Axel's Thesis Scenarios

Thesis Scenarios

The two following scenarios, playing in the post-thesis time, visualize the space in which we attempt to support design activities, hint at interesting problems that arise, and indicate a contrast to the current work practice of designers with the written media. We recap both scenarios in the end.

The Kitchen

An architect and her client are in a discussion about a kitchen extension and work in front of an Architects Electronic Sketchboard.

The client brought the plan of his map on disk, and it opened in a translucent patch in the center of the electronic sketchboard, partly covering some work the architect worked on the day before. While discussing the client and the architects both sketched and drew on the sketchboard, extending the original patch by some area to sketch the kitchen extension, sketching a vertical view of the extension in some empty space, and creating a new patch for some comments the architect wrote down. The architect knew the client would want a rough cost estimate right away, so she started to write down the main cost items during the discussion.

Now, at the end of the session, she writes a number next to each item, a double bar below the list of numbers, and then chooses to apply a calculator interpretation to the patch. The interpretation recognizes the list of numbers, and adds them up below the double bar.

The client decides that he might do without the additional door. The architect uses a gesture to create a patch including the door item, and some gesture to delete the patch with content.


The door item in the list is deleted, and, due to the calculator interpretation, elements below are shifted up, and the sum is re-calculated.

Relational Datamodelling

A data analyst of ParRel Inc. and some top and mid level managers of American Steam Inc. a large shipping company, are meeting to discuss the benefits of relational databases. ParRel Inc., a manufacturer of parallel processing relational database machines, has done such pre-sales meetings with great success in the past. The strategy used is always the same: designing the essential relational model of the business together with the client while also collecting the relevant business questions. Finally the analyst shows how the interesting business questions can be answered with queries over the relational model. The clients like this approach since the model, questions, and queries are right in their work-context, and the benefits over their current disparate hierarchical and network databases become obvious very fast.

They are also impressed with the technology ParRel Inc. is using during these sessions. Before using portable electronic whiteboards, the analyst used to tape large sheets of paper around the walls of the meeting room. She walked around the room, updating questions, and tables. It was always hard to edit the tables according to new information coming up in the discussion. Foreign keys needed to be propagated, renamed, added and removed, or tables split into two, ... it was always a mess. After the meeting she used to take all the sheets to her office and had to transcribe all information into a relational modelling tool, the business questions and pseudo-SQL queries in a word process, finally merging everything together in order to send the client a followup document describing the results of the meeting.

The process is much smoother now. She is still using a pen, but the Modelers Electronic Sketchboard running on the portable electronic whiteboards supports her during her modeling session in a variety of ways. Initially, she asks the audience about the important business "things" and starts to write down and order those nouns. Using gestures she can edit them and move them around. Most of those nouns will eventually become tables. At the same time she asks about important business questions, writes them somewhere on the electronic whiteboard, and marks them as such. That will automatically collect them into a list she can come back to later. Sometimes she displays the list on one of the boards, and collects the nouns on the other. When the stream of business "things" from the audience trickles down, she moves all nouns collected, over to another board, and starts working on each of them individually. This is the time, when she starts to draw tabular lines around the particular noun, and asks how particular instances of this "thing" are uniquely identified, and what attributes and example values they have. By now she attached a relational-table-interpretation to the sketch of the table. That will ensure the integrity between tables when keys, or columns are renamed, added, or removed. Foreign keys refering to a primary key will automatically update when the primary key changes. Using the relational- table-interpretation will also avail her some operations she would not have gotten with a regular table-interpretation, e.g. she can drag in other tables into attribute columns, and they will manifest as foreign keys, the primary key is separated from the attributes by a double bar, etc.

Usually she does not beautify the tables at this stage of the process. This is a presentation trick, she learned that too "exact" looking tables don't entice as much discussion. Nevertheless she can use regular table operations like adding or deleting columns and rows. She executes them either via a gesture, or via pop-up menus by double-tapping. All labels, as well as the data is still kept as ink. There is no advantage at transforming them at this point.

Having done more than twenty such pre-sales modeling sessions, the analyst has a good idea when the essential parts of the business are modeled. She then proceeds to go over the list of collected business questions. For each question she asks the audience for the relevant tables, pulls them up and distributes them over the electronic sketch boards. Usually the audience is very quick about figuring out how to combine the information in those tables to solve a particular question. Through gesturing she attaches the tables, the necessary joins and the business question to a query object, for further reference.

At the end of the day, she beautifies tables, and questions. Since her handwriting is quite nice, she lets the system straighten out the tabular lines, and attaches a calligraphic renderer to the content of tables, and to all questions. She hands out a hard copy of all tables and business questions, and explains, that she will send out a follow-up document in the next days. She also remarks that the model created today can be used as a starting point for a more thorough modeling session once the client decides to go with ParRel Inc.'s offer.

Recap

Both scenarios show a group involved in a collaborative design process, in which the written medium is used to express ideas and document intermediate results. The integration of computational facilities in the paper medium enables a smooth transition from the "sketchy" stuff generated when expressing ideas to the more rigid structures needed for documenting and disseminating results.

Furthermore, turning the paper medium from a passive medium into a computational medium avails manipulation and transformation of content previously impossible. By attaching interpretations to selected parts of the ink the content becomes rigid enough for computational processes to work.

Yet, structured and unstructured, interpreted and un-interpreted content can stand side by side, and be transformed into each other. Thus different parts of a design can be formalized to different levels.

The use of an electronic pen as the input device allows a fluid transition from textual to graphical input. Graphical symbols and annotations can be drawn with more precision than with the mouse. At the same time the pen can be used for issuing commands via gestures, or with a double tap through pop up menus.

A spatial distribution of the content is emphasized through the sketch book metaphor and supported by the mix of textual and graphical information. Furthermore, the use of an electronic pen lends itself naturally to spatial access and spatial distribution of information.



updated 5 June 1995, (c) 1995 by Axel Kramer