New Thread: Why is tagging a "must have" these days?

From another thread:

It seems that everybody and their brother is clamoring for the ultimate in tagging as a feature of DTPO, and I’m wondering why. I see way to many posts here nagging the DTPO team about including tags or they (the poster) will never buy DTPO.

I’m an old fart and perhaps beyond learning new tricks but I’m also a curious old fart. If I recall, tagging seemed to pop up out of nowhere into the mainstream of computer application use over the past three years or so. Being curious I tried a few programs that permitted me to tag various files in different ways. I tried the tagging feature of specific apps like Leap, which also produced a tag cloud. I started tagging my images within iPhoto.

After a few fits and starts I decided that tagging was simply more trouble than it was worth (for me). I was spending hours calling up files and images and graphics and stuff, trying to think of appropriate tags for each one so I could easily find what I wanted in the future.

The trouble was that I quickly forgot what tags were in use and so when I went searching on what I thought was a tag, nothing ever showed up. When I simply searched on filenames or filecontents I usually found what I was looking for, quite possibly because I create file names and something in the file contents that I do remember in context. That, to me, seemed a lot more productive and less consumptive of time than going through hundreds of thousands of files, images, whatnot and trying to append tags thereto.

So I’m throwing out the question to all you tag “addicts” out there: Why precisely does tagging make or break a deal for you? Is there no other way to organize your thoughts and files? For me DTPO offers so much in the way of AI and organizational capabilities that I wonder just how tagging will improve DTPO so much that it becomes a “must have” application. For me, it’s already a “must have” and I couldn’t live without it simply because it works and works well.

Positive remarks and responses welcome!.


I think you basically answered your own question :wink: you were using the file names as tags :smiley:

The real difference for me is the difference between find information by filing it into an hierarchy and finding by searching (where tags or file names makes it easier for the search to succeed). Both these methods can be combined and the both has their advantages and disadvantages. Personally I use both

Hi Tod-

Congratulations for calling out the whiners on the forum, and their impolite badgering of the Devon engineers. I think the interest in tags has to do with a reluctance to really deal with organization on some peoples’ part. Certainly can create a folksonomy with millions of users contributions, but on a single user computer? Not the same thing, sadly.

That said, tags are an important organizational tool. A lot of what I deal with shades between philosophy and science, for example. Or an author like Melanie Mitchell, who has written papers on both evolutionary computing and cellular automata. I’d like to keep her papers together, but it’s also nice to see them in both categories when I’m browsing.

Best, Charles

Clearly, tagging is not a new concept. “Tag” is just another name for keyword, and keywords have been used to classify texts for ages. In libraries for example. In our university library - years before computers could be found in German libraries - we had two kinds of index card catalogues: A catalogue by author and a catalogue by keyword. Tags are a way to organize information - not a way to avoid organization. For some people hierarchical folders are a limiting way for organizing information. Depending on the kind of text you work with, they often belong in more than one folder. Devonthink offers replicants as a solution. But as I said in another post, this can be very time consuming. With the combination of tags - and the search for those combinations you can create folders on the fly - which makes tagging very flexible. Why would someone need tags when he or she can use a full-text search? Often the topic of a text is not mentioned. You have to add it in some way. It might be a special text about a certain author of the eighteenth century in England. But this might not be mentioned in the text because the author assumes that everybody knows it. It is very convenient to tag this text with “literature”, “England”, “18th Century”. When you work with more than one language you have to add connections to texts. Doing this by tags is very useful. Like all methods to organize information, tags have their limitations. They always represent your point of view and research interest at a given point in time. But they are not completely useless because of their limitations. Just like folders are not completelely useless because of their limitations. I would really love to be able to use the tag view in Devonthink and I find it hard to understand why it seems to come last. A lot of people have expressed their disappointment - but this might qualify as whining and impoliteness. I don’t see it this way.


Charles, like you I deal often with bridges between philosophy and science, as well as concepts shared among different branches of science.

A difficulty in tagging for such relationships is a priori recognition of them as content is added, which is further complicated by the use of different terminology in the various disciplines. But one of my greatest joys in exploring my references is discovering important concepts shared by disparate disciplines.

In the sciences, such concepts are often quantitative relationships. Often, See Also can help me see such relationships. One of the examples I frequently use was See Also’s suggestion, when I was viewing a paper about how invasive species changed the population dynamics in an ecosystem, of a paper about factors that influence chemical reaction equilibria. Yes!

Calhoun’s classic study of the fluctuations of field mice population on a plot of land (and related predator populations) as related to fluctuations in food resources is a similar example. While viewing that paper, See Also will suggest some papers in economics and sociology as well as historical cases of diseases, epidemics and famines affecting human populations.

40 or 50 years ago political scientists jumped on the bandwagon of quantitative methodologies with enthusiasm (they changed the departmental name from ‘government’ to ‘political science’). Some of the resulting work was productive; some was just silly. I teased a political science friend about having been influenced by “Tom Swift and His Electric Factor Analysis Machine” (I made up that one). See Also suggests that there is similarity between use of polling data and my references on sampling and evaluation of environmental data. Of course! And in both fields, uncertainties often loom large and public opinion and politics play major roles.

I’m continually discovering new relationships between ideas and bits of information. I don’t use keywords, unless I’m cataloging things. I do make searchable notes when I find something interesting, with links to examples. I rarely spend much time on organization, except when I’m working on a project and that kind of work becomes helpful. Especially for work at that stage, I’m also looking forward to the DEVONthink tagging scheme – but I will never, ever use it to tag everything.

Well, as a father several times over when my kids keep asking for something that I can’t afford and I keep telling them no and try to explain why not, then it degrades into whining and badgering. My kids didn’t see it that way either but it’s the truth.

The DT folks have heard all the requests for tagging/keywording and have been very patient here on the forums when people keep making the same request, and Christian or Eric reply with basically the same answer. I for one am tired of this broken record and would like to see discussions move forward.

I try not to bicker and whine too much.

Everyone tags with metadata. Simply placing a document into a hierarchical group of folders is tagging data.

What many of us tagging fiends want is to dissolve the “physicality” of the computer folder for something that is even more virtual and fluid. I spend too much time managing my hierachical folders and fretting over if I should group data by datatype or should I group them in a project manner.

It then dawns on me that the folder structure is merely a metaphor carried over from an potentially inefficient filing cabinet system in the real world. By limiting my thinking to this paradigm (managing folders) I’ve failed to grasp what a computer excels at. Presenting data in a vast amount of way in order to more effectively communicate.

So tagging comes with my brain input as to what tags are relevant to my personal needs and what tags are not. I tend to go with my first intuition about tags because that’s what is going to come up most often.

I can no longer hope that I can pull the right search string in order to find the document I want. I would rather tag my files and create smart folders that watch for certain tags. This way I can whimsically change how and what data resides in the folder without really changing a directory. I’m simply changing my view of my entire dataset.

When done right it’s powerful and liberating and hopefully posterity will master it and thus master the intelligence lurking in our data.

Murch - thanks for a good example. It was a good read and provided a lot of background info that I hadn’t considered in the past.

I’m sorry, but in my opinion your remark is just impolite. Tagging is nothing that has not been promised or explained as being impossible. It is announced as a feature of 2.0. Perhaps I have missed lots of posts on this topic. But so far the only answer to the questions about tagging I have read is that it will be part of 2.0 and that it is related to groups. I have just asked for further descriptions of the tagging interface. I don’t see any broken rekord here. Okay, questions like this are regarded as impolite on this forum and should not be asked. People asking them are treated like misbehaving little children. What is this forum about, then?


I’d just like to jump in and quickly show my love of tags:

1 - first, very happy that DT is including tags/keywords as an organisational tool, I have built my own tag-system in DTP for now, looking forward to have it done properly :slight_smile:

2 - Tag-organisation and folder-organisation are two very excellent ways of achieving both the same, but also different, goals. They complement each other beautifully. And they both really suck for certain tasks. I see the value in both. In particular with data that can be indexed in multiple ways, tags are really neat.

3 - I think most people have been using “tags” for years without ever realizing it (or caring). And they don’t have to realize it either.

3 - I think tags (or keywords, whatever one should call it) would be a much bigger success, more established, and more people would “get” them, if they were adapted and provided at an operating system level, just like the file-folder system is adapted at that level

@Superballoon: Thanks for some interesting and cogent points. I especially agree with what you’re saying in points 3 and 4 above. With responses like yours I’m gaining a better sense of what tags are as well as how they’ve evolved from “keywords” and other earlier organizational tools.


Yup, tags can dissolve the location-dependent boundaries of folders.

Uh, huh. From sjk » 29 Dec 2008 11:07:

… am ultimately more interested in alternatives to traditional file management that’s built around last century’s still-stubbornly-dominant desktop computing metaphor.

… which links to sjk » 21 Oct 2008 11:56, where I uttered:

The file/folder/desktop metaphor is so pervasive and taken for granted that it’s hard for many people to consider viable alternatives (excluding obviously irrelevant computing contexts).

There’s plenty of related discussion in that long thread.

Cherry picking a few of my numerous comments from other times and places, in no particular order:

Being able to easily create/modify/destroy different virtual views that aren’t forced to mimic the filesystem hierarchy is one way I can imagine organizing and managing all my “files”.

One limitation of the desktop/file/folder metaphor and content management UIs is being too tightly coupled to the hierarchical filesystem.

While there’s definitely room for improvement I prefer iPhoto (and iTunes) handling the backend storage details automatically and transparently, which relieves me from tedious hierarchical file/folder management of the actual data.

The traditional style of “file management” is a rigid, tedious, location-dependent decision making process. Do create/copy/move/delete operations really need to be so strictly tied to the hierarchical storage of the filesystem? I’d rather not be forced to rely on that structure in order to organize and create different relationships between data objects.

Hierarchical filesystems enforce an unavoidable, ongoing “where does this belong?” decision making process. Manual tagging can add a “… so it’s easier to find later” component (among other things), but also adds another decision without removing the folder location dependency. The combination of choosing a filename, folder, and tags might be confusing and overwhelming for a lot of people, possibly lessened if tags were true file/folder peers.

I’ve mentioned Gmail as an example of making the folderless/tagging metaphor more familiar to the mainstream by replacing mailboxes with simple labels. And playlists/albums in iTunes/iPhoto can be considered virtual folders, with the other folders (poorly chosen name, IMO) later added to those apps adding hierarchy and location if desired. Even the latter folders can be seen as a virtual view of the main unhierarchical Library view (which itself is a virtual view of an underlying file/folder storage hierarchy).

Ad nauseam. :slight_smile:

Wrapping this up with a remark from - WARNING: A long, rambly exploration of the state…:

Every geek I know shares, to some degree, the notion that the “desktop” metaphor for computers is outdated.

For me tagging would be useful for non-content-rich material. For example, I’ve been using DTP to take some notes as I watch various series on Sometimes a screenshot is worth a thousand notes – however to place that screenshot in DTP to find it again, I’d need to either embed it in a .rtf document with some typed text, or create a folder for it and add a text file to that folder describing it.
And therefore I have screenshots still on my desktop, unfiled, waiting for me to remember do jump through some hoops and deal with them.

filtering speed is the main reason why I like to work with tags. At the moment, tagging can be simulated by “replicating” the file into a different folder, or cmd-option-dragging the file into that folder.

In the free reference manager ( for example , tagging is possible via drag-and drop. It is also possible to change the name of the tag for all items, which is useful if the concept changes. I am looking forward to see something like this implemented in DT, although I am much more eager to see fuzzy search working again.

I think I’ve mentioned it elsewhere, but technically, tagging and folders is the same thing. Metaphorically, they are different (sticking a tag onto, versus putting something into a folder). To the comptuer, it doesn’t matter, they can be marked with the same kind of metadata. They can be made to differ, by allowing only one location for an item, but several tags for it, but as we know, devonthing does not do this. Tags can be hierarchical just like folders.

In use, there is a world of difference between tags and folders. Tags show “where else” or “what” something is, by showing all the tags. Tagging is usually more convenient (if done right) than dragging things into folders. Also, by having both tags and folders, once can separate “where” and “what”.

Sometimes it can also be useful to switch views, to show the tags as folders, or the other way around. One of my favourite qualitative data analysis tools, Bookends, can both display a hiearchy of categories (folders) and when looking at an item it shows all categories used for that item (tags).

Thus, “tags” and “folders” are just different views. Both are useful to have, and often useful to have at the same time. (No, I’m not just speculating, I’ve built a system that can switch views, some years back, just to make sure that I’m right about this… and a few other things…)

Like Christian recently said:

And basically tags will be groups and vice versa.

Nothing new there then. Also Interaction design needs to be really excellent for it to become two distinctly different and yet very useful things. Maybe that’s been mentioned too already? (Interaction design has so far been ok in DtPro, but it has not amazed me yet. Maybe this implementation will be it.)

I am not a whiner and am waiting for tagging ability because I need it and want it. I think tagging is a basic tool that should be available in a program like DT. They are a quick way to designate a file according to type for future reference. That’s all. No need for protracted arguments or to be told how I’m doing it all wrong, lazy, unorganized, how to do it a different way, etc. All other ways I’ve tried replacing the tagging function in DT are cumbersome. Tags work for me and other programs offer tagging. I want to be able to use DT for all info warehousing and tags are simply another quick and easy way to do this. I look forward to them being available in DT soon, and find it irritating that they haven’t been and that it is taking so long. I have been waiting in silence, being irritated in silence, and have tried other programs because DT didn’t have this functionality. But I like DT, have been using it for years, and still like DT best for information storage.

So yes, please, DT folks, add tagging support quickly. Thank you.


and some random thoughts of mine…

  • as mentioned above by many forum members, tagging is actually managing a flat structure of folders with a different UI approach;

you don’t put your documents to folders, you assign these ‘folders’=tags by entering their names; this makes your workflow a bit different, it may be easier to simply enter some names (and even more easier if there’s an autocomplete feature for tag names available) instead of using drag’n’drop

but that’s actually the same like having a text field where you can enter your folder names; you would enter these names, DTP would help you by autocompleting what you type - this would be faster than using drag’n’drop in most cases, especially when putting a document into multiple folders

  • are tags easier for search issues? don’t think so; I had the same problem using tags in Evernote (‘how did I call this tag?’) like I may have now with DEVONthink (‘what was the name of this folder?’)

  • there are some advantages of the folder-based approach: the most important for me is the one with using duplicate names for many folders - I often use the same folder name for several folders with different parents;

example: ‘Hobby/Books’ folder where I collect the books to buy/read and a ‘Shopping/August/Books’, ‘Shopping/September/Books’ etc. folders for grouping the books I’m planning to buy each month - how would you solve that with tags?

  • I can imagine there are some users who’ll enjoy working with a flat structure of tags; but I do need hierarchies - and I’m not the only one - have a look at Evernote tagging, they introduced tag hierarchies which makes this feature similar to folder hierarchies - BUT with a single, fatal exception - you can’t have multiple tags with the same name in different places of your tag hierarchy - which destroys my concept (see previous point)

So, to summarize:

  • limiting the information management software to tagging ONLY makes no sense - it’s NOT THAT flexible as people think, see Evernote and my examples

  • tags are nicer as far as the UI usage is concerned; assigning tags by entering names is usually faster than drag’n’drop to folders although the (auto-)classify feature of DTP makes this issue less problematic!

  • I’d love to see tags support at DTP but I would use in a quite personal way - as a way of managing meta-metadata (like DTP labels but with no count limit)

@bartekb81 I’ve enjoyed reading your posts. Your enthusiasm is infectious!

I’m with you in that I think in hierarchies, too. But to solve this problem, you need to step back and think “flat.” For example, if I planned to buy a book entitled “Flora & Fauna of the Devonian Period”, in August, I would tag it “Hobby”, “Books”, “Shopping”, and “August” (After my purchase, I could change the tag “Shopping” to “Bought”). To visualize this, see our old friend the Venn diagram below:


The overlapping area would include the book “Flora & Fauna”. If I were to search for the tags, “Hobby”, “Books”, “Shopping”, and “August”, that book would pop up.

I have no idea why they did that. I think it would make more sense to make hierarchical notebooks, not tags. No wonder it doesn’t work right.

Personally, tags kind of turn me off. I have no desire to file a document in my neatly organized hierarchy of folders (well, I try to be neatly organized) and then have to tag everything, too. However, other people think differently and find tagging works better for them.

As my grandma used to say, " ‘Each to his own notion,’ the lady said as she kissed the cow."