Hacker News new | ask | show | jobs
by cytzol 1853 days ago
This post illuminates two ways in which users' expectations of computers have changed since the '80s, which is when this UX was designed:

1) Users don't deal with external storage nearly as much anymore, so the "disk image" paradigm feels out of place. Before the Internet, it was more common to insert a floppy disk, drag the application from the disk to the hard drive, and then eject it again. Users were accustomed to opening and ejecting disks based on the icons on the desktop, so when disk images started to be a thing, it worked exactly like floppies did.

2) Users think of applications that have to be explicitly installed, rather than just copied, like any other kind of file.

The second one is the really surprising one: when we think of how to get applications onto our computers these days, the word is "install". You need to install a program before you can start using it. On Windows, you ran an installer program. On Linux, you ran a command. How else can you do it?

But on Classic Mac, an application was just a special kind of file, which you could drag, drop, and manipulate just like any other kind of file.

It didn't even need to be in the Applications folder! You could put it in a subdirectory, you could put it on the Desktop, you could put it wherever you liked, and it would work just the same. I think it's a darn shame that this isn't a guarantee anymore. The other day, I tried putting all of the Microsoft Office applications into /Applications/Office instead of just in /Applications, and the next time the Office updater ran, it installed into /Applications, even though that's not where the apps lived, so now I had two copies of everything. Bleh.

Why is this so much better than running a program or running a command? Well, think about how you would uninstall a program. On Windows, you'd have to run an uninstaller or find the entry in Add/Remove programs; on Linux, you'd have to run a different command. In both cases, you'd have to do something you hadn't done before — you have to know how to do something before you can do it. On Classic Mac OS, you can just delete the application by dragging it to the Trash — something you already know how to do.

Anyway, I'm not defending this approach to installing applications. But I do think it made sense at the time, and the computing landscape changed around it.

4 comments

Spot on. I had a similar thought.

Another major change is that the App Store is now the default way for 99% of users to download and install software. Pre-App Store (say, in the Tiger-Snow Leopard days), Mac users with even limited experience knew the `Mount DMG → Drag to Applications` ritual. At the time, it was one of the two primary ways of installing Mac software, the other being using Package Installer to run .pkg installations.

As a Certified Technical Consultant in those days, I found that almost everyone was able to do the Package Manager method; the DMG method created a lot of chaos for about 50% of people. You'd find people running Firefox or Chrome from the DMG, which they'd leave mounted and visible on their desktops in perpetuity.

As a Windows switcher (in 2006) I always felt that the drag-and-drop model was really elegant, but it was unintuitive because, to your point, I had been exposed to the installation model for so long.

> As a Windows switcher (in 2006) I always felt that the drag-and-drop model was really elegant, but it was unintuitive because, to your point, I had been exposed to the installation model for so long.

https://portableapps.com/ started around that time: builds of open source programs that ran out of a directory without any installation. They were meant for flash drives, but worked just from the desktop or wherever as well.

But you haven't described how to uninstall an app. You described how to remove the program directory, which would be the same as deleting the folder in Program File on windows.

Changes to system config won't be rolled back and other files won't be cleaned up.

That's why on Mac OS you sometimes have to get an uninstaller, which sucks because now you have an inconsistency.

> Changes to system config won't be rolled back and other files won't be cleaned up.

These didn't exist on the original Macs[0], and I would argue that they shouldn't exist on any Desktop OS.

[0] Unless you count Extensions, which were still drag-and-drop but to a special folder under System.

From the UI standpoint proper uninstalling can be tied to deleting the app.

Otherwise a user-level app only used to have a preferences folder and uninstalling an app should not delete the preferences -- maybe you'll reinstall it later. Or at least it should ask for it.

> It didn't even need to be in the Applications folder! You could put it in a subdirectory, you could put it on the Desktop, you could put it wherever you liked, and it would work just the same. I think it's a darn shame that this isn't a guarantee anymore.

It is even worse than that, many applications these days cannot even be installed anywhere other than the default location! At least on Windows you used to be able to do that with a high degree of reliability.

I think the original Mac did it right and everyone else who didn't do the same thing (RiscOS, NeXT kinda) is doing it completely wrong. Which is why I'm a fan of AppImage when I have to use Linux Desktop and "portable" applications when I use Windows. I don't use modern MacOS.

I avoid installers and prefer portable releases 100% of the time. Sometimes, you can extract an installer like a zip and it still works. Other times, I suffer through the install process once and then zip up the resulting folder to get my own portable for next time. I think it's crazy how even the simplest of programs today want to run installers.

I wrote more about it here: https://voussoir.net/writing/master_of_my_domain