Saturday, April 26, 2008

Weinberger on Pepper on Topic Maps

David Weinberger of Everything is Miscellaneous fame writes about a talk by Steve Pepper on Topic Maps in his blog. He seems to endorse the claim that there is something wrong with the "document-centric" approach that characterizes much of the thinking and praxis of knowledge management. We should become more "subject-centric" because "documents are about subjects. Subjects exist as concepts in our brains. They’re connected by a network of associations. Docs are how we happen to capture and communicate ideas. “Hypertext has been barking up the wrong tree” ever since the memex. ... We should be organizing information around topics/subjects, not around documents."

Topic maps are also said to "reflect the way we think."

I am not sure that this is true. Perhaps topic maps reflect one of the ways we think, perhaps they reflect some of the implicit presuppositions of the way we think, but perhaps they do nothing of the sort and simply reflect what some people have thought. But I will have to think about it.

Monday, April 21, 2008

A Victory of Information over Experience?

In an article published in The Times of April 18, 2008, Ben Macintyre comments on writers who have written travel diaries about countries they have never, in fact, visited. His first example is from around 1850, the second from 2008. (See Economical with the truthiness.) He concludes that (i) "bogus travel writing has a long and inglorious history," and (ii) that the more recent example "is representative of a wider and more modern malaise: writers reviewing books they have not read, politicians claiming to have braved dangers they never faced, novelists depicting places they have not seen, memoirists describing a past that never happened, journalists making up stories about people that never existed, and, most pernicious of all, writers simply cutting and pasting words they have not written." He then claims: "This is the victory of information over experience. In Wiki-world, where so much semi-reliable information is available at the push of a button, there is no need to see something first-hand in order to be able to describe it with conviction and authority. A comparison of Paris guidebooks reveals entire chunks of identical text for some tourist spots: why actually visit somewhere to find out what it is like when one can merely paste together a version of reality?"

It's difficult to disagree with with his first conclusion: "bogus travel writing has a long history." In one sense, it would also be difficult to disagree with the second conclusion, namely, that we are facing a "wider" problem or "malaise" with semi-reliable information and plagiarism in various forms. But it is difficult to see why this is a "more modern" problem. As his lead-in shows, the problem or malaise is as old as modernity (and probably a little older).

It is just false to say that this "is the victory of information over experience" or a characteristic of the "wiki-world." Mere information and experience, or perhaps better: knowledge, have always existed together, and information always brought with it the danger of crowding out knowledge and experience. It has always taken critical minds to distinguish mere appearance and truth. How quickly semi-reliable information is available does not fundamentally change our ability to investigate and examine it. The "victory of information over experience" is also a mere appearence. Furthermore, it has little to do with "wiki-world," simply because there is no such thing. A wiki is a tool, not a world. Like any tool, it can be used to do a variety of things: some of them good, some of them less good.

"Why actually visit somewhere to find our what it is like when one can merely paste together a version of reality?" What about: "To find out which version is preferable?" Experience and knowledge are not just occasions for a descriptions "with conviction and authority," they can be ends in themselves. Wikis can be helpful in achieving these ends.

Saturday, April 19, 2008

ConnectedText 3

I have been using ConnectedText since August of 2005 (Version 1.0.0.2), that is, almost from its very beginnings. Before ConnectedText, I used Notebook, a tcl/tk application by William Duquette that I felt implemented the idea of a personal wiki in a most elegant way. There was, however, one problem that became more and more important to me, as my database grew larger, and this was the limitation of the search function. Notebook allowed (and allows) you to search only for a certain string. It does not have any "full text" search, or the ability to search using operators like AND, OR, NOT, etc. ConnectedText did have this capability. Furthermore, Notebook's database was a flat text file. This made searches fast, but it also meant that everything was in one large text file, and the question arose for me how well that would work, if I did not just have one or two thousand of entries, but ten-thousand entries or more. ConnectedText is based on an SQL database, and that seemed to be better. It also allows several databases to be open, allows searches across all open databases, and even inter-project links. All these reasons persuaded me to switch, even though there were some things that Notebook did better, as, for instance, the ability to embed searches in pages by using a macro.

It took me a while to transfer the thousand or so entries into ConnectedText, but since the Notebook file was a mere text file, it was not difficult to translate its format into the format required by ConnectedText and to remove information that was irrelevant. (This just goes to show how important it is not to be locked in by a proprietary format.)

I have never regretted the decision to move to ConnectedText. It has never crashed, and I feel my information was never in danger. In any case, eventually there were added export capabilities that allowed you to transfer all the information to text or html files. The newest version goes on step further. It can save all the information twice: once in the database, and another time as a pure text file in a backup directory (which, by the way, can be indexed and accessed by any desktop search program.) In other words, ConnectedText does not only not lock you in, it also plays nice with other applications. Import could, in my opinion, be stronger. It allows you to import text and html files, but it does not allow you to import one large text file as many topics, using a separator like "~", for instance. But that does not bother me at all, since I put all the information I collect and all my thoughts into ConnectedText project (or database) anyway.

Version 3 solves, by the way, the only misgivings I had when I changed from Notebook. It now has something called "Smart Topics," or topics that can contain search expressions and can thus give you an idea of what information there is in your database on a certain topic. Indeed, since ConnectedText's searches are much more capable than those of Notebook, the feature is much more sophisticated as well. I now have more than 6000 entries in my main project, called "Notebook," and thus this feature is more important than ever. If you index your project and don't create huge pages, that is, pages that are the equivalent of more than 50 printed pages, smart topics open almost instantaneously (just as the results of even highly complicated show up immediately).

There was one feature of ConnectedText that I did not immediately appreciate, but that has become more important to me as time went on, namely the ability to assign categories to topics, where categories can be used very much like tags. In any case, it allows you to categorize as much or as little as you want.

Another feature, which I thought would be more important to me than it turned out to be is the Navigator, which gives you a visual outline of how the topics are related to one another, very much like a mind map (or like the visual map of a program called The Brain). I sometimes use it, but less than I would have predicted. I prefer the "mental map" of the direct connections between topics created by the direct wiki links and by the indirect linking through categories. But there are others who find this visual map more important.

Between late 2005 and April 2008 so many new features were added that one might almost say that ConnectedText became a different application. But the "almost" is important, as it remained true to the spirit of its original conception as a personal wiki designed for note-taking. Some of the more imprtant of these features were: export capabilities, the ability to filter categories and topics, inline queries, print preview, a clipboard catcher, anchors and footnotes, an autolink function for topics, an outline view, a quite capable outliner that allows one to drag topic names as links into the outline, inclusion (transclusion) of other topics in any given topic, a table of contents view, various plugins for useful applications, and scripts, etc., etc.

Version 3 adds many tweaks and minor improvements, but the most important feature it adds in addition to a new editor that handles proportional fonts, indexing, and backup is the new semantic extension enabling markup of properties and attributes, which can also be used in full text searches.

Properties provide meta-information about a topic that is not displayed directly in the topic itself but can be queried for in a full text or semantic search. Thus, if there are certain topics that need further work or revision, one could add the property ($PR:) pending (Status:=PENDING). It could be something as complicated as the following: $PR:Status:=OK|DateDue:=20080519|Value:=1000.
This is very useful for "slicing and dicing" in ways that mere categories do not allow.

Attributes are used to mark information within the topic itself. They allow you to assign a value to a certain phrase, as for instance the value 3,000,000 to "population." If the topic contains the text "the population is [[population:=3,000,000]], the topic would display "the population is 3,000,000". Attributes can also be searched for in full text and inline queries, thus adding a further dimension to the data.

I do not think that it is an exaggeration to call ConnectedText now a "semantic desktop wiki". As such, it is almost without competition.

In any case, I am impressed with the new capabilities of ConnectedText. Time will tell, how valuable the semantic extension will be to me personally, but the beauty of the program is that you can use this feature as much or as little as you wish. It grows with you, or, perhaps more importantly, it allows you to grow with it.

Friday, April 18, 2008

Distributed Cognition

An great post for note-takers:

Tools for Thought,

without further comment.

Monday, April 7, 2008

Capturing and Keeping Notes versus Developing Ideas

There are many good applications that help you to capture, keep, and organize notes so that you can recall them in various ways at a later date. Most come with good search functions, and many also allow you to sort or order the notes in some rudimentary ways. Some use trees or a hierarchical structure for organization. Others use categories or tags to cluster notes in a more flexible way as networks of related topics. Still others, like EverNote, allow also a strictly temporal order.

Ideally, an application would allow for all the different ways of looking at the notes one has collected. Starting out from what might be called the "shoebox metaphor," that is, the idea that you have one large container into which you drop all your stuff so that you have it all in one place and thus have created the necessary condition for being able to find your "stuff" again, capabilities are introduced that allow you to impose order on that stuff that makes it still easier to find this stuff. Such capabilities also make it easier for you to get clearer on what you have collected, to see patterns and structures that you might have (would have) overlooked, if you had not put in the effort to order your stuff by putting it into a hierarchy or categorizing it in a variety of ways.

If we call this ordering of prima facie unrelated notes or ideas "external structuring" (in analogy with "external outliners"), we can distinguish this activity from what might be called the "internal structuring" of the particular notes that we have taken. Most note-taking applications are very good at external structuring, but they are not very good at internal structuring. They don't help you in getting clearer about the ideas contained in or represented by any one of the particular notes one has taken.

Indeed, some of the applications that make capturing and keeping information from various sources very easy get into the way of internal structuring of notes, as they make it easy simply to cop what one has found on the Web, in an electronic document, etc. But this can (or should) be only the first step. The information needs to be analyzed or decomposed into more basic ideas, which can then be used in various other contexts or connections. Ideally, I have argued before, there would be one idea in one note, and no more. This needs judgment, of course, which may or may not be supported by the note-taking software. And none of the tagging and organizing of notes that are themselves fuzzy or contain ideas that are unrelated or loosely related won't overcome this shortcoming.

One needs to think about the note or the idea itself. While getting clearer on what the idea is that the note contains or is about, will help also with classification. But this is not all, thinking about the idea will also make you realize whether there are other notes that are directly related to the note that you are considering or whether it leads to other possibilities you have not thought about before. In both cases, a direct link to such ideas might be called for.

This is what makes applications that allow of easy linking of different notes or ideas so interesting. Indeed, the principle here seems to be: the easier the better. And that is why an application that allows of wiki-type links, like ConntectedText, has in my view a crucial advantage over applications that have more clumsy ways of linking existing notes or creating a new note on the basis of a mere link.

Now, it might be objected to what I have just said that this is just another way of talking about external structuring, as I am just talking about linking notes. Perhaps, but I don't think so. Ideas have connections and getting clear on what connections they have, or creating a new connection, is also getting clearer about what that idea is itself. It develops the idea. The starting point is different. It starts with the idea and its internal structure, not with the "stuff" and what it might mean.

To be sure, the two activities of internal and external structuring are complementary. They are certainly not independent of one another. But they are just as certainly different from another. And both need to be supported, or so I would argue. Nor are direct links the only way to develop ideas. The ability to outline particular notes and "do" other things to them or with them is also essential. Note-taking, or "note-making," as some people like to call, is to engage actively with the world -- and yourself.

ConnectedText 3.01 is out. I will write a review of it and its advantages in the next few days. This review will also develop further some aspects of the idea presented in this entry.

Sunday, April 6, 2008

Personal "Outboard Brain"

Here is a Jon Udell interview with Phil Libin, the CEO of Evernote, Personal Outboard Brain. in which he describes the planned future of his product as a "personal memex." It's very ambitious. The interview has some relevance for anyone interested in note-taking.

I am not sure the online "solution" really is a solution.

There are other interesting interviews in the Interviews with Innovators series.

See ITConversations

I found especially interesting the interviews with Ward Cunningham of Wiki-fame and with Jeannette Wing from Carnegie Mellon University on "computational thinking."