Re: text conventions/spec. round off &c

["Huber, Mike" <mrhuber -at- SOFTWARE -dot- ROCKWELL -dot- COM>]

Allow my to expand on this: don't expand the precision of a converted
value to match your conversion table.

"I went for a ten mile hike." does NOT translate to "I went for a 16.10
kilometer hike."

I had an editor change an article I wrote once. My article said that a
program required "about 580K RAM to run effectively". The editor changed
it to "593920 bytes", and didn't understand why that's wrong.
Eventually, the editor changed it to "60000 bytes" just to make me go
away. "593920 bytes" implies more precision than the tests were designed
for. We didn't check performance at 593900 bytes.

Excess precision in this kind of situation isn't so much a matter of
misleading the reader (unlike the case of technical diagrams) as wasting
the scarcest resource a communicator needs - the reader's attention.

---
Office:mike -dot- huber -at- software -dot- rockwell -dot- com
Home:nax -at- execpc -dot- com


>-----Original Message-----
>From:  Ridder, Fred [SMTP:F -dot- Ridder -at- DIALOGIC -dot- COM]
>Sent:  Tuesday, April 14, 1998 9:07 AM
>To:    TECHWR-L -at- LISTSERV -dot- OKSTATE -dot- EDU
>Subject:       Re: text conventions/spec. round off &c
>
>...
>One case where false precision can creep into dimensions is when
>converting between metric and inches. If an object's dimension is
>defined to 1/10 inch tolerance, there is no justification for giving a
>metric conversion to anything finer than 1 mm (roughly 1/25 inch).
>Only if an object is dimensioned to 1/1000 inch tolerance should its
>metric dimension be given to hundredth of a millimeter (and at that
>point, it's probably also appropriate to explicitly indicate the tolerance
>since 1/100 mm is less than half as big as 1/1000 inch). Also, be
>sure that the conversion factor that is being used has an appropriate
>number of decimal places for the number of significant digits in the
>dimension itself.
>