Identifying Design Principles
From Visualization Sp06
Lecture on Feb 28, 2006
Readings
- Pictorial and verbal tools for conveying routes, Lee & Tversky (pdf)
- Rendering effective routemaps, Agrawala & Stolte (pdf)
- Identification and validation of cognitive design principles for automated generation of assembly instructions, Heiser et al. (html)
Optional Readings
- Designing effective step-by-step assembly instructions, Agrawala et al. (html)
Demonstrations
Ryanaip - Feb 27, 2006 11:26:43 pm
The Lee & Tversky reading makes a point about the ability to translate between using depictions and descriptions for conveying directions. I think the paper is quite interesting (particularly having seen LineDrive previously), but I'm not sure of the generality of the relationship between depictions and descriptions. It may be the case that they are equivalent only in very limited domains.
Textual descriptions of directions are not abstract -- they don't make use of the facility of language for describing abstract ideas. Rather, they are a direct translation of the sequence of physical actions required to get from one place to another. I would imagine that, given a more naturally abstract domain (say, instructions for working together on a team), it would be more difficult to translate from a description to a depiction. The similarity in structure between depictions and descriptions of directions suggests that it's possible to construct either descriptions or depictions for concrete domains, but I'm not convinced that the same holds true in abstract domains.
I wonder if people naturally prefer to give visual or verbal directions, in the absence of constraints. My guess would be that visual encodings would be prefered, except in circumstances where drawing a good map would be difficult (e.g., over the phone, or with most computer mapping programs). More abstractly, do people prefer to use depictions for any sufficiently concrete domain?
Cynbot - Feb 28, 2006 01:48:14 pm
I like the way that the assembly instructions paper breaks down the effective communication of geometric information. So many times in paper-and-pencil drafting classes (which I took in high-school for two-years) those steps were performed by trial and error which resulted in aggrivating erasing and redrawing.
The step by step assembly paper also reminds me of this site:http://nationallampoon.com/nl/08_features/ikea_desk/ikea_desk.asp
Lesliei - Feb 28, 2006 09:14:12 pm
One of the aspects that struck me about the readings and the discussion in lecture was the role of text. In class, our informal poll showed that some prefer text-based driving instructions. However, the assembly instructions experiments showed that participants did not find text useful. Has anyone investigated guidelines for when text is useful and/or can replace other representations (such as visual ones)?
Maneesh - Mar 01, 2006 12:06:40 pm
Choosing the right representation (text or digrams) is a question that cognitive scientists have spent some time studying. A classic paper on this topic is Larkin and Simon's Why a Diagram is (Sometimes) Worth Ten Thousand Words. Don Norman also talks about choosing the "right" representation in his book Things that Make us Smart. In addition, Zhang and Norman have written a really interesting paper on the representation of numbers.
Alexd - Mar 01, 2006 05:22:27 pm
Don Norman came today to speak at the ISchool (South Hall). His presentation was called "Cautious Cars and Frightened Wives." His presentation revolved around automatic driving systems, from automatic parallel parking systems to advanced cruise control and braking systems. His overall argument is that automated processes should be seamless, fluid, and work well, or, not automated at all.
He has a lot of interesting things to say about design, as Maneesh mentioned. For example, if you look right-hand column on his website http://www.jnd.org/ he has an interesting take on product manuals: "Is a manual important? Yes, but even more important is a well-designed product, one so well conceived and constructed that either the manual is not needed at all, or if it is, where the manual can be short, simple, and easy to understand and then to remember."
His take on product manual writing and design could be summed up by Einstein's quote, "Things should be as simple as possible, but no simpler."
Maneesh's work identifying congntive design principles seem to support Norman's claim. It would be interesting to see if the design principles advocated by Norman's model technical and product manual writers also translated into a solid foundation for clear, effective design/assembly diagrams.
Sharena - Mar 01, 2006 09:18:25 pm
I had not seen LineDrive before and I thought it was very intersting because i have often had to print out 2 version of a map, one showing the whole route and one closer up showing for example the path from the getting off the freeway to the destination. So line drive seems to simplify that, however I do wonder about seeing more surrounding streets to give more context about the area. But it is a very interesting idea of taking how people think about and depict maps and directions (as they talked about in the Tversky and Lee reading) and turning that into a mapping system.
Gwyu - Mar 01, 2006 10:17:47 pm
I agree with Sharena's point about LineDrive that showing more streets to give more contextual information about certain areas (especially around difficult intersections) would probably be helpful. Also, some of the examples of participant-drawn maps in the Tversky and Lee reading include both landmarks and crossing streets that indicate that a driver has gone too far (Figure 3, which shows Stanford Stadium and Serra Street). I'm not sure if adding these elements would be helpful or add unecessary clutter.
Bryan - Mar 02, 2006 09:59:19 am
The act of identifying design principles is not something I had thought of so explicitly before. However, from reading the papers it becomes apparent that as soon as you want to accomplish something automatically, you are suddenly required to specify things which are implicit in the skill of a designer. I gained respect for automatic generation.
I wonder what the limits of specifying design principles for automatic generation of maps, instructions, &c are. For example, how complex a manual task does it take before it is much more effort to specify a principle than to manually create design elements for each of the visualizations? Screws and nails seem simple enough, but what about tying of knots, cutting of shapes, folding, and other common construction tasks?
Pushkar - Mar 02, 2006 11:26:37 am
Tiny comment about the design of driving directions: I find it very useful to know when I've missed a turn and gone too far. Something like "Exit on Street X...if you see Exit Y, you've missed the exit".
Modifying the online driving directions to specify this kind of information seems straightforward, but I've yet to come across a service like this. Is there a special reason why not? (Or am I in a minority of people who actually like having this service?)
Noaa - Mar 02, 2006 08:54:30 pm
I agree with Bryan's comment: it seems like designing furniture assembly diagrams would be something that someone would manually do to make sure everything looks right and is clear. The fact that you can automate something like that really makes you wonder how much of design needs to be hand done and how much can be automated, and where else these principles can be applied. I was talking to Sharena about this in class but I wonder whether something like how to knit can be automated: you input a certain design and it shows you exactly how to do everything, since in my experience with books it's really hard to tell how to do things correctly.
Mattkam - Mar 02, 2006 11:24:46 pm
I would like to follow up on Pushkar's comment since it's so interesting. My impression is that even though we have been talking about providing context to map displays, this appears so far to be a static notion of context. It seems that with location-aware devices becoming more popular these days, maybe it is possible to incorporate a more dynamic notion of context. As an example, a GPS-enabled portable map display in a car could infer that the user/driver has made a mistake once he deviates from the prescribed route, and hence provides a slightly different context by zooming into his current location and giving suggestions on how to return to the prescribed route.
Cynbot - Mar 03, 2006 12:58:18 pm
Right it is possible that IKEA and other prefab places have found a niche macket where: 1. images are foremost because they ship to so many countries and can not possibly make detailed text in all those languages and keep costs low, etc. 2. the operations have to be simple enough for people to do at home
Owners manuals for cars and electronics are often volumes of information utlizing more text than images.
Nchentan - Mar 04, 2006 09:35:54 pm
Adding to Pushkar's comment: It seems that there might be some advantages taking into account the likelihood that the driver will make mistake, in deciding which context to provide in the map. The likelihood could potentially be a set of simple heuristics. For example, on a highway, it is rather likely that the driver will miss a small exit. Therefore, providing context that specifies something like what Pushkar comments would likely to be the result.
Brien - Mar 07, 2006 03:55:54 am
Sometimes it's hard for me to match landmarks drawn on a map with their physical appearance. For example, everyone might know mention a specific church that's not clearly labeled. It would be interesting to pull images from emerging urban image bases like urbz and put them in the map somehow. That might enhance the landmark effect and might give a greater pool of available landmarks to pull from. For example, the wall with the cool spiral graffiti in a city could become a visual landmark (until it gets erased by “beautification”...).
Yi-Tao - Mar 07, 2006 10:16:58 am
One of the advantages of text over a diagram is that it explicitly tells me what to do. For example, in a route with many turns, I don't have to look at the map and figure out if the turn up ahead is a right or a left. However, text directions are useless when the action is unclear given the actual topography. For example, once the directions told me to continue on the road but the road clearly forked and there were no street signs. A diagram would have helped but I'm not sure if the diagrams shown in class would work since they distort the layout. For complicated areas (like a 5 way intersection), an exact layout would be far more helpful because there's no doubt for the driver.
Also, I wonder what if there's a study on whether the distance distortion in the maps help or hurt people. In the cross-country map example, I looked at it and wondered why they had directions that makes the person go in a perfect circle. Then I realized it was just the on-ramp expanded in size. Such representations are counter-intuitive and probably take time to get accustomed to.
Raymond - Mar 07, 2006 01:08:06 pm
In my UI class, we just talked about 3 design principles (from Bauhaus):
1. form follows function
2. economy of form
3. integrity of materials.
Looking at the examples produced by the automated program, I believe it incorporates all three principles efficiently. I also agree with the other's comments that a manual isn't needed as much if the product/application has a good design. In essence, one should incorporate usability/design in the beginning.
Cynbot - Mar 08, 2006 10:31:26 pm
Raymond Which I classs is that? I would like to take that class.
