Some enhancements/feature request that would be nice

Mostly, driven from experimentation with Yojimbo. I personally have been using DEVONthink for about 2 years. I have tons of stuff in it, and it has been extremely reliable, but I’m always tempted to stray – by one or two things that are done “more nicely” in other applications. After a couple days of experiementing, though, I typically fnd that DT is nicer in most ways so it’s not worth switching.

Okay, blabbing aside, here are a few things in Yojimbo that I are great (or comments about tweaks to DEVONthink which could help here).

  1. It would be very nice to be able to view the entire library in one shot place. Just have a “Library” option, which would be a listing of every single item in the database. Makes it nice in the case where there are all sorts of nested folders to just access something if you know it mostly by name – without searching.

  2. “Smart” folders. I know this has been requested, and it’s in the plans for 2.0. As such, probably not a whole lot I’d need to add here re: details. Both Spotlight and Yojimbo offer ways to implement – both with more simplistic searching than DT offers.

  3. Tagging support. This is somewhat possible in DT today by using “info”, but Yojimbo’s support is a lot easier. You can have lists of pre-populated tags, and they are automatically generated when you type them in manually. Just makes it a lot easier to maintain some common lists of tags – this coupled with the smart folders would make it a lot easier to use search to keep lists, rather than hierarchial folders and replication as I do too often today. Pretty easy to see what they are doing with their implementation, and probably not a huge stretch to implement something similar on top of what DT already has. Another example of something very much like tags is metadata info on photos – for example how Aperture does this. It’s very neat to grab 5 or 6 keywords and just drag them on top of a photo. DT could offer something similar, to drag keywords to a document instead of typing them in the info panel.

I think there is some real elegance in how Yojimbo forces the “flat structure plus search and smart folders only” on users. Not that I think it should be mandatory, but implemented well I’m finding it’s often easier than digging through hierarchial folders – because I’ve very often found I can’t find things via this. I never remembe what folder I’ve used, and even with 2 or 3 replicants, I still miss the right folders :wink:

Hi, CatOne. Try the History view (Tools > History) as it is related to a couple of your issues.

First, it does offer a flat view of all of the documents in your database.

The initial view is sorted by Age, but you can sort the list in a variety of ways, including Name, Kind, Word count.

Slipped up and accidentally dropped a new document into the wrong group, and don’t know where it is? The History view may be helpful. Click on the document (which will have a recent Age if just captured from the Web or created by you) and press Command-R to Reveal the location into which you placed the document. Now you can move it to the correct location, either using the contextual menu option or drag & drop (for the latter, I usually use the Groups panel).

Currently, smart groups are created using scripts, and are updated whenever they are opened. One of the smart group scripts creates a smart group that displays items added within the past 7 days. As this is a script, it can be modified.

Searches provide another kind of ‘flat’ view of search results and the results can be sorted in a variety of ways. Remember anything at all about a document? It can likely be found by a search/sort. Remember that search results can be moved or replicated into a (previously created) new group for searching/sorting on that group.

Did a particular word occur in that document? Option click on that word in any other document and a drawer will slide out displaying a list of other documents that contain that word.

Some users create a ‘tagging’ document that contains a list of commonly used tags. That document can be kept open at all times. Just copy a tag to the clipboard and paste it into the Comment field of another document, or select it and drag it into the Comment field. (I keep the Info panel open all the time on the right side of my screen, so that the Comment field is always available for any open document.) Remember that Exposé lets one almost instantaneously switch between the open windows of the frontmost document.

I don’t think of my organizational structure as hierarchical, and I’m not forced into a rigid hierarchical structure. I often violate hierarchy by, e.g. replicating documents into multiple groups. I think of a group as holding a ‘cluster’ of related documents, and/or a ‘cluster’ of related groups. Some items fit appropriately into more than one such ‘cluster’.

I’ll refrain from talking about version 2.0. :slight_smile:

Hi Bill,

The history view is nice – it might be nicer if it were available as a “folder” in the regular list, and looked pretty much like a normal group. For now, it will work for me (actually found a couple huge documents with ~10 MB attachments which I can prune out)… so thanks for that.

Thanks for the info on the smart groups. I’ll tinker with them… I see right now they’re displaying or creating replicas for everything. Still… it makes finding things that are misplaced easier. Actually the smart groups reminds me of something else I didn’t note – a simple flagged field would be a nice addition. Flagging and smart groups is something I use with Mail quite a bit, and the other organizing tools have had it for quite a while.

As far as tagging and a separate document… I could do it, but if you haven’t tried it (no need to admit you have :wink: give a gander at how Yojimbo does it. It auto-completes all the tags, and does this automatically… it’s a lot easier than cut/pasting in the info box. Now I did note that you can multi-select documents and do a “get info” and change the comments for all at a bunch and the end result is good, but the whole process of tag entry and management of, say, Yojimbo, makes things very nice. You can do the tagging right within the main window for a note – makes it very quick to do this with no keypresses necessary.

As far as replication as compared to smart groups… the main thing is that replication is a largely manual process (well, DEVONthink’s smart groups as AppleScripts do this), but the overall elegance and speed is less than what YJ has today.

I will say though, in the testing I did, I exported the PDFs I have in the DT database (okay… actually I grabbed the “files” directory out of the package and put it on the desktop – dangerous but I know enough about what I’m doing here :wink: It’s about 2 GB. I tried to import into YJ and it crashed nearly every time. So I tried an import of 1/2 of them and it imported, gave me a ~1 GB Core Data database where the performance was absolutely awful. So DT definitely is by far a more scalable product at current… I’m probably locked which is just fine with me as DT has been so good to me… but I do crave a little more elegance in other ways to find the info. It’s possible to make it work in DT (seems there’s very little it cannot do), but it feels like you have to work at it a bit extra.

As far as DT 2.0 comments… fair enough I’ve read enough comments here. Hopefully there will be a fairly extended beta program where feedback will be solicited :slight_smile:

Hi, CatOne. Personally, as I’m managing well over 100,000 documents in various databases I regard tagging in general – no matter how elegant – as not only unfeasible but unnecessary except in special cases (like tax records). Unfeasible, for me, because I often dump hundreds or thousands of items into a database, and I’m not inclined to mark each item individually.

In a broad sense I do make use of classification – organizing related documents and groups into appropriate related clusters – as a form of tagging.

But even then, when I’m doing a particular project I’ll quickly create new clusters of groups and documents better related for that research purpose than my existing organization. Some of those new clusters are built using replication or duplication (duplication, when I want to be free to annotate or mark up documents without altering the originals) of search results. Some (often the most useful) are based on notes with hyperlinks that I create when following trails of ‘See Also’ explorations.

During the course of such a project I’ll mark drafts by state (complete or not) and will sometimes use labels on notes and references. That’s tagging; but when the project is finished I’ll search for and unmark those tags, as they would be irrelevant to the next project. And I’ll export the project group to a new database and delete it from my working database, probably retaining only the final project output (imported from an external word processor) and a note as to the project and the name and location of its database.

So yes, I suppose I do use tagging, but almost never a priori tagging beyond the level of classification – and even initial classification often isn’t appropriate to a particular project. I’ve never seen a tagging scheme that would cover all of the ways I may find a document useful, except for something simple like my database of financial records at tax time.

Feel free, of course, to ignore my personal idiosyncratic comments about a priori tagging. If you’ve got a scheme that works for you, follow it. :slight_smile:

Hey Bill,

Well… my current scheme for tagging (in DT) is reallly… not tagging. What I do today is pull in PDFs, web archives, or rich text cut/paste snippets from a variety of sources, as well as import RTF documents that I create in TextEdit (as meeting notes) – I do them in TE because I have a laptop where I take the notes, and a desktop which is my “one true database,” and there’s no good way to synch the laptop and desktop.

For everything but the notes I’ve created myself, I just chuck them into a “reference” folder with a number of more specific subfolders. But this categorization is very arbitrary – because sometimes I forget the subfolders exist or because it might fit in a few places and I don’t feel like creating multiple replicas. For the notes, they’re applicable to a specific customer, so I file them in a specific folder and drag them to my CRM application, Daylite, which does a better job of relating them to contacts, companies, and tasks.

So the problem I have today is that I can rarely find things that I have filed by digging for them. My filing system isn’t sufficiently consistent that this actually works. So I search (which works pretty well in DT, I must say – in particular it’s the only app of this kind that actually displays the results from PDF in context – which is HUGE). What’s the point of filing anything if I must search 90% of the time?

All this categorization and dragging though actually in many cases seems like more work than I might need to do. I think tags could be quite useful for creating smart folders in lieu of this filing – type in a few words that auto-populate from an existing tag list and then just leave them in the “library” folder. Now this is largely possible from DT today – using the “info” field combined with smart folders as you showed me. But, compared to the way YJ can do this, it’s extra work. Maybe not a huge deal, but an extra 10-15 seconds per note. I really think the usefulness of easy tagging and smart folders are co-dependent… one without the other provides little benefit.

Anyway, so my point is that I don’t reallly have a system that works well for me, given DT as it is right now. The YJ system (in 1.3) with tagging is a definite improvement in some ways, but YJ as a product is not nearly as scalable or functional as DT. I was calling out a few things (flags, tags, smart collections) that could serve as examples of “well implemented” functionality… if nothing else something to keep an eye on as 2.0 rolls along :wink:

Thanks for that, Bill. I’m personally at a loss as to why anyone would want to “tag” stuff on their own hard disk when searching is so much more flexible and speedy, and requires no overhead pre-formatting.

Tagging makes sense for non-textual data like, say, photos. It might also make sense in a social context, where users might want to group things for reasons other than content. Finally, tagging makes sense where the full text isn’t available. So makes sense because it ins’t full-text and is social. Flickr makes sense because it’s social and non-textual. Research files on my disk make no sense whatsoever because they’re not social and contain full text.

For non-textual data, why bother with DEVONthink? I mean, you can’t use DT’s excellent AI for images, so… why not just leave them in the Finder?

That said, the history panel (which I learned in a comment from Bill a couple of months ago) has revolutionalized my use of DEVONthink. And real Smart folders would be worth a paid-for upgrade, even without a new database.

When it first got popular i also thought tagging might be a wunderful thing - but as with sophisticated file-naming-conventions - i am just not using it, because it takes too much time.

So i stick with giving vaguely useful filenames and putting things in the vaguely appropriate folder (DT group). Thats not perfect, but better than the best tagging-intentions you had but never realized…

Hi, I’m testing now devonthink pro office. Some question to power-user.

Yesterday I scan some tax document to put those in devonthink. But I feel the lack of possibility to add field to a document. For example: I scan a fiscal document. Now, I need to insert some datas about this document: where the paper is; why I have this document; who handles the document, et ceterae. The only way I found is to build a sheet with datas and put the scaned document in a group with the sheet.
This is not a ‘bad’ solution, but it is a little redundant, and I build a sheet to use it for only one record.
Another missing feature is I feel is the lack of a model of sheet; I make an example: I build a sheet containg data for generic document. Everytime I scan a document I make a copy of the blank sheet, write datas, and group it with the scanned document.
After some work I see I need another column on this kind of sheet. Now: I have to open and edit all the sheet I have done and add the new column. There is not a way to have a model of sheet, so adding a column in a model, adds the column in all the sheets using this model.

As I wrote in another post, I think the sheets are a real great resource for a database as devonthink, but actually they seems to me quite rough.

I’d like to know the way how other users of DT uses the program to have datas for the documents archived.

Thank you.



I use an iMac as my main and desktop machine and have a MBP and use Chronosynch to keep them synched up … rview.html

The software lets you set up several synch scenarios, but basically my MBP is a mirror of my iMac - photos, music and DTP databases - it also will do your browser favorites and email - basically everything. The flexibility is pretty slick. I also have an external HD I use to do a weekly back up of my iMac.

While I am looking forward to Time Machine in Leopard, ChronoSynch will be what I use to keep my two machines in step.

Hi terciero,

I am referring to tagging textual data. Say I have a bank statement. Say its name (as downloaded) is xx77lksdf_110706.pdf.

Do I want to rename it? I guess I can file it in a particular folder. Or, with good tagging, I can just type ba and have a tag auto-complete as “bank statement” and then at any time I later want to find it, I can just type bank statement and there it is. Now, this assumes you have good tagging infrastructure. Sure, DEVONthink supports something close enough to tags (keywords) that you can enter via “get info.” But there is no pre-populated list; things don’t auto-complete, and you have to manually bring up the panel and type in every word every time.

Just try Yojimbo and see how it works – really. It’s simple, but it works very, very well with smart groups.

Now I’m not saying tagging has to replace search. And if you threw 100,000 items in the database it would be infeasible to tag every one. But I’m using DEVONthink for a number of things:

  • Customer notes
  • PDFs of product documentation
  • PDFs of “hobby” information – specifically photo newsletters from the past 15 years
  • Web archives of stuff I’ve captured
  • PDFs which are printouts of my tax records
  • PDFs which are Visa and bank statements
  • etc.

Sure would be nice if I could have tags for some of those categories. With a couple keystrokes I could auto-tag stuff, which would then make it so when smart groups are implemented, I could find these things without need to to dig through folders – assuming I don’t know any keywords in the document. And I have mis-filed before… it’s a pain spending 3-5 minutes trying all manner of search terms trying to find the phrase that finds your “needle in a haystack” of a document.

Again, this is all possible now with DEVONthink, in some ways. But it’s balky when you put it up against what Yojimbo 1.3 offers. Now YJ 1.3 is new, and it offers tagging support vastly better than 1.2 did (which basically offered “get info.”). Have a look and see what I’m talking about – sometimes it’s quicker to tag than file. Not always, but sometimes. And it allows fewer groups as well (say, a big folder with 10,000 items called “reference” that doesn’t need further subcategorization, because the tags will help you find stuff that you filed).


I use an iMac as my main and desktop machine and have a MBP and use Chronosynch to keep them synched up … rview.html

The software lets you set up several synch scenarios, but basically my MBP is a mirror of my iMac - photos, music and DTP databases - it also will do your browser favorites and email - basically everything. The flexibility is pretty slick. I also have an external HD I use to do a weekly back up of my iMac.

While I am looking forward to Time Machine in Leopard, ChronoSynch will be what I use to keep my two machines in step.

Hi crhooker,

I actually use ChronoSync :wink: It works quite well… and in fact I have it set so that when I connect via AFP to my laptop, that my DT database is synchronized. The problem is that this synchronization is essentially one-way… because the DT database is “opaque” to ChronoSync (it just copies files), if you add files to both the desktop and the laptop, ChronoSync will not know what to do. So I have to avoid adding anything to the DT database on the laptop – I save web archives and RTF documents to the desktop and then pull them into the DT database on the desktop when I have the chance.

Two cautions about using synchronizing software:

[1] Don’t assume that the synchronizing software can ‘read’ and copy individual files added to or modified in the monolithic body of the database, which holds all of your text-type files. ChronoSync can’t do that. So two-way synch of a database that resides on your desktop machine and your laptop will lose information.

I don’t trust any synch software to do something as simple as managing a text edit document that I’ve edited at different times and in different ways on both computers. The logical problems are simply too complex. The only truly satisfactory solutions would be to place two copies of that TextEdit file on both machines, as overwriting the older copy by the newer version would result in loss of information.

[2] If you are using backup software to copy your drive to an external drive, please make sure that you have first closed your DT Pro database. Copying an open database can result in errors; perhaps small errors, but they may cascade over time should you need to resort to that external backup. BTW that’s true for copying many file types that are open at the time of copying. Perhaps there’s unsaved information still in memory, for example.

Unfortunately, typical backup software isn’t smart enough to recognize open documents and close them before making a copy.

This is of course very true, so my process is to only work one machine at a time and always sync before switching


Searching is not a decent substitute for a tagging system. If it were, then I wouldn’t be spending multiple hours trying to hack the stupid functionality into DEVONthink.

Yojimbo’s implementation is great. I downloaded it and looked at it lustfully. It’s too bad that you can’t nest folders or smart folders like you can in Journler. Journler’s implementation is great too – better, really. But I hate Journler’s database management – everything’s thrown together in a proprietary database, and it’s slow, and it can’t handle huge collections of documents.

Here’s to a DTP2 with tagging and real smart groups: c[_] <-- beer mug

Why not? Just saying it isn’t proving it.

I think I understand something of what CatOne is talking about. I presume the problem is that we fundamentally differ in our purposes for DEVONthink. CatOne uses DT as, essentially, a scrapbook. I use it as a research assistant.

My personal feeling is that DT is a kinda lousy scrapbook. It’s not easy to file stuff, replicants are more work than they should be, and the interface isn’t nearly as flashy as many. Many other programs (e.g. Yojimbo, SoHo notes, Circus Ponies) are easier for data input, whether by index or import.

But nothing in the world can compare with DEVONthink’s AI capabilities. And for that, as far as I can tell, tagging is pointless. I value DEVONthink’s see also function above anything on my computer. Seriously. It is the most important application I use. But tagging has no place at all. And, from what I understand, DEVONthink’s AI takes hierarchies into consideration.

CatOne mentions that tagging isn’t feasible when we’re dealing with thousands of documents. Hello: I deal with thousands and thousands of documents. And, (I may be very wrong about this next statement, and am willing to concede should it be incorrect), I think the thousands-of-documents, exploit-the-AI is DEVONthink’s sweet spot and target audience.

So my current use case uses lots of replicants in a hierarchy. I file things instead of tagging them. And the interface is pretty sucky for that. It’s tolerable, but not great. But once I get my stuff in, DEVONthink is magic. DEVONthink is the #1 reason I will never switch to a PC.

If tagging would improve that, then please explain it to me. I’m not seeing it yet. Being able to search for “Bank Statement” rather than, say, the name of the bank that’s in the heading of every bank statement seems redundant to me.

Here’s a use case I’d like to see explained: I’ve got a 40-ish page article on marriage in seventeenth century England. Sure, I could tag the file extensively: marriage, sex, england, 17c, james, stuart, catholic, protestant, puritan, religion, virginity, wife, husband, church, priest… I could go on for a couple of pages here. It is, after all, 40 pages of text.

And then what value would it provide? All those words are in the text. They will likely appear multiple times. They file will come up in a search even without the tags. And, more importantly, it will show up in a “see also” when I look at an article about the attitudes about virginity during the interregnum. Without tags.

Maybe you’re talking about something shorter than 40 pages. Let’s take a 100-word document, just a note I’ve taken from a reading. I could tag that as well. Or, I could count on the note itself. But the same rule applies. If it’s not in the note, then it shouldn’t come up in a search anyway. Why tag the note with information that isn’t in the note itself? (NB: I always include complete bibliographic information when I take a note.) My note will come up when I search of the right terms, and will be found when it is appropriate in a “see also.”

The only reason for tags in my situation, as I mentioned before, is for non-textual data. Of course pictures won’t show up in searches without some textual metadata, and, as I have already stated, that seems like a valid place for tags.

But honestly, can anyone tell me how tags would improve my life? What am I missing?

I’ll give you an example. I work for a fairly large computer company. One which gets a lot of press coverage, and which people blog about all the time. People write about our products, people review our products, and people often compare our products to the competition. And sometimes they go into more depth on the products than I know of (and I’m a sales engineer for the company).

So I get an article that’s a mix of all of the above. Maybe it does an in-depth review; it compares/contrasts it on features/functionality compared to the competition, and clearly spells out the good and bad things. Maybe there are a couple customer quotes in this article as well.

So this article is a review. It’s also competitive information. And it’s a reference. Where do I file it? Do I really file it in one, and replicate to all 3? Gee, that’s at least 3 actions – right click->replicate right click->replicate drag->move. Okay. Well how about just typing in review competition reference and then dragging it to the “stuff” folder. I can find it by a search, or it will automatically appear in those 3 smart folders.

Sure, DT is a great reference and research tool. It’s also nearly as good as YJ and other “snippet managers.” With a couple simple tweaks it can be the best at both – so why not suggest it? Tagging is optional – again, I’m not going to do it on all articles… just the ones I want to go into smart groups I create… and tagging + smart folders is way faster and more sensical than replication for this. ESPECIALLY if you figure that you want to add another smart group later.

Also, recall that neither competition, review, or reference may appear in the article at all… so searching won’t find them via those words. Sure, the product name will appear… but with let’s say 10-15 articles a day coming out when new products are introduced… getting back 100+ articles as a search result isn’t always optimal. I’d rather do “product name” reference and have them all come back in a limited set.

But what you mean for tags? Someting like columns in a sheet?
As I said before (but nobody answer me!) I have doubt about devonthink for non searchable documents. For example: how can I use devonthink to catalogue a handwritten papers? I can use the office version for scan the document, but I do not use the OCR and I need to insert data like the paper original position, what the document is talking about, who created the document, et ceterare. And I need those data for each document, I cannot create the same fields for each scanned document.
The only answer I thinked is to create a sheet with ony one record, copy to the folder I’m inserting the document and group it to the scanned document everytime I scan. But I was wondering if it is the best solution.


I’ll try to do just that.

Background: I’m writing a series of ten books. Long books. Life’s work.

  1. Full-text search doesn’t deal well with things like book-planning, where most of the words you’re likely to search for show up in every plotline, storyline, character description, book summary, chapter draft, and foreword that you create. Boom. I think we could stop right there.

  2. Full-text search returns all replicants. Is there a way to turn that off? Because even if my search results weren’t cluttered by the presence of 95% of the documents in my database, they’d certainly be cluttered by the presence of ten copies of every single document that I have to have because of the replicant system.

  3. Full-text search doesn’t deal well with synonyms, homonyms, implied connections (ie, something that refers to something that refers to what you’re searching for)… in other words, search will never be as intelligent as I am. It’s fine and dandy to click on “See Also” if you want to see some vaguely humorous associations that the software makes… but if you want to control your database, you have to be the ultimate authority.

  4. In a program with a well-designed tagging system, tags could be applied to or removed from a mass of documents in less time than it takes to auto-group or auto-classify them, and with 0% chance of errors, and with no extra work required. I’ve auto-grouped twice in DTP. The first time basically seized control of my computer for twenty minutes or more. The second time I did it on accident.

  5. If DEVONthink had a tagging system, it would be an immense timesaver to people like me who had to cook up a Wikipedia-esque system of [[Category]]s and [[Tag:_________]]s and [[Class:_________]]s just to replicate that functionality. As you yourself have stated, it’s not easy to file things and the replicants are more trouble than they should be. So why don’t I use Yojimbo or Journler? Because, as CatOne mentioned, they don’t deal with large databases well. I tried them – believe me, I tried.

  6. I believe that you and Bill are looking at tagging in the wrong way (or at least a way different from the way in which I look at it). Tagging is not a tool for finding documents or finding associations between documents. It’s a way of binding a group of documents together in a way that makes it possible to find each and every one of them instantly in the future. Sort of like putting them into a folder, except with tagging and smart groups you can put one together in 3.24 seconds flat instead of dragging a bunch of replicants to a new place. Sort of like the existing smart groups, except that you don’t have to write a custom applescript for complex boolean operations that are going to drag ass in Applescript anyway but that update in real-time and can have actual arguments passed to it instead of needing its own document (which then gets returned as a search result, natch) or a specialized markup language in the comments section or a title which is passed as the arguments and makes it impossible to sort these “smart” groups in a way in which you can actually find the group you’re looking for at any given point in time. Exactly.

  7. Sometimes it’s necessary to have descriptors of a file NOT be in the original file. An example would be images and PDFs, sure. Another example would be stories, which should not have all the publication information (e.g. lit mags in which they were published, dates submitted in them, responses received and the content of those responses, etc). It’s a liability in many situations even to have links to other documents containing that sort of information. You could throw all of that information in a few documents or sheets, but it would take roughly the same amount of time just to tag it – and once the information is in tags, you can find all of, say, the stories that you haven’t published. Or all the ones you’ve published in Mid-American Review. Or all the ones you published last year, so that you can look at your count from this year and be depressed. Any sorting along these lines would require multiple extra steps without tagging and result in either extra files, editing of the comments (which is hardly as quick or as nice as editing tags in Yojimbo), or a horrendously crowded group hierarchy. What I would prefer would be to create a smart group or hierarchy of smart groups on the fly and then knock them down when I’m done. This is only quickly possible with tagging.

Now, I understand that tagging may seem unnecessary to some. I maintain that it’s essential to me. You can say that DTP can’t be all things to all people, and that’s true – but additional capabilities are never a bad thing so long as performance is preserved.

I hope this answers your questions somewhat. If you have more, I’ll do my best to answer them too :slight_smile:

Sorry for double-posting, but I read your post through again and decided that I wanted to tackle your examples directly. I might not be able to convince that you that it’s useful for you, but hopefully I can convince you why it’s right for me.

That’s an excellent example. If I had that document in my database, I would tag it as follows:

[[Maya Busomis Alwaysheaving]] (character name)
[[Family History]] (explaining how this is relevant to the character)
[[Alwaysheaving Family]] (which family to which this is relevant)
[[Volume One]] (which volume in which this is mentioned)
[[Volume One:Chapter 47]] (which chapter this is mentioned in -- Volume One prefix is necessary)
[[Volume Two]] (another volume in which it's mentioned)
[[Volume Two:Chapter 23]] (if the prefix weren't there, it would show up for  Volume One, Chapter 23 and Volume Two, Chapter 47, even though it appears in neither)
[[Sociological Articles]] (this is a cultural article, after all)
[[Event:1654 11 23@21:03 -- Maya loses her virginity to Lance Throbbing.]] (associates this with an event in the series -- Maya losing her virginity)
[[Event:1656 05 12@15:00 -- Maya and Coitus Interruptus get married.]]  (After Lance runs off with a circus midget, Maya is married by her father to the local ADHD-stricken scholar of Roman descent)
[[Event:1656 05 12@22:35 -- Coitus discovers that Maya is missing her hymen.]] (In mid-poke, no less)

Now, from those tags (none of which appear in the original text), you can imagine how many smart groups in which that would appear. Now imagine that I realize that Coitus and Maya should – no, must! – have gotten married a month earlier, and I have two hundred articles that reference aspects of that event. If that doesn’t cause your head to sink down into your hands, and tears to well up, and your sphincter to slam shut, well, you’ve just never been in that situation. You search all those articles, try to do a group-edit to remove that tag and replace it with another, and –

It’s NOT possible because no two documents have exactly the same tags… if you use DEVONthink’s article comments as the repository for tags. If you use external documents or sheets to store tags, then your troubles are just beginning. If your articles are all user-editable and you don’t mind having tags cluttering up the document and requiring a full-text search to use, then AppleScript will bring you to wildly gesticulating heights of fury as it drags ass through each and everyone one of those documents… after you write the custom script.

But it’d take five seconds in a system with an evolved tagging system.

The short and sweet summary is that metadata is often necessary in any project requiring considerable planning and careful construction. Especially years of planning and careful construction. Why? Because it is used to explain how the bricks fit together to make a building. It’s DNA, essentially.

Most interesting discussion!

Here is my two cents:

Somebody above made a comment about using DT as a “Research Assistant” versus a “Scrapbook.” I think this is a crucial distinction. I use DT to transform masses of mostly online material into a reports. As such, I see DT as an Information Processor which puts me squarely in the Research Assistant Category. I have no need for tagging because as Bill pointed out, it would just slow me down. If I need something in more than one place, I replicate. Having bunches of replicants around doesn’t bother me because when I am done with a report, I generally just archive the DT database and don’t look at it again unless I am going to update that report.

As I explained in a previous post, I archive all my important stuff outside of DT in the normal file structure using a series of techniques to “stamp” the reference information. I do that because I learned the hard way not to commit myself to any proprietary file system, no mattter how good it looks at the current time. In Scrapbook mode, I do feel the need to tag from time to time because keyword searching can sometimes absolutely miss the real context. For example, I have a whole bunch of stuff that I think would be good for presentations that I tag that way because how the hell can I remember months down the line all that disparate stuff I thought would be usable?

So, bottom line, Scrapbookers are going to want tags and I understand that. Information Processors like me probably don’t see the sense, particularly when you are trying to process over a thousand documents at a time.