noteworthy wrote:
] As your once and future agent will kindly tell you, 2005 will
] be remembered as the year that remembrance agents went
] mainstream.
]
] This essay by Steven Johnson appears in the Sunday NYT Book
] Review.

Linked here is a related post from his blog...

noteworthy wrote:
] As your once and future agent will kindly tell you, 2005 will
] be remembered as the year that remembrance agents went
] mainstream.
]
] This essay by Steven Johnson appears in the Sunday NYT Book
] Review.

Linked here is a related post from his blog...

[ Interesting... this is an area of particular interest to me. I downloaded the tool under discussion, called DEVONthink, and gave it a whirl.

Initial impressions :

* johnson is right, it's useless for large documents. It'll certainly tell you that a word or phrase exists in that document, but then you still have to seach that document for the relevant locations. some mechanism for automatically breaking the imported docs up into segments (i.e. on a paragraph break) would make this more useful.

* it shows all the imported files in folders and so forth on the left. this is kinda dumb. if i wanted a hierarchical folder structure, i would use one with actual hierarchies of folders. oh, right, i do that, and it sucks. the whole point of a program like this is that hierarcies don't work well for complex data, and rigid groupings are completely contrary to the task. this leads me to my next point...

* the groupings should have nothing to do with the folders the documents were in orignially, nor should i *have to* create groupings on my own. the computer should categorize them for me. yes, i know, this is hard. if it was easy, i'd have done it 2 years ago, and someone else would certainly have done it by now. if i *really* want to create my own category, and add objects to it, the software should let me, but by default, it should be able to say "this document is about 'internet', 'law' and 'file sharing'" and put that document object in those catogories. in fact, really, it should break all documents into pieces and do that with every single piece, and anytime a piece is displayed, provide a button to show the whole document, if i want it. mostly though, if i'm interested in finding things about the legal issues surrounding file sharing, i don't give a fuck where the file is on disk based on my previous shitty filing scheme, or, in fact, about the 75% of the document that doesn't pertain to my current query. ok, enough.

* it doesn't handle pdfs. oh, it imports them, but they may as well be images. that's not useful to me. i can search filenames already, and, again, i don't want to have to categorize things myself. i do that now, with folders. i can appreciate that this may require paying money to adobe (does it?), but still, for a 40$ product, i'd like to be able to search the actual content of my pdf.

i don't want to sound like i'm bashing it too hard. in a sense, yeah, i'm complaining that my ford focus doesn't go 200mph and get 500 mile to the gallon, but i need to be honest about the utility of this kind of product. it's about as good, and probably better, than other similar products of it's type, and as a kind of jot board, for storing and managing notes and short rants or whatever, that you type directly into it, it looks to be extremely useful, but it's not yet what I need. Which is to say, an external memory with recall that's akin to, but better than, my own brain, which has trouble remembering things that it's seen before.

if he's right, and this is the year for this kind of stuff, i feel pretty sure that someone else will implement what i want well before i do, but i'm pretty confident that i'll know it when i see it. -k]