Alex Payne Hates Devonthink (Ok, not really)

http://www.al3x.net/2009/01/31/against-everything-buckets.html

I’ve spent quite a bit of time getting DevonTHINK set up to my liking, and I’m happy with the results. For a year or so I thought such applications were overkill, but I eventually got tired of fighting with the finder.

I disagree with much of what Alex said, but he’s obviously a smart guy.

Is he off his rocker? Or does he make good points?

Structuring documents falls down when a single document might logically reside in two places. How do you remember which one it was in? Oh, so then you search.

The issue is, while Spotlight is very good at indexing documents, it is not as good at DEVONthink at search. DEVONthink I’ve found is very good at finding the exact document or documents I’m looking for. Spotlight usually brings back results that are overly broad, in a way I don’t like. As such, for that reason, DEVONthink is worth it to me.

Then there’s the OCR workflow (well, when it’s working :wink:… which is a huge timesaver.

I also say DEVONthink isn’t necessary an “everything bucket.” I mean, it has a hierarchy itself. And hopefully soon it will have real tagging, as opposed to the hierarchical based tagging in the 2.0 betas which is no better than the old groups way of 1.x.

I agree with him to a certain extent, but our conclusions are different.

I’ve ranted and raved on here about how none of the existing information management software is very good, and that saying “DEVONthink is the best” is not really saying all that much. Alex’s answer is to simply use the Finder because of Spotlight, the largely bugless folder metaphor of the filesystem, so on, so on – and I can dig that. The first hurdle any information manager has to leap is to replicate the functionality that already exists in the Finder! And not only that, but most information managers don’t replicate the Finder’s functionality very well at all.

sgmiller has stated a similar corollary – that we need to have all of the functionality of DEVONthink plus announced features like user metadata built, rock-solid, into the OS itself. And I agree with that too.

My conclusion, though, is that we need better information managers. Alex complains about having a lot of unstructured RTF and TXT files – and I’d add to that PDF, JPG/PNG (a little difficult to structure those right now), and any other format that people commonly use. Even OmniOutliner files are unstructured by this standard. But that problem exists in the Finder as well – not just within DEVONthink et al.

Unfortunately, the filesystem is not going to answer all of our prayers any time soon. I see the ideal information manager as effectively creating/emulating a filesystem that does provide all of the features we want from a filesystem in this day and age.

It looks like the Department of Veterans Affairs is sending me back to school to get a B.S. in Computer Science… while I’m there, I’m going to make one of my part-time jobs the design and construction of an object-oriented information manager with support for all of the things I’d like a 21st-century information manager to have – user-defined object classes with class inheritance and protocol adherence, complex extensible metadata, a pervasive interpreted language, scenario simulation, a mixed database model uniting object and relational databases, etc. So I wouldn’t expect this to come out even in alpha form for three or four years, and probably never, but if it comes to nothing I’ll at least appreciate what I’m up against :wink:

Thanks for the responses guys.

I haven’t been using devonthink for very long, so I don’t feel qualified to give an educated opinion. I’m sure the way I use devonthink could be replicated in the finder, but I find it much easier to use devonthink. Maybe if I spent more time trying to get the finder to do what I want it to, I don’t know.

Anyway, that’s why I was hoping for the opinion of more experienced users who left the finder behind a long time ago.

And reading the two responses so far, I’m even more confused. Your response, kalisphoenix, is very interesting. Not only your ideas, but the fact that I really can’t find reasons in it why devonthink IS better; why you ARE using devonthink. Of course, the reasons could be going over my head. :slight_smile:

I am wondering if this guy has ever used any of these applications he’s trashing, OR if he’s even read the documentation for DevonThink. Right out of the gate, DevonThink documentation encourages you to have a VERY structured folder structure. Doing this allows there to be a great deal of granularity in the folder structure, that DT can then automatically figure out where to put like documents so that they ARE in a structured organization. So his whole rant about people throwing everything into one big bin is completely opposed to the DT approach.

He also implies that these applications don’t do anything. The primary reason I’ve started using DT was to digitize and index a file cabinet full of paper. By using DT to do this, I can now easily find the things that used to be misfiled in my paper file cabinet. That is huge. Plus, I can drop a document link into other applications if and when I need to.

This guys seems to be kind of clueless as to what DT’s philosophy is and how it works. The only thing I agree with him on is that its best to stay as close to the non-proprietary beaten path as you can to avoid getting left on an island if some proprietary company blows up in the future. Short of Apple integrating all of this stuff directly into the OS, there isn’t a way to do that today. My method of protection is to create searchable PDFs in DT and give the files meaningful names at intake. Then if and when DT were to ever blow up, I can always export all the documents back into the file system in the same folder structure that matches the group structure in DT… and I’d loose the functions of DT, but not loose my files and organization.

Well, Alex Payne’s understanding of Devonthink seems to be derived from Steven Berlin Johnson’s descriptions, or something less, but not usage. I think most users, at least those in this forum, don’t use the database at all in the way he’s suggesting.

What I like about DT is exactly its ability to structure data, and make connections across that structure when the data violates or sidesteps the strictures of the structure. Finding and relating information in this way is continually the challenge for people who analyze qualitative data.

Sure, in the Finder (and increasingly in Path Finder), I structure my files in nested hierarchies that relate to their function- teaching, research, manuscripts, committee service, hobbies (for me those are all about bike racing), music, photos, etc. I know where to find pretty much anything I need to find.

Spotlight is a powerful indexer and search tool-- but not for the type of search functions my research demands-- fuzzy spelling, context of categories, boolean operators, often restricted, etc. I’m not looking for files, but bits of data. It’s not the same thing.

I guess I could make Spotlight act more that way with extensive attention paid to tagging in Spotlight comments- maybe using Default Folder X. But, the returns are still unwieldy. Spotlight for me works best as an application launcher or to find whole documents- which I can find in Finder just as easily.

I can’t comment on the other programs-- I played around a little with Yojimbo, Tinderbox, Evernote, Together-- but they didn’t work for the way my brain processes my research. I may well be, though, that is because of the type of research I do. In any case, Payne does seem to get the real utility of a program like Devonthink for qualitative researchers.

And why back on rtf and pdf files? If one really wants plain text, that’s an option too. Still, in my experience through the years rtf and pdf are much more durable file types than all the notes I took in WordPerfect back in the day, even more so than .doc files now being replaced by .docx. Is there a single word processor on the planet that can’t open an rtf in either mac or wintel platforms? Is there movement anywhere away from pdfs for online book and article databases like jstor?

Apologies for the longish counterrant.

If I followed that advice, I would have an application folder filled with one-trick ponies. Sure, each program would do it’s assigned task well, but I would have to learn how to use all of them, switch between them, switch data between them, keep them up-to-date, buy upgrades, etc. I’d rather use one program that does several things well enough.

None of us want to have our data marooned on an island if the application of the year is no longer developed and supported. We want a way to export our data and files to another application or OS.

Alex doesn’t recognize DEVONthink as any different than the other pure search index data management solutions, this is evident when he makes a case against “throwing everything into the system”. No one who actually uses DEVONthink would do this. Even text as far back as 2005 suggests storing snippets of text and not dumping entire volumes.

I don’t really think he is being specific, but unfortunately calling out specific applications deserves equal scrutiny on what stereotyping he is providing.

I don’t think the filesystem needs any scrutiny, Alex completely loses sight on CAPTURE… the reason people use programs like these is with a single press of a key i can get “data” from virtually any application into a system walled off from the system wide search index. Spotlight is a frackin nightmare, really now. Anytime I do a search with spotlight via FInder it’s like looking for a diamond in the rough. Usually my document surrounded by a 50-100 other random files on the system - none of which appear to have any relationship with each other either. So even the info mgmt solutions using a pure search index are still walled off from the system-wide search index.

Alex doesn’t get research vs recollection. So many Google references in “The Search Illusion” with text like “remembering a phone number”. Yes, Google is really good at looking up things like that – hell this is even discussed in the video tutorials on DEVONagent in DEVONacademy. Is there any antedote for his argument that suggests he’s even aware that DEVONthink isn’t simply a pure search index, nor can anything remotely compete with the advanced boolean logic BOTH DEVONthink and DEVONagent offer. Least of all secondary query. Nope not a drop of argument about anything more complicated than storing a phone number in your everything bucket.

Alex fails to recognize what a failure Google is as a research tool. If you are looking to innovate or otherwise researching unpopular material, Google is useless. You are forced to eyeball a long list (usually) of search results looking for your proverbial needle in a haystack while navigating advertisements and the investments to get better ranking in search. Business corrupts even search engines.

Someone said it best early but using one-trick ponies does often generate fail because often none of them are interoperable – instead you are getting lossy data as you export and import your way around applications. His clamor for everything tools that do each thing mediocre does not resonate as well with me as, my data comes into the system and never leaves until cooked into a production document – yet the source data stays in DEVONthink. He doesn’t appear to appreciate the beauty in this. Anyone that has had to drag data through a gauntlet of applications is aware that it’s rife with inefficiency and unforeseen showstoppers.

The worst part is the conclusion - it discredits the entire article. It’s sort of a jab (but not really) at DEVONthink, but then he doesn’t finish the argument… It sounds like someone saying, and this one thing DEVONthink has all this techno mumbo jumbo, but in the end if you just think about the task you probably don’t need all that crap… Check out what your friends use on I Use This or MacUpdate and find an app – Wait! you haven’t even offered an antedote after you’ve just called b*llsh!t on everyone who makes an everything bucket app…

Some good points have already been made here. An interesting point is made here:
markbernstein.org/Feb09/Ever … ckets.html
by the creator of Tinderbox (for our purposes, it’s not important what that is or if it’s any good or not). What he says may sound a bit nebulous but it’s quite true.

As for storing only snippets in DT, it depends on your usage. One of my uses of DT is to store pdfs of papers on theoretical physics. I have hundreds of them in a single database, classified in folders according to area etc. As I work/have worked on various areas, some have very little connection to each other, while others seem very different but actually use similar concepts and methods. It has happened that DT’s “see also” feature has turned up connections I hadn’t noticed before. This is probably because I only have papers in that particular database.

On the other hand, storing snippets is also useful. As an experiment, I stored some snippets from a book into my database, and DT found some unexpected connections to papers in a completely different area (which I did already have in mind, but I was very surprised to see turning up). The success rate was higher than normal, so it seems a good idea (although laziness prevents me from doing it more often).

If one was to just dump random stuff into a database, then I guess it wouldn’t really work any different from a hierarchical file structure, though. For example, storing huge course notes from the web into the same database as papers simply results in those notes turning up in “See also” for everything.

DTs search feature is also a lot better than spotlight for my purposes: it returns results in order of relevance and allows more focussed searching than the spotlight GUI (or HoudahSpot, which is the other front end I tried).

Bernstein makes a good point, and this is precisely what I find appealing about DTPO. It’s the “Swiss Army Knife” concept applied to data organization. DTPO offers a powerful set of tools, way beyond what’s offered by the Finder and Spotlight to enable someone to organize their data. One of the most important is DTPO’s scriptability.

Right now I’m working with several thousand NY Times articles, using DTPO, Applescript, Acrobat, and Skim to produce structured text notes for use in a program like Tinderbox. Without even talking about its “AI” capabilities, DTPO’s advantages over the Finder in quickly culling through PDFs, renaming files, etc. is a huge boon. If I was never to use DTPO’s search capabilities, it’s enabling me to do a job that would be unthinkable with the Finder, or software like Yojimbo.

Best, Charles

This has been a sometimes useful discussion, but after reading it all I thought “Who is Alex Payne anyway?” I thought he must be some deep thinker about information organization/mgmt.

So I looked at his original article about “everything bucket” software
al3x.net/2009/01/31/against-ever … ckets.html

and his own bio page
al3x.net/about.html

and his page about what he does—work, talks given, book written.
al3x.net/books_talks.html

I saw nothing indicating any particular expertise regarding the management of large amounts of information to produce something, such as managing many scholarly articles to support one’s enquiry into a question, or one’s writing of an article to advance thought on that question.

Alex works for Twitter and speaks and publishes about Twitter and its API (application programming interface —I admit, I had to look that acronym up). And about APIs in general.

We all have a right to our opinion about software or anything else, and equally the right to evaluate for ourselves the value of the opinions of others on any given topic. The topic of information management for a creative purpose does not seem to be one regarding which Alex Payne has any particular expertise. His opinion is his opinion. Others have pointed out that his post does not give evidence of real familiarity with Devonthink, since “everything bucket” does not accurately describe it. Devonthink is an “everything bucket” in the same way that a physical library is one: yes, all sorts of things (CDs, books, journals, maps) get housed in a library, but the library has a staff of (non-artificial) intelligences to sort it all out, and make cross-references and cataloguing decisions. Same with Devonthink.

Please, Alex Payne, I’m not attacking you, just expressing my opinion that what you said isn’t relevant to us here.

The person who started this discussion by referring to Alex Payne’s article was, I think, a relatively new user of Devonthink who was not sure if the “bucket” criticisms applied or not. If we agree that they do not apply, we can talk more clearly about Devonthink’s remarkable strengths and versatility. Perfect, no; solution to everybody’s needs, no; but an incredibly valuable tool for managing information, yes.