Misanchuk, Schwier & Boling: Icons [My Response]

This paper provides a wonderful example of what happens when one chooses their subjects well. This does appear to be a common strategy in HCI. I for one, STILL don't know what a "LASSO" is or does - in spite of having tried to use one recently. The image fits the word nicely, but the rationale behind the choice of that word is not clear. The data in this paper states that ALL their subjects understood this symbol., both in its static and animated form.

Much of this is similar to what Misanchuk, Schwier & Boling say. As such, I have little to add here.

The most exciting phrase to hear in science, the one that heralds new discoveries, is not "Eureka!" (I found it!) But "That's funny ..."

- Isaac Asimov

Petroski overstates his case to the extent that I find it insulting. This chapter at least, is very one dimensional. He has also left out (or at least trivialized) the role of art, and, even more importantly, the influence of the 'found objects' and 'necessity' factors. I dare say many, if not the majority of made objects were influenced more by the "Oshit" factor than by a conscious, directed, engineering design. It is all too clear that the author is very much a stereotypical engineer - one who places engineering at the apex of skill development. To his great discredit, he trivializes craft and creativity. A very engineering thing to do.

Some things are designed the way they are for esthetic reasons, and many things have both qualities: functional form and artistic, or esthetic form. Many designs happened because we stumbled across them, found them while looking for something else, or were in a hurry. Check out the variations in the design of cork-screws (just below). It is most definitely not always about convenience or economics.

What if I happen not to accept this functionalistic view of all made objects? I tried Norman's trick of looking around my room and I did indeed find many objects whose form and structure appear to follow from their intended uses. However, I also found a number of objects whose forms do not really fit into this philosophy.

This lamp is almost purely a decoration - it would serve as a night light, perhaps, but that still doesn't explain its shape. Petroski would dismiss this particular 'design game' as irrelevant. This particular piece may indeed fall into that category, but there are countless others that went on to serve as the inspiration for other, more useful things.	These corkscrews all 'do' essentially the same thing. It would be interesting to find out which ones were designed by engineers and which were designed by 'craftsmen'.

Petroski's description of how a spell-checker works raises a very serious red flag for me. Word processors do NOT use the algorithm he suggests - it would be way too inefficient. This is an algorithm I might expect from a rank novice (which, by the way, is a typical programming style of many engineers). His explanation implies that the search must be exhaustive - checking each word against every other before being able to determine that a word is misspelled. This makes it an O(n²) search [read: order N-squared]. Our beginning CPSC students know this is a silly solution by the time they finish their first programming course. I jump on this example because I know a little bit about searching algorithms - there are many techniques, including various hashing algorithms that can effectively eliminate the need to 'search' much of anything in cases like this. This was true a dozen years ago when this book was published. I believe that at the time at least, spell-checkers typically used a 'bell-hash' - a process that transforms a word into a number, in such a way that words with similar letters and spellings will end up being transformed into similar numbers. Using methods such as this, the spell checkers don't even go near the vast majority of the words in the dictionary. A good hash algorithm would allow the spell checker to ignore as much as 98% of the words in the dictionary.

Spell-checkers do not search like we do when looking for a button in an unsorted pile. That's really quite absurd. Although still less sophisticated than the truth, spell-checkers operate much more like we do when looking for a name in a phone book. Roughly speaking, this is a binary search with a search efficiency of about O(log₂n). When looking for a name in the phone book, do you really need to eliminate ALL others before being able to determine that you've found the right name? Get real!

What I find so alarming about this description is not that it is so wrong, but that it indicates a profound lack of understanding. In and of itself, Petroski's erroneous claim about how word processors check spelling is neither here nor there, but it does indicate that he is willing to talk about things he doesn't really understand and to use them to support his arguments. For me, this brings the rest of his claims into question as well.

I would also suggest that equating a craftsman with an engineer, regardless of how 'savvy' he may be, is a step down, rather than forward or up for the craftsman. Craftsmen (and women) have something many, if not most engineers lack: an esthetic sense; the 'art' of what they do; a deep-seated appreciation for the beauty of what they do. If I were a craftsman and someone told me I might eventually be scientifically "savvy" enough to be labelled an engineer, I'd be incensed! In less than one statement, Petroski manages to insult craftsmen and scientists both. Engineers are NOT scientists, nor are they craftsmen. Perhaps that is indeed one of the problems with design today: engineers should aspire to become more like craftsmen and not vice-versa!

In his final paragraphs, Petroski trivializes aesthetics by saying that when esthetics are allowed to dominate, objects often leave much to be desired. He then goes on to claim that there is not much room for variation on chess pieces. While he is entirely correct in saying the number and function of the pieces is not really negotiable, their form certainly is. I am offended by his implication that aesthetics interfere with function.

If he had qualified his statements at the outset by admitting he was really only interested in "industrial design" (read 'boring'?), then his claims might have been somewhat more palatable for me.