come_to_think: (Default)
Reading: The Design of Design, by Gordon L. Glegg (Cambridge U.P., 1971)

A delightful little book by a British professor of engineering.  I bought & read it shortly after it came out, but it disappeared at Twin Oaks.  Recently I recalled a joke that reminded me of it, so I had Amazon scare it up (deaccessioned by Northern Michigan University Library, where, it appears, no-one had ever checked it out).

A compendium of sage advice for budding engineers ("Disciplined thinking focuses inspiration rather than blinkers it"), illustrated with lively accounts of blunders, many of them committed by the author.  He was a sportscar enthusiast, whose attempts to push the envelope could be hair-raising.

Foolishness, of course, extends beyond the engineering profession to its customers.  One must "beware the danger of assuming that what people need is necessarily what people want to buy".  He invented a cheap, simple device to solve the perennial problem of judging the mixture of water & cement.  But:


Hardly anyone wanted an instrument like that on a contractor's site.  The man on the mixer who had been controlling by sight and slump did not want to learn new tricks.  His supervisor did not want to introduce a means whereby his boss could make random checks on what was happening.  The efficient contractors had control and inspection systems already established.  Some did not want to know too much, others thought they knew it all.

I can't remember how much difficulty I had following his examples the first time around, but this time, perhaps owing to senility, I found it mostly impossible.  Here is the first mystification, concerned with the design of car chassis & suspensions.  He says that early designs, lasting into the 1930s, were imitated from carts, ignoring the fact that in cars, torques are much more important than they are in carts.  "The total length of the frame members in a typical chassis...came to between four and five times the length of the wheelbase."  There is no drawing, but I can imagine that.  A typical cart is rectangular and hangs out beyond the wheels at both ends, so the side members will each be longer than the wheelbase.  The cross members will be shorter, but probably long enough to make the total more than 4 times the wheelbase.  Glegg does not explain, however, why that particular ratio is useful as a figure of merit --- is that supposed to be obvious to an engineering freshman?  And on the next page is a drawing of an Austin 7 chassis frame, of which he says that "the total frame length was reduced from 4--5 to 1 2/3 the length of the wheelbase".  The drawing shows two converging longitudinal members extending from the rear axle to the front axle, so each must be slightly longer than the wheelbase, and the two must contribute >2 to the ratio, not to mention the contributions of three short transverse members.  One and two-thirds?  Baffling.

By now, it seems, even when I am not reading about human beings, I quickly arrive at a point where I become, as the saying goes, unclear on the concept.  About 20 years ago, a friend gave me the famous book on C by Kernighan & Ritchie.  I could understand it until it got to pointers.  I couldn't see the point of them.  Likewise, in an elementary book on Lisp, I bogged down at the first example of a double recursion.  I could follow the steps, but was blind to the idea.  Luckily, in the dialect of Lisp that I need for customizing my Emacs, there is no need for recursion, and in fact it is recommended against.

This is the joke:


All engineers should keep in mind the story of the small boy who went into a shop and asked for a package of detergent.  'What do you want it for?' asked the proprietor.  'To wash my budgerigar in', said the boy.  'That won't do it much good', replied the man.

Next day the boy came back.  'How is the budgerigar?' said the proprietor.  'Dead', said the boy.  'I warned you', said the man.  'It was not the detergent that did it', replied the boy, 'it was the wringer.'

Designers, too, often fail to recognize where the real danger may lie....

Profile

come_to_think: (Default)
come_to_think

November 2022

S M T W T F S
  12345
6789101112
13141516171819
20212223 242526
27282930   

Syndicate

RSS Atom

Most Popular Tags

Page Summary

Style Credit

Expand Cut Tags

No cut tags
Page generated Jun. 10th, 2025 05:54 pm
Powered by Dreamwidth Studios