Why We’re Building Collections

Cool! Here's what I would really, really like to see as part of a file manager re-imagining: a unified, programmable activity stream for my actions across both my local machine and the Internet, stored locally but accessible via API.

Imagine this:

I save a photo, with Photoshop, to my Pictures/2012 folder. It's immediately added to my local timeline. It's private to me.

I copy the photo to my Pictures/Web folder. Again, this is added to my timeline. I've programmed the timeline to also send the photo to Flickr when I do this.

I copy it to my Pictures/Clients/John Smith folder. It's in my timeline, but here I've programmed it to send it privately to John Smith via a web service.

I'm on a company network. I copy a file to a shared drive. It shows up in the timelines of all the other people who have access to that drive (although some of them may have chosen to filter these actions out). A couple of them have actions programmed in; they automatically copy the file to their own private folders, or to their mobile devices, etc etc. One of them decides to leave a comment on the shared timeline.

I can search my timeline by type, person, action type, etc etc. And do the same even when I'm disconnected from my computer and network, although not all of the files are necessarily available.

I'm squinting at the pictures and wondering. How on earth is this thing going to handle real projects? You know, where you have like twenty different kinds of files all related to the same project, which should be grouped together because they're about the same thing, even though they're in different apps.

I mean, here's what I find in the directory for my current comics project:

  a bunch of Illustrator files (this is what I draw it in)
  a few CBZs of the content
  some PDFs of the same content
  a bunch of web-res gifs of the pages
  some Indesign files (related to publishing the book)
  a couple .csv files used in an Indesign data merge
  various other gifs/jpgs/tiffs
  a Word doc
  a link to an external directory full of print-res TIFFs

Plus a subdirectory structure to chunk all of this into stuff related to book production, model sheets, fan art, and whatnot.

This is the use case where every "We're going to replace and simplify the file browser!" effort seems to fall down. Recognizing that real people's projects sprawl across many file formats.

Seems more like Adobe Bridge than Finder.

http://www.adobe.com/products/bridge.html

It really bothers me when I click a download button only to be greeted by "Hey, you can't actually download our app, but why not sign up for our newsletter?"

Just build a new Norton Commander for OSX and we are happy. Anyway, don't get why YC/PG invests in such ideas. Though nice this idea is no venture case (easy to copy when successful).

I was a boy when I first saw that icon of a hard disk on a Macintosh System 6 desktop. The moment I saw it, I knew, instantly, EXACTLY what it implied and felt a deep sense of satisfaction that the person who came up with that (ostensibly a PARC person, not an Apple person, of course) was indeed a poet, probably someone who could gaze at a Magritte painting for an hour, just enjoying it.

The file-system-as-a-tree was and remains a powerful and useful abstraction ... that very few people are aware of. Not morons: my wife is brilliant, but when she saves that complex XCell doc with all its pivot tables, formulas and summations, she still appears to have no clue what happened to it.

And since the rest of the engineering community has also decided that the file system tree is "too hard" for people to understand, they are doing away with it on tablets now too: ever save or download something on your ipad (or apad for that matter) and struggle to find it?

So if mere mortals cannot understand where a file went on the file system tree, I sincerely feel like these poor guys writing Collections are handing a machine gun to a cave man by which they will be clubbed to death with.

A sophisticated solution to a putative problem of non-sophisticated users that only sophisticated users would use.

Is this not Nautilus/Eazel all over again?

http://en.wikipedia.org/wiki/Eazel

I think you're hanging out in a reality distortion field where people actually do try and use Dropbox, Drive, iCloud, etc. I get it - all my friends use them and talk about them and stuff. But my friends are tech nerds. Normal people grow their file system organically. They don't use dropbox - they email stuff to themselves. They are disorganized, and if they ever tried to get organized, they'd just end up forgetting where stuff is.

This thing looks like it's going to add a whole bunch of work to my life while I try to figure out where the hell stuff is.

How is this app going to make money?

This product (might someday) solve a real problem I have: one significant impediment to picking up a new web app is the need to learn yet another way of interfacing with the content I "own" on it.

If Collections can provide a compelling consistent interface on top of existing web apps, and an API for new web apps to target, then they might get to own some valuable conceptual real estate in user's minds: "I'm willing to try this new service because I already know how I'll be able to manage my content on it".

This also makes the world a better place by making it easier on the newcomers--it's unfair to them that established players occupy the "I know how to use this already" space in users' heads, that their service has to not just be better, but be that much better than the established players, to reach people.

Then they could use that conceptual real estate to promote those new web apps, and that promotion could yield revenue. Another option would be to wrap their own reference implementations of the services they're abstracting over in a premium layer. Another would be to work with web apps to provide value-added interfaces to the web apps' premium services and take a cut of whatever the web apps charge. Another would be to offer a premium corporate version that plugs into internal corporate datastores (in a way that, presumably, doesn't suck, distinguishing them from other products).

Collection's play for native integration (e.g. extending that consistent interface over all your local content) distinguishes them from Dropbox, which prefers to own that content.

A similar problem to this, that Collections isn't targeting (yet), is to provide an abstraction not over data but over operations. It's already far too complicated to juggle email, Twitter, Facebook, Instagram, Gmail, G+, SMS, tumblr, etc. That means that I have to really think hard about letting a new web app into my routine.

An intermediary that presents a consistent interface to all those services, and opens itself up for use with new services, could try to win over valuable conceptual space with users and make the world a better place in exactly the same way.

It would be cool if your slogan "Finder for the cloud" was replaced with "Explorer for the cloud" (or whatever windows users call it) if they are on windows.

You could detect this pretty easily and it would prevent alienating a large percentage of visitors.

I have absolutely no problem with the finder and would not like a file manager that links to my social websites. I would mind a separate app that would aggregate that info, but wouldnt want use it to manage my local files

I was literally thinking about how much I needed this five minutes ago, but I want it for Flickr! Any chance you'll be adding support for other services in the future, or adding a plugin model so I could do it for myself?

This looks amazing, but it sounds too good to be true.

This is the quint-essential problem of our current time, for techies and non-techies alike. Media overload.

No easy way to organize everything and find it.

Whenever I take pics with my iPod and put them on my PC, I don't take the time to rename them from DSC_001.JPG to something useful. Just drop them in a folder that is named appropriately and am off.

It is suboptimal. So if you guys can figure out how to fix that, there is much potential with that.

But....iono...I am a bit skeptical, would love to be proven wrong though :)

Just my 2cents: have you read About Face ( http://www.amazon.fr/About-Face-Essentials-Interaction-Desig... ) In this book they explain a whole vision about computer things (memory, files, interfaces) that Apple is slowly implementing in tablets and Mac OS X (extending it with iCloud). I'm always pestering people at work with: "have you checked in the green book before doing this?"

Maybe I'm not representative, but I live in zsh on mac and linux/freebsd.

What I'd really like is a way to specify files within zsh, then open a new finder window with just those files in it, without moving them. Right now, I can do "open ." and get a Finder window with the current working directory, but what I want is something like "openw [1-5]*/" and get a new finder window containing just things matching that. They could easily be in multiple directories too.

Yay that's exactly what we need to make file management more coherent - another baroque piece of crap that only adds features instead of refactoring existing ones.

uh... it was actually not always this bad.

The problem is that we have opted too much for the visual metaphor for a desktop/folder etc...

Back in the day (~1987?) there was a file manager called "PathMinder" -- this was a GUI file manager for DOS -- it was AMAZING!

You had a full keyboard navigable system for your file system... now, this was a directory centric model (the idea of collections/groups/tags/etc had not evolved yet) -- but hte fluidity of navigating the structure was AMAZING.

I had fully memorized many many paths (just like memorizing the path to various dungeons in Bard Tale) via the keyboard.

It would be great to have the ability to have a meta-tree of navigation.

E.G.: assume you navigated to \Downloads - you could then highlight a file, say, a .PDF -- then in one more stroke, show me ALL .PDFs -- or what if I had tags on my files - even AUTO-TAGS (like Received Via [work/personal] email) - show me all PDFs I have received via my work email. or Show me all attachments received from my mom.

Collections of content is a weak premise... I would rather auto-tag content based on how I receive/procure it...

It is more about threading the communication channels than it is putting shit in the same bucket...

I find it interesting that this was in the same class as Filepicker.io

Is cloud-based file selection something that should be offered by a web developer, or by the OS?

Good software idea as you're going after a market that existing cloud vendors cannot (in terms of interoperability) in fear that it might cannibalize their own market share. In the end though, I can't see this being much of a consumer app, but rather a power user app. In the consumer world, I see whatever file manager and eventual cloud system that's baked into the OS ruling.

Windows 8 starts to head along this path with cloud-based file selection from within 'Metro' style apps. For example you can add a picture from Facebook or Flickr to Onenote directly from the file picker in Onenote.

It'd be nice if they did what Collections proposes and extended this by allowing apps to natively integrate with Explorer and not only the Metro file picker.

Interestingly enough, one of the first YC companies from Summer '05, Memamp, tried to solve the Desktop search problem with replacing Finder.

I remember watching their demonstration during demo day and and being very impressed. Now the problem space is even harder, with cloud and hosted solutions in addition to all the desktop files.

I'm always open to Finder replacements. I use TotalFinder to add tabs, but there's definitely room for improvement.

I think it sounds promising. As much as my habits are solidly in the "file explorer" space, I can see its limitations.

But, in order to even convince me there's a viable alternative, the user interface of something like Collections has to really knock it out of the park for me. So much so, it seems daunting.

This could be great. I have endlessly thought of how obsolete the current UI is on the system trees are. Really wish the best for you guys, of course signed up!

This is important in order to track all of my personal assets that reside (somewhere) on the web.

Looking forward to a good product - don't dissapoint! ;)

I think the whole concept of storing files in folders and folder hierarchies is outdated. Why don't we use tagging for our files yet?

Interesting idea. Looks promising. Good luck!

As much as I usually love native apps, why can't we just have a web version of this? A new type of "web portal."

Don't forget to include links from pinboard.in as this is an important part of personal knowledge

An interesting take. Collections seems to be a sort of compromise in the native vs cloud debate.

If this is making local copies and syncing them like dropbox it could maybe be awesome but is likely read only. If it doesn't work like that then it will be terrible and constantly be loading, imagine trying to open a big image collection if it is making the requests right then.

However if it is dropbox style it might use a lot of people's hds.

Is this Midnight Commander 2012?

Is the new Ubuntu / Unity Dash somewhat like this (without as many services)?

How will this handle music?

"Request an invite => "Subscribe to the list"

Something's wrong here.

This looks neat. Looking forward to trying it out!

Pet Peeve:

The big friendly button labeled "Get the App"... doesn't.

I think this is a good premise - the current form of file management is definitely outdated and not nearly as functional as it could be. It seems like you guys want to do for every storage/sharing service what Dropbox does for Finder/Explorer by making management as simple and as transparent as a regular directory on your filesystem. Here are my thoughts on this: first, I think you (or the users) are going to run into a problem eventually of redundant data. If I am sharing my photos across 3 or 4 different services, how do I keep track of the individual files? If I make a change to one, do the changes get propagated across the other services? Or do I have to do manual copies? What if I upload pictures from a mobile app or a web interface, and then upload the same files to another service? From the screenshot it seems as though you divide up functionality by individual service, which makes sens, but what if you included a 'Pictures' or 'Music' or 'Whatever' view, much like smart folders work in Finder now? And like smart folders, these views would be customizable. Would you have this integrated into spotlight or roll your own indexing feature? A good search is going to be essential for this kind of app. I like benwerds idea of a programmable activity stream, but in order to reach a general audience with this app that kind of functionality would have to be built in. I think its safe to assume that most users aren't interested in extensive customizability and personal scripting, so even though a truly programmable file manager (possibly to replace both the command line and the file manager?) is a wonderful idea for technical users and I think something like that is in our future, but I feel like that is beyond the scope and focus of this app.

Here are a couple of specific things I would like to see:

- transparent music management. If I am browsing my music 'collection', I would like to have them appear side-by-side with my favorited songs on soundcloud, or my amazon cloud drive or spotify or youtube. When I click on the song, it will stream it from whatever source it belongs to, but will all use the same play/pause/scrub controls. This idea could also be applied to videos, or even photos and text. However of course this would mean creating your own viewers/players (so maybe not something for a 1.0) but I think a feature like that would be necessary to create the total "transparent cloud" experience, because otherwise the task of music listening becomes a chore of managing different apps, web and otherwise. "Buy/Download" buttons next to each remote song would be cool, and once bought, all that would happen is the button would disappear and the song would show up as being on your local (or on your cloud drive or whatever). It would still maintain the same position on your playlist. Now that I look at the screenshot more, it looks like you're already doing some of this with pictures.

- Dual/split panes, a la commander. Please. It makes life so much better.

- Command line support. It doesn't need to be scriptable, but I think somehow exposing your objects to the filesystem like dropbox does would be a good solution, even though you may not be able to do it like they do by storing the files on both the local machine and the server, since I feel that would defeat the purpose of using cloud backup services. Which objects to expose and how might be a tricky question for some services (like twitter), but I just want to be able to cd around and cp files to and fro, so for services that just store/share media, this would be pretty straightforward.

- Hotkey show/hide. I love TotalFinder (http://totalfinder.binaryage.com/) and TotalTerminal for this reason. It makes general maneuvering about on my box so fluid (thats what she said), and its become a dealbreaker enough for me that I've passed up a number of "better Finders" for this reason. It doesn't have to slide in from anywhere, but I just want it to show up when I press something and hide when I press something.

More than I expected to write; I guess I've been thinking a lot about this kind of app.

This post blew up quite fast. 31 votes and 6 comments in ~30 minutes?

Go go Collections!

Emptiness.