Twitter developer Alex Payne recently criticized apps like ShoveBox, lumping them into a category he calls “everything buckets”:
The Mac software ecosystem faces a plague. A plague of Everything Buckets… These applications claim to be “your outboard brain” or “your digital filing cabinet” or similar. They go by many names: Yojimbo, Together, ShoveBox, Evernote, DEVONthink. …
An Everything Bucket, since you‘re probably wondering, is what I call applications that encourage the user to throw anything and everything into them. They‘re virtual scrapbooks, applying a lightweight organization system to (often) unrelated data of varying types.
He goes on to make essentially two arguments that get muddled a bit:
- Computers work best with structured data.
- Use the filesystem, you dolts!
My counter-argument is:
- Structured data is not always best for users. Software should be designed around users first, and computers second.
- While, ideally, it would be best if programs manipulated data closer to the file/directory level, there are some deep problems, some technical and some psychological, with the way the Mac OS and its applications work with the filesystem.
Structured Data Isn’t Always Best
Structured data is great when you know what structure your data needs to take. And even when you do know, the trouble is there’s an up-front cost to capturing it that isn’t always amortized over the rest of the life of the data.
What I like about ShoveBox and even my competitors’ apps is that I can quickly enter in some data that’s nothing more than a thought, an AIM handle, an interesting article I plan to read later, or even print a receipt to it.
Then, at the end of the day, I can quickly drag these bits and pieces into todo list programs, address books, web apps, and more. I can take the time to massage the data a bit so it fits into other systems. Or I can delete it and decide it wasn’t worth it.
Being a east coast guy myself, I’m convinced that the "Worse Is Better" philosophy of engineering is right on a lot of things.
Does anyone remember when search engines used to look like this?

Google didn’t become popular because it had a powerful interface for nested Boolean queries and let you specify where to look for your query within each document. It became popular because of a very simple, unstructured approach.
It ended up being great for users, but it certainly wasn’t the easiest thing for a computer. It required vast infrastructure, not to mention some incredible linear algebra. Users don’t care what’s easiest for computers; users care about what’s easiest for users.
And though it makes geeks cringe, Excel is probably the most popular database, despite not being a database at all. It’s simple and you can whip together solutions around loosely-structured data in an ad hoc, generative fashion. Huge departments in multinational banks rely on Excel rather than custom solutions, just because it’s the easiest to start using.
The idea of rich structured data is very appealing to engineers like Alex and I. It’s nice to think of a world where every one of my documents has rich (correct) metadata, where I can strew my data all over my disk (and cloud-based sites) and still pinpoint them in seconds.1
But the fact is, most people don’t have time for rich metadata and consistent, taxonomically-satisfying structure. They just want to get their work done.
And even as we find better ways of entering and using structured data, there will always be classes of data that are best left unstructured. That ill-defined category of data is what ShoveBox was made for.
The Trouble With The Filesystem and the Windowsization of the Mac OS
Before I continue, I should emphatically state that ShoveBox is not meant as a replacement for the Finder. It’s not a file-management tool. It imports data in a great many ways, and the only reason you can drag certain types of files into it is that, well, it’d be a shame if you couldn’t.
The term “Everything Bucket” is a slight misnomer — if you’re putting truly everything you work with in it, and keeping it there, you’re not using it correctly.
But from listening to developers of ShoveBox-like apps, I can tell we all get inundated with requests from users who are misusing our apps. For instance, the Yojimbo release notes pleads “Please do not try to replace iPhoto, Aperture, or Lightroom with Yojimbo: You’ll make us cry.”
It seems to me that that the problem that Mac users and fellow developers of “everything buckets” are groping around in the dark to solve is that, out of the box, the Mac OS’s tools for managing data are unsatisfying and confusing. But to be fair, I think most novice users find the very idea of a filesystem unsatisfying and confusing.
But It Wasn’t Always

When the Mac came out, it didn’t have directories, it had folders (and, originally, only one level of them!). It didn’t have rm, it had a Trash. It didn’t have symlinks, and when it finally did in System 7, it called them aliases. It didn’t have a tools menu, it had desk accessories. And everything had an adorable Susan Kare-designed icon for it.
The innovation here was not the desktop metaphor, but its implementation. Everything was concrete, visual, and almost tactile. Windows stayed where you put them when you closed them. Icons didn’t sort themselves. One window, one folder. One file, one icon. And there were almost no modes.
Windows has never optimized its folder structure very well for users, and tries to discourage direct manipulation of the file system. Programs are free to dump files in any location they please, even to the extent where many programs on Windows won’t run on a non-Administrator user account.
In fact, in recent versions of Windows, when you first open up your C:/ drive, it warns you before even allowing you to see the contents. As if to say “What are you doing, Dave?” (though, in its defense, it does say “My Computer” not “Your Computer.”)
The Mac has remained fairly resolute in making the filesystem and its folder structure something users feel comfortable taking a stroll around. Even files nestled deep in /System/Library have beautiful, seldom-seen icons to convey a little of what they’re actually for. And thank goodness, drives still mount on the Desktop and don’t warn me when I try to open them. But it’s definitely yielded to external pressures.
I’ve never been one of those whiners saying that the Finder is broken, but I think it plays a large part. Some have complained that it’s moving away from its spatial roots.
Back/forward buttons, column view, Smart Folders, Cover Flow, QuickLook, Stacks, and others are all very handy features that have clear uses. But they dilute the wonderful concreteness of the Desktop metaphor with abstraction.
Oh — and when I open my Documents folder, the number of files that I remember putting there myself are outnumbered by those that I don’t. Microsoft Office put a 150 MB folder in there whose purpose I’m not sure of. Any EA game I install puts one or more folders in there as well. And iChat, Colloquy, and Linkinus seem to find it a good place to dump log files.
I don’t always feel comfortable using it directly2, and I’m a computer nerd, for crying out loud!
There’s something comforting and pleasing in using an app like ShoveBox or Yojimbo for everyday files, even though it’s not a terribly good idea.
For a lot of users, it’s kind of like living with some inconsiderate housemates, finding a shelf in the kitchen, and agreeing “If it’s on this shelf, it’s mine. Don’t touch it.” It’s a secure place with a well-thought-out, consistent, streamlined interface. And best of all, Microsoft Office won’t swagger in at 4AM and mistake it for a sink.
The Awkward End of the Desktop Metaphor
We’ve seen consumer-oriented apps from Apple that abstract away the filesystem entirely. iTunes doesn’t play MP3s, it plays music. iPhoto doesn’t organize JPEGs, it organizes photos. Part of the reason for this is that they are not presumptuous enough to assume basic computer skills on the part of the user. The other reason is that tying it closer to the filesystem would make their interfaces less apt for what they’re supposed to be doing.
Spotlight is firmly in the camp of "one file, one datum" and Apple apps that disagree with this have either been forcibly converted to this philosophy or provided a Spotlight interface through a rather inelegant hack (take a look in your ~/Library/Caches/Metadata folder).
We’re in an awkward phase in user interface design where the desktop metaphor is dying but there is no clear successor. More abstract ways to deal with data have been gradually introduced, but we’re still in a sort of limbo.
Do we return to spatial ways of dealing with files, or are the abstract ones better?3 Do we try to keep data structured, or do we free it? Should the filesystem be more like a database to the end user, or should it be left as is? Is flat better than hierarchical?
I do feel odd writing an additional layer of indirection on top of what seems, superficially to be a perfectly fine metaphor. But it wouldn’t be necessary if “pieces of paper in manila file folders” was the best way to think about everything.
Maybe some things are best as messages in an inbox, others points on a map, others monsters in a dungeon, some as crazy zooming squares, and still others as rows in a database. The future will tell.
Alex paints these apps together with a sort of broad brush, but they’re quite different. Developers of apps like ShoveBox, Evernote, and Yojimbo recognize a major problem that many users have dealing with certain kinds of data. We’re all contributing ideas to a vast, distributed, multi-developer experiment on the Mac platform.
Even the lovely Together, probably the closest to the Finder of any of these apps, has a tremendous following that I envy.
ShoveBox’s contribution to the experiment is trying to apply the email client metaphor (inboxes, flags, sorting rules, etc) to unstructured data. And, if I may say so myself, I think it’s the most streamlined app in the category.
But judging by the vast, diverse preponderance of “everything buckets”, there’s a real need here. We’re all just throwing stuff at the wall to see what sticks, and learning more about how to work with computers more productively in the process.
Footnotes
- As an aside, that isn’t to say we’re not getting closer to that reality — one positive example of structured data succeeding is Facebook. It astounds me that the average college kid, still hungover from the night before, will tag their photos with metadata about who’s in them. This, along with other aspects of Facebook, strikes me as a huge win for structured data.
- I’ve carved out a few sub-folders in my Documents folder that I keep in my Finder sidebar: Projects, Schoolwork, and Resources.
- When I compare abstract versus concrete, I’m not talking about having an abstraction versus not having one. Clearly, any user interface for dealing with data is a way to take magnetic particles on spinning platters and make them more understandable. When I say concrete, I’m referring to interfaces that mimic the physical world versus the mental world.