The future needs files

For many mobile users, files are like dinosaurs, a holdover from the bygone desktop era. Sure, they “work” but, they’re mostly there because, you know, ancient history. I’ve discussed this issue for the last 2 years and I usually get some version of “get over it grandpa”. 

I’m not here to tell you exactly what should happen, but more what you should want. For me, it’s a travesty that people don’t understand why files are so powerful and more importantly, how they need to evolve for mobile. I want all OSs, including mobile ones, to properly support real files as they are amazing, inspiring, and possibly the future of how we build our digital future.

Note: I’m using iOS as an example throughout this post but Android (and others) are doing nearly the same thing. Please don’t mistake this as some type of attack on Apple, this applies to everyone.

I’m not a luddite

I can understand your skepticism. Am I the dinosaur, overly attached to the past? In my defense, I was on the Apple Newton team in the 90s (even working on an unreleased “Newton Phone” concept) and also managed Google’s Mobile UX team from 2005-2009, I was there when all hell broke loose and saw firsthand how mobile changed everything. Mobile is clearly a juggernaut far bigger than desktop. But too many assume a market win means a perfect product. It’s never that simple. Mobile won for a variety of reasons, but throwing away files wasn’t one of them.

Misconception #1: Mobile already has files!

Whenever I broach this topic on Twitter I always get some smart aleck posting a screenshot of the Apple Files app.  Sigh… Yes, there is a Files app, Bravo… But it’s so poorly integrated into the experience that it creates confusion and extra work. Let’s back up a bit.

In 2007, the iPhone was a radical simplification over the desktop. There were no windows, no menu bar, and there weren’t even visible scroll bars! The iPhone was primarily a content consumption device. This was a brilliant insight. It didn’t rule out content creation, it just made it an edge case. The iPhone was first and foremost focused on browsing and scrolling. In fact, it’s maniacal focus on scrolling introduced “flicking”, which allowed a super fast scroll to the bottom of lists. (there’s a whole blog post I could write  just on the difference between the Newton and iPhone scrolling behaviors)

But the iPhone didn’t stop there, it radically simplified other parts of the UI, the most notable was removing the file system entirely. Remember, this was a consumption device, so files weren’t strictly necessary. You had file-like things, but they were locked up inside the apps themselves. The Notes app is a good example. 

And to be honest, if you have just a few notes, this isn’t bad. The problem is that if you have lots of notes, or want to do anything interesting with these notes (e.g. get comments on them, post to blog, or import previous work) you’re out of luck. My issue with that initial 2007 iPhone was that while it was well intentioned it took things too far. Instead of hiding files away, it killed them off entirely.

But things have improved since then right? There is a Files app after all. Notes can import from Files!

Well, not quite. Let’s just look at the most recent (2021) version of the iOS Notes app. It’s significantly different from the original 2007 version, with lots more functionality, but below is a screenshot of me trying to save a note to Dropbox.

Notes 2021

Notice something? There is no “Save to Files” option! Even more confusing, Notes has its own parallel folders that don’t show up in the Files app. And if you feel like being a smarty-pants and say “Scott, look, those are iCloud folders!” Not so fast there buckeroo. Here’s my iCloud drive:

Web iCloud

Those Notes folders are nowhere to be seen. They are ONLY visible in the Notes app or the iCloud Notes app! A tight little ecosystem you can’t escape from.

To further confuse things, when I took a screenshot of the Notes app and tried to save this to the Files app, that actually was possible! The Mac prided itself on “learn once, use everywhere.” That’s clearly not the case for iOS apps.

Screenshots support Files app

Side Note: You actually can use Files from Notes but it’s hidden. Instead of “Save to Files” you have to chose “Send a copy” menu item that will export a version into Files. So while it’s Notes does indeed support the Files app, it’s unlike others and clearly only focuses in the Import/Export use case.

This odd-man-out approach for Notes shows an underappreciated challenge for any paradigm shift. iOS started off without files so when Apple suddenly added a Files app a decade later, it’s not surprising that most apps didn’t immediately start to use it uniformly. 

Of course, things may improve over time but it’s been years with little change. I worry things are intellectually calcifying, or in Notes case, bifurcating. Part of my motivation in writing this post is to get us fired up about the value of files so it we appreciate this is happening.

Misconception #2: Sharing is all I need

The power of files comes from them being powerful nouns. They are temporary holding blocks that are used as a form of exchange between applications. A range of apps can edit a single file in a single location. On mobile, the primary way to really use files is to “Share” between apps. This demotes files from a powerful abstract noun into a lackluster narrow verb. 

 For example, I can import a text file into the Notes app but it’s really nothing more than a glorified copy/paste, not an editing of an object in place. This makes a cloud storage service like DropBox nearly useless as I’m not editing “the thing” but a copy of the thing. I need to save it back out to Dropbox if I want anyone else to see my changes. That’s vastly underutilizing the power of the abstraction that comes from files. 

By sharing a file into an app you’re effectively making a copy. If I’d like to make a few changes to a photo before posting it, each app I use makes an internal copy of that photo. In order to pass the new photo to another app, I have to export it out, so I get not only a copy of the photo in each app I use, but it’s result needs to be copied out yet again to a service like Dropbox so that I can share it back into the next app.

Of course, people don’t do this type of flow often but that’s because mobile is mostly about consumption not creation. If we want mobile to expand and grow it needs to handle the flows “knowledge workers” do routinely. Part of my frustration in talking about this issue is that people are so trapped within the present. Just because no one needs something today somehow justifies our pain forever. If we’re talking about the future, we need to talk about new tools and new workflows. The current model of files on mobile is drastically restricting this.

Misconception #3: But I can share with iPhone users!

Yes, you can “Share” notes with other iOS users but that’s a very Procrustean Bed you’re making. You have to ask “at what cost”? Are you really willing to bet your creative productivity to a single app from a single company? Remember, this approach prevents your notes even from being used by other iOS apps as well!

The most powerful aspect of files is that they liberate your data. Any app can see it and do something useful to it. DropBox (et. al.) were able to seamlessly merge with desktop usage as it required zero changes to your workflow. Files were just magically synced to the cloud, unlocking not only multiple computers working on the same file but multiple device types. 

The current mobile model does indeed sync your data but through the wrapper of apps which forms a restrictive shield around your data. It’s so much more powerful to sync your data through files.

Misconception #4: Files are just blobs of data

Files are mistakenly conceived as only content, something holding your notes, spreadsheet data, or a photo. But files also have metadata, information about the information. The obvious examples are the file name, creation and modification date. The only one of these that is really used much on mobile is modification date as when you use the ‘file picker’ on mobile, it usually defaults to ‘most recent’ files. This actually does work well, if you’re trying to include something you’ve just created. Unfortunately, it just doesn’t scale much beyond that use case.

A more subtle example of metadata is the folder a file is in. This allows you to group files from different apps, into a single place. If I’m planning a wedding, it’s very helpful to have all wedding things together. This is data first vs app first organization. This was extended when the Mac created the “Desktop”, a temporary holding place for files. People needed folders for longer term storage but it was also powerful to have a temporary ‘working area’ for recent files. The original Mac even had a “Put away” command that would return a file from the Desktop back into its original folder location (sadly removed in OS X). This small bit of history shows how adding a tiny amount of metadata can have a significant positive impact on a user’s workflow.

Mac OS9 Put away command

The same applies to previews or content indexes (e.g. Spotlight on the Mac) This allows the Finder to display your files in more helpful ways and even allows you to quickly find things based on their content. This metadata is hugely powerful and not always available on mobile. 

But it’s helpful to remember that this metadata only went so far. Using “just files” started to break down with apps like iPhoto and iTunes, which tried to have it both ways. Both used the file system to store the many large files needed but they also required an app to add additional metadata to group and sort the content. This created a schism, splitting the metadata between two different masters. This meant you couldn’t just ‘reach into’ your iTunes folders with the Finder to rearrange things (or convert the files from WAV to MP3) without causing serious app confusion. In some cases, if you did this the music would simply disappear from iTunes.

2002 iTunes 2.0.4 CD on OS9 - Take 2 |

There were attempts to fix this. BeOS allowed arbitrary data to be added to files and this was reflected it’s Tracker (file browser) app. This allowed iPhoto-like ‘apps’ to exist entirely within Tracker. WinFS from Microsoft carried this even further with a more robust metadata mechanism. Both were valiant attempts but most people have no idea either existed and have ended up, like Dvorak keyboards, to be considered a mostly ignored branch of computer history. This is too bad as we’ve already seen that things like Mac Spotlight are incredibly helpful. I strongly feel that we should be looking harder at bringing back metadata systems like WinFS/BeOS. But not for me, for the AI.

Our AI Future

My goal isn’t to talk about “fixing mobile”. Mobile will, eventually, get there. Too many people think “Mobile is the Future” but we are so far past that. Mobile is the present. We need to actually be thinking about the future that is coming and what we are going to need.

Mobile started off as a consumption device. That brilliant simplification unlocked an explosion of basic consumption tasks. But if we want to move everyone over to phones and tablets, we clearly have a long ways to go. Yes, there are small niches of people, like writers that are using their iPad for creation. But that isn’t a very high bar, extremely simple devices have existed for this for a long time. Besides, how many companies have successfully migrated their entire company to tablets? I’m sure a few exist but it’s not exactly an avalanche is it?

I’m talking about moving from consumption to creation and not just for today’s tasks, but for the tools we are just starting to use. I’m referring to Machine Learning systems. These are the type of agents that can run through the data on my phone making inferences, corrections, and suggestions that make my life easier and more productive. Things like:

  • Cleaning up my contacts (and searching for additional info on them)
  • Tagging my photos with text inside them
  • Proofreading my writing
  • Indexing and linking “statistically significant” words in audio/video files
  • Creating semantic links between all of my work

These are just baby, brainstorm-ish ideas. We know this will evolve to be so much more nuanced and impactful. Relegating these services to the OS is a safer option, certainly from the security point of view, but that creates an innovation chokepoint. If we’ve learned anything from our history, we need to have more open systems to create an opportunity to try out many many different services. Not just a few more but orders of magnitude more, which is far more than any OS can provide. If we’re happy with Dropbox, we should have no problems with 3rd party ML systems scouring our data, especially if we have folders as a mechanism to gate access.

This isn’t some feeble political statement to liberate my data from a company.  I want files to liberate my data from my own apps and create an ML explosion of activity! Files are at some level a hack, I get that, there are limits but they are an extremely useful and flexible hack. Like the QWERTY keyboard, they are “good enough” for most tasks. Files encapsulate a ‘chunk’ of your work and allow that chunk to be seen, moved, acted on, and accessed by multiple people and more importantly external 3rd party processes. It is a fever dream to think mobile is adequate today. It isn’t adequate and we desperately need the power of files to unlock the future on mobile.

Special thanks to Gordon Brander whose musings on his new app Subconscious revived this 2 year old idea into this blog post. If you’re not reading Gordon, you’re missing out.