Conceptual Problem Burdens ADM 3 and Other Programs
Posted by srdiamond15
on 12/6/2004
srdiamond15
12/6/2004 2:34 pm
People who find ADM 3 cumbersome may reflexively find the main problem in the gui, but that's an illusion. The main problem with ADM is conceptual, its excessive and self-defeating complexity as it tried to implement two roads to the same end.
ADM starts with powerful though imperfectly implemented outlining features, to which it has tried to add extensive database functions. Unfortunately, its designers implemented features before exploring their logic, and ended up making the program less effective if anyone tries to use all the features.
The two functions are outlining with cloning and key word categorization. As I pointed out in a post a while back, categorization in an outline, where an item can be assigned multiple parents, is functionally indistinguishable from assigning items to keywords arranged in a hierarchy. When ADM allows and even encourages users to use outlining with assignment of keywords without any hierarchy, it is functionally as though the user has besides to his outline a second outline with a flat hierarchy, the two outlines being separate entities. The user is allowed to do some things with key words he can't do with outlines (e.g. form Boolean combinations); and certain things with outlines he cannot do with keywords (e.g. find items at different levels of generality or abstractness) but each way is handicapped by the inability to combine the information in key words and outlines seamlessly in a single procedure.
Here is how it should be implemented. The basic mechanism for assigning keywords should be the outline, and Boolean operations should be performable on any outline headings. It should be possible to assign items to headings by an alternate *mechanism*, assigning keywords to the item. Keywords would be equivalent to a "Favorites" listing of outline headings, that can be applied on the fly, with the results being manipulable in the ordinary way within the outline.
Only two programs that I'm aware of properly treats outline headings also like they were key words, allowing Boolean operations directly on the outline headings: Idea! and InfoHandler. (Both German companies. Does anyone know if Germany is in he vanguard of knowledge management.) Of those I can speak with certainty only of Idea!, not having actually used InfoHandler.
The next best is UltraRecall (http://www.kinook.com brought to my attention on this forum. It sets up the outline categories properly and doesn't introduce redundant keywords like ADM,but it doesn't allow Boolean operations directly on outline headings. This might be partly remedied by putting outline heading names in its full-featured search, except for a bug that will be corrected causing the search to ignore keywords in the "Data Explorer" (the overall outline). UR is conceptually inferior to Idea!, but has the better writing environment, including (I think alone among outline-based databases) unlimited universal undo/redo. It has a tabbed multi-document interface and multiple selection in the outliner, including non-contiguously. For me the choice is between the two. I have bought both UR at the special intro rate and I have a license for Idea! Professional. My tentative apportionment between them is to use UR for writing projects and Idea! for work flow.
The other products that have interested me a lot--ndx Cards and Miss LonelyNotes--have come close to following ADM in drowning in the keyword versus outline-heading bloat. Both of these programs use outlining to organize a project such as a writing plus use a flat keyword list for more permanent organization of knowledge. This is less problematic than ADM's approach, but only because the problems it creates of separate organizational schemes is less pervasive. Since in general the more thoroughly you cross-classify data the better you will be served by a database, and since failure to adequately cross-classify is the biggest problem in the use of such databases, there is no reason why the user must scrap a system of organization that has served him for one project. It is usually better to super-impose it on the permanent organizational schemes, as but one more set of hierarchically organized elements. But if the user makes his outlining organization a permanent fixture, he runs into the same problem as in ADM, two insulated systems of classification serving the same purpose.
ADM starts with powerful though imperfectly implemented outlining features, to which it has tried to add extensive database functions. Unfortunately, its designers implemented features before exploring their logic, and ended up making the program less effective if anyone tries to use all the features.
The two functions are outlining with cloning and key word categorization. As I pointed out in a post a while back, categorization in an outline, where an item can be assigned multiple parents, is functionally indistinguishable from assigning items to keywords arranged in a hierarchy. When ADM allows and even encourages users to use outlining with assignment of keywords without any hierarchy, it is functionally as though the user has besides to his outline a second outline with a flat hierarchy, the two outlines being separate entities. The user is allowed to do some things with key words he can't do with outlines (e.g. form Boolean combinations); and certain things with outlines he cannot do with keywords (e.g. find items at different levels of generality or abstractness) but each way is handicapped by the inability to combine the information in key words and outlines seamlessly in a single procedure.
Here is how it should be implemented. The basic mechanism for assigning keywords should be the outline, and Boolean operations should be performable on any outline headings. It should be possible to assign items to headings by an alternate *mechanism*, assigning keywords to the item. Keywords would be equivalent to a "Favorites" listing of outline headings, that can be applied on the fly, with the results being manipulable in the ordinary way within the outline.
Only two programs that I'm aware of properly treats outline headings also like they were key words, allowing Boolean operations directly on the outline headings: Idea! and InfoHandler. (Both German companies. Does anyone know if Germany is in he vanguard of knowledge management.) Of those I can speak with certainty only of Idea!, not having actually used InfoHandler.
The next best is UltraRecall (http://www.kinook.com brought to my attention on this forum. It sets up the outline categories properly and doesn't introduce redundant keywords like ADM,but it doesn't allow Boolean operations directly on outline headings. This might be partly remedied by putting outline heading names in its full-featured search, except for a bug that will be corrected causing the search to ignore keywords in the "Data Explorer" (the overall outline). UR is conceptually inferior to Idea!, but has the better writing environment, including (I think alone among outline-based databases) unlimited universal undo/redo. It has a tabbed multi-document interface and multiple selection in the outliner, including non-contiguously. For me the choice is between the two. I have bought both UR at the special intro rate and I have a license for Idea! Professional. My tentative apportionment between them is to use UR for writing projects and Idea! for work flow.
The other products that have interested me a lot--ndx Cards and Miss LonelyNotes--have come close to following ADM in drowning in the keyword versus outline-heading bloat. Both of these programs use outlining to organize a project such as a writing plus use a flat keyword list for more permanent organization of knowledge. This is less problematic than ADM's approach, but only because the problems it creates of separate organizational schemes is less pervasive. Since in general the more thoroughly you cross-classify data the better you will be served by a database, and since failure to adequately cross-classify is the biggest problem in the use of such databases, there is no reason why the user must scrap a system of organization that has served him for one project. It is usually better to super-impose it on the permanent organizational schemes, as but one more set of hierarchically organized elements. But if the user makes his outlining organization a permanent fixture, he runs into the same problem as in ADM, two insulated systems of classification serving the same purpose.
zeoli
12/6/2004 9:25 pm
Stephen,
I think I agree with your conclusions -- if I understand them correctly -- and would like to add a comment.
Most information database programs are built around the outline or data hierarchy. If they allow for keywords, the keywords are secondary, an additional way to categorize. In my opinion, it should be the other way around. The keywords (or categories, whatever you want to call them) assigned to a piece of data should be the primary data structure, and the outline should be a more fluid, even temporary way to organize the data for any given project. Don't get me wrong, outlining is crucial, but it is always going to be changing depending upon how you need to use your data. Information databases built around an outline hierarchy quickly become overwhelming, in my opinion.
That said, I also am not a big fan of being forced to add keywords to every piece of data in order to find and retrieve it (this is the problem with Personal Knowbase and Literary Machine, for example). This is why I love Zoot. You never have to do any of this categorization with a piece of data... just clip it or create it, and it is easy to find and collect when you need it. At most, you just need to know WHICH database to drop it into, but even that really isn't necessary. Zoot's two main drawbacks are its lack of text formatting capability and its really clunky outlining. Truly, you can't really outline with Zoot at all. However, it is easy to clip stuff from Zoot and drop it into BrainStorm. These two programs together are magnificent... except for the lack of text formatting in either.
Steve Z.
I think I agree with your conclusions -- if I understand them correctly -- and would like to add a comment.
Most information database programs are built around the outline or data hierarchy. If they allow for keywords, the keywords are secondary, an additional way to categorize. In my opinion, it should be the other way around. The keywords (or categories, whatever you want to call them) assigned to a piece of data should be the primary data structure, and the outline should be a more fluid, even temporary way to organize the data for any given project. Don't get me wrong, outlining is crucial, but it is always going to be changing depending upon how you need to use your data. Information databases built around an outline hierarchy quickly become overwhelming, in my opinion.
That said, I also am not a big fan of being forced to add keywords to every piece of data in order to find and retrieve it (this is the problem with Personal Knowbase and Literary Machine, for example). This is why I love Zoot. You never have to do any of this categorization with a piece of data... just clip it or create it, and it is easy to find and collect when you need it. At most, you just need to know WHICH database to drop it into, but even that really isn't necessary. Zoot's two main drawbacks are its lack of text formatting capability and its really clunky outlining. Truly, you can't really outline with Zoot at all. However, it is easy to clip stuff from Zoot and drop it into BrainStorm. These two programs together are magnificent... except for the lack of text formatting in either.
Steve Z.