but figured I'd use it all the time when I got an Intel. I was half right. It's about twice as fast. Which means it can still spend 10 seconds thinking about nothing of apparent difficulty before letting me type a word into my dictionary. I only have six widgets. Dashboard needs to be rewritten from the ground up for instantaneous speed.
• Spotlight. Far too ubiquitous in the system. If I want to find a file, I go to the Finder and hit the standard key combination Command-F. If I want an address, I go to Address Book. If I want an email, I go to Mail. If I hit Command-Space, anywhere, any time, it's by accident. Usually, I don't even mean to open the Help menu. Lurking behind either accident is a judder as the Mac thinks hard about opening... a search box.
• The default Finder window. What IS all this crap? The titlebar is half an inch tall. I've got a search box which, if I click on it and begin typing, MAY start searching in the current folder, or may jump out to the whole disk. It may decide to do Contents or File Name. If I "x" out of it, it will dump me back to list view when I was in column view. There's a gear-widgety thing that I've never used; it does nothing but duplicate the File menu. Most of the stuff on the left makes sense, but once again I've got search; can we just pick a place for search and leave it there? The bottom of the window is another quarter inch tall, and empty except for one line of text that could fit a dozen times in all the wasted space up top. Finally, there's a little tic-tac in the top right that hides all the useless stuff -- as well as the useful stuff in the left-hand pane.
• Disk Images. I'm old enough to remember copying between two disks on a Mac Plus with no hard drive -- using "Eject" instead of "Put Away" so you could drag and drop a file from one disk to a greyed-out picture of the other, and spend the next several minutes swapping disks physically. Why do I bring this up now? Because the virtual disk concept is lost on all but the most propellerheaded users. Look at your mom's computer: she's still got every installer and .dmg file she ever downloaded. She doesn't get that you download a file, the file becomes a "disk," you copy from the "disk," then you "eject the disk," and delete the file. (No mom, deleting the disk image won't delete the copy of Firefox you just installed...) Applications shouldn't have installers and outside-the-bundle files in 2010 anyway, but that's a much wider abuse of the Application concept in modern graphical user interfaces.
• One-click application opening. Have we forgotten how to double click? How often do you accidentally begin the slow loading of a large application because it's next to the one you wanted on the dock? Here's a consistent workflow: Hover reveals nothing new but indicates that the object can act; Click hilights the object and reveals options; Second click performs the default action. (In an application's case, starting up.)
• Dot-3 file extensions. What kind of teletype-era UNIX cancer is this? The file NAME is the file NAME. Nothing. More. The file TYPE is the file TYPE. You had this solved in LisaOS, never mind in the first version of the MacOS that was written when I was four. Two hidden pieces of data, each more than three characters long: The file type, and the file creator. The whole point of a graphical user interface is that the file system shows us -- with it's graphics-ical things -- what type of file we're dealing with. Which leads us to...
• Thumbnail icons. Unless the image the file contains is not much more complicated than an icon, an icon-sized preview is unlikely to reveal much of value to the eye at a glance. Below a certain size, say about that of your fist at arm's length (your thumb thumbnail is far too small), images become jumbles, like a big table of numbers. My text documents all have thumbnail icons. They all look the same. Their icons do not reveal the file type or creator, which was supposed to be the point of an icon. It's possible we need to abandon miniature icons in their entirety.
• No application should ever steal focus from another application. Period. If I open three applications and then click back on the one I was working in, I should never be ripped back out of it. If I insert a CD, iTunes has no right to demand my attention whenever it gets around to deciding what the CD is.
• The jumping icon on the dock is much too annoying for developers to have access to it. Too many abuse it. Progress finished? So f***ing what? Leave me the hell alone. Apple's own Software Update abuses it, constantly demanding attention until it's allowed to tell the user that, eventually, a restart will be required.
• The Finder vs. the open and save dialog. The special view of the file system in the open/save dialog made beautiful sense when we could only have one application open at a time. That was twenty years ago. There's no more need to duplicate the functionality, especially in a truncated manner. How should we decide where a new file is saved to? We click Save, and the top Finder window pops up with the file awaiting a name.

And briefly noted:

• If an application is going to take more than half a second to open, I should be able to tell it to stop.
• It's easy to browse FTP servers from the Finder, and it feels like a natural use of the filesystem. So why is it read only after so many years?
• Safari Command-arrowkey navigation doesn't work when a text box is hilighted. The page itself gets to decide when a text box is selected.
• Connecting to and rebuffering an internet stream locks up the entire iTunes interface with a modal dialog.
• *Sorting in column view. I'm asterisking this one because I understand it's finally been implemented in Snow Leopard. The trouble is, we should have been able to sort Finder columns the same ways we can sort Finder lists a long, long time ago.

I was making Hypercard games when I was ten on a Mac Plus. I learned Animation:Master when it was Playmation on a Quadra. I was modding Escape Velocity with Ray Dream Designer and ResEdit on a 60Mhz Performa. I won most of my film festival awards for a 12 minute short animated on a 500Mhz G3 iMac bubble. I freelanced after college with a G5 tower. I remember Strata, KPT, Aldus and Fractal Design, and I'm old enough to remember the MCP when it was just a chess program! I may know what I'm talking about.

How to Tell If You've Been Ripped Off by the Developers of Your Corporate Site

It's thoroughly embarassing what even Fortune 500 companies will accept from their website designers. There's a lot of snake oil out there, like Twitter and almost anything to do with Facebook, not to mention the Russian roulette of gaming search engines, but we're just going to look at your website. If you're a not too tech savvy manager trying to figure out if your vendor is ripping you off, read on.

• On any page, go to View Source, and search for the word "<table>". Did it come up? If so, is there any actual spreadsheet-type data on the page? No? Then you've been ripped off by a firm that can't use modern code to build a site. They'll want twice what they should be charging for any minor layout tweak, because they'll have to practically rewrite the page from scratch every time it needs changing.
• When you enter a phone number in any reasonably logical manner into the phone number field, does the site fail to understand it? Does it give you an error, and tell you how it wants the number entered instead? If so, you've been ripped off by a vendor who thinks your customers' patience is less valuable than half an hour of its billable time. Teaching a computer that 555-555-5555x102 is (555)555-5555 ext.102 and 5555555555ex102 is not taxing. An incredibly badass phone number validator might run to 50 lines of code. Might. Five minutes on Google will turn up dozens of good ones, ready made and free to use. See also: Social security numbers, dates and zip codes.
• Does clicking the link to log in redirect the user to a different website? It's easy to create a fake site at a plausible-looking address. There are thousands of them on the web at any given time, designed to trick the user into typing in their account information. Redirecting from TinyCreditUnion.com to login.2MyAccount.co.cz/TinyCreditUnion?=mainsite?=login.redirecttome.com goes against every shred of good internet consumer advice in existence. If you train your customers to make bad decisions with their account information, some of them will.
• Does a password have to contain numbers, symbols, etc? Bear the following in mind: There are lists available on the web, compiled by legitimate security researchers, of what the most common passwords are. Beyond screening for those, the difference between the password "buylamps" and the password "4f9s^fg)3" is statistically meaningless. Brute-force attacking an eight character password containing nothing but lower-case letters would require an average of over 104 BILLION attempts. If the site designer is smart enough to take the tiny precaution of ignoring more than one login attempt per second (which your customers will never be bothered by, on the receiving end) it would take 3,309 years of constant guessing just to get an even chance of breaking into an account.
• Is there a meter that tells you how "strong" your password is? It's bullshit. It's calling 3000 years "weak," 30,000 years "fair" and ten lifetimes of the universe "strong," and chances are it's not even checking against a list of common passwords. If the small steps above have been taken, it's as meaningless to you as mathematician talk about the different sizes of infinity.
• If you enter the wrong password a few times, does the account lock you out? More fake security. The obvious trouble is, it's only ever going to be invoked by already pissed off customers who can't remember the machine-exact keystrokes with which to enter this particular password on this particular website. I know you want to think of yourself as the center of your customers' financial lives, but you're not. So unless you want to pay someone at a call center to deal with every minor issue the website was designed to handle, fire the bums who try to sell you customer lockout.
• Does the password need to be changed at regular intervals? This is not even fake security, nor merely irritating: it's actually negative security. Your customer can't memorize a new password every month (nor can you) never mind half a dozen. That means the password has to be written down somewhere. Which means it's not a password anymore, it's a bearer bond. Suddenly, leaving an address book in the restaurant, losing a $10 thumb drive, having a purse snatched, or just plain leaving a sticky note in the wrong place can compromise half of a customer's financial life.
• Does the site use security questions? If you can make a security question specific enough to be unambiguous (mother's maiden name) it will be a matter of public record. If it's any vaguer than that (your first crush), it will lead to so many possible answers that no customer will be able to consistently enter the same, correct, exact, machine-confirmable string of characters in answer to the question each time without -- again -- writing it down. Most users settle on a consistent string of profanity, which is guessable and still usually needs to be written down. Guess a few strings of profanity yourself if a vendor ever tries to sell you this.

UI: R. Clayton Miller's 10/GUI

Problems:

1. You just doubled the amount of space I need between myself and the monitor.

2. Multitouch allows for more kinds of interaction: true! However, this interface steals ALL of them away from use by the applications.

3. Left and right sides of the screen aren't discoverable. Might as well be top and bottom -- i.e. bottom of the screen for application launching (call it a "dock") and top of the screen for context-specific options (a sort of "bar" of "menus").

4. Linear spatial overload of windows is no better than two-dimensional spatial overload of windows. Labelled zoom-all-the-way-out cheat no better than Expose and application switcher.

5. Where does file management fit into this scheme?

Lukas Mathis calls 10/GUI "one of the most dramatic reimaginations of the desktop user interface I've seen in a long time" but on examination it's an incremental hardware update with no real interface breakthroughs. Keyboard + mouse has gone on for far too long, as has the W.I.M.P. interface. A better direction would be a tactile multitouch surface which can be anything it needs to be, including a keyboard (for any language), coupled with a GUI that represents tasks and actors rather than objects in a space. 10/GUI does nothing about window and document clutter, squinting, scanning large lists, or making the computer's workings and status an organic part of its presentation. The video may be a slick investors' reel, but shows no real progress.

Email to Twitch Film

Reader for about six months. Love the site content, and the hard labor of love work you guys put into it. I've discovered so many great films because of Twitch.

Hate the new layout. Here's why.

The readable 1-3 paragraph intros of the previous format made it possible to browse articles and glean a bit of information about each project. The nice big images were equally browsing-friendly. It was much easier to guage your interest in an article without additional pageloads. The wall of tweet-length teases and postage stamp-sized images in the new format provide almost nothing in comparison. The new layout reads more like a reference site, where individual articles may be teased but most readers are expected to come for the search feature, than a day-to-day news blog.

The Slashdot/BoingBoing-style blog layout was a much better fit for the great content you guys provide. I hate seeing the alchemy of SEO plastinate another great site into a cluttered 2000-era portal.

Return to SpaceToast.net