Thursday, December 8, 2011

Computerized Tickler File

Andy Glew's blog recently presented a link to an article about organization on his personal wiki. This article brought out a few thoughts.

It seems that a computerized Tickler File could organize items so that the ticler alert is triggered at an appropriate time (e.g. during a lunch break).

An even more sophisticated system might use information beyond time of day (e.g., position [seated, standing, walking; workspace, conference room, other's office, hallway, outdoors, restaurant], activity [software application with focus/receiving input; input rate/type; frequency and time of last display update], psychological state [heartbeat pattern, perhaps movement and activity patterns, perhaps information about stress inducing events and healing events]) to trigger notifications (and even time the receipt of messages) to fit one's state.

E.g., the system could exploit 'necessary' context switches caused by delayed responses to inputs as well as introduce productivity enhancing distractions (e.g., perhaps when it appears one is spinning one's wheels working on a problem), well-timed encouragements/refreshments (e.g., a loved one might leave messages of encouragement--like lunchbox notes--which the system could pop up at an appropriate time or a reminder of an upcoming positive event like a work group BS session or one's child's first school play), or even advisory notices (for most people advice has to be carefully presented; rather than "You're getting tired; you should consider taking a break" some might respond better to "Your favorite muffins will be fresh in the cafeteria in five minutes" or "Remember that you wanted to talk with Joe about such and such.").

A human manager cannot afford to be aware of fine-grained conditions (and most people's sense of privacy and self-sufficiency would be violated by such). Even a human 'executive assistant' would not be able to properly support a worker at the granularity that a computer could (and being impersonal, a worker might not have as much concern about privacy--if only the computer knows--or dependence/competence; on the other hand, being corrected by an idiot computer can be more painful/frustrating than being corrected by a semi-competent human being).

In the article, Andy Glew also noted that the artificial size limit of ScanCards was inappropriate for a computer, but it seems to me that sizing can be a disciplining factor to push for concise presentation and might also be useful to present a visual hint of the nature of a note (size of a note and size of font can give clues about complexity, importance, etc.). It might be good to have a standard size note and use hypertext, elision/abbreviation (which can be conditionally unelided/expanded), and font shrinking to include more information.

I am not certain how to handle hypertext display. Endnotes can be frustrating because even with back navigation one can lose one's place in reading; even parenthetical expansion--where one would place a note mark in parentheses and an activation would expand/make visible the note--can interfere with visual orientation because the layout of a paragraph will change (vertical layout changes could be handled with a sidebar that changes color/shade/pattern based on semantic boundaries like sentence starts--interestingly, such a sidebar mechanism might be used to mark content density [side marks have been used to mark importance and other aspects of a text]; such would allow a reader to orient by the sidebar even if insertions moved text).

It is also frustrating that a single mechanism is used to provide notes of different kinds; notes can differ not only in length but also in tightness of the relevance--applies strictly to word, sentence, paragraph or ties to word, sentence, paragraph but applies more generally--, level of knowledge expressed in the note and expected of reader, nature of the note--e.g., a historical note like "this varies from version 2.3 . . ." differs from a warning note like "this is atomic not synchronized" or an explanatory note like "this is a variant of the Smith-Jones algorithm". Sometimes something like "tool tips" might be appropriate (like many browsers handle 'title' elements), possibly activated by a click/press-hold-and-gesture; sometimes a persistent frame--requiring explicit dismissal--might be used; sometimes an insertion into text might be appropriate. (Notes for a Shakespearean play or a poem by Alexander Pope might explain an archaic term or non-obvious allusion, a point of textual analysis, or a reminder of further information being available. Good versions tend to place the first type of note as side notes matched to the line--so they can be easily referenced without disrupting the flow of reading--and the middle kind as foot notes. The last kind of note might be best suited to an end note or be included in an opening or closing commentary.)

It does seem that a dynamic and intelligent display offers significant opportunities for improvement over paper-based information storage and presentation.

No comments:

Post a Comment