|
Floyd made landfall on the NC se coast, but significantly weakened, so it
looks more to be a matter of rain and flooding. So far I have noted no website
disturbances.
I wandered out on the town for some errands today, one of which was to find
a suitable "scientific" pocket calculator for Edward and school use. Unlike
the PC market, this appears to be a relatively stable niche market these
days -- math is after all math, and the feature/pricing steps were all pretty
much defined years ago. My eye fell on a TI-30S as a happy medium, and the
price was right.
Understand, when I went through high school, the Faber-Castell duplex
slipstick was king, and multiple LL scales were awesome if you knew
how to use them. I still have my F-C novo duplex in a drawer --
look, no batteries! Some people spent personal fortunes on the high-end precision
sliderulers, and cleaned and polished the sliders religiously. Ah, those
were the days...
Then when I reached University, the new HP calculators were the status symbol
of the day. People walked around with hip holsters and had verbal shootouts
about the relative merit of RPN (Reverse Polish Notation) versus standard
algebraic entry (i.e. the way it is usually written). The carnage was awful...
You could in later years distinguish the slipstick user from calculator ones
in various ways. The most obvious was that the slipstick users could work
out rough but order-of-magnitude correct answers in their head (accuracy
without precision), while the calculator users would give you many decimals
but could easily be several orders of magnitude off (precision without
accuracy). Strange, but simply punching buttons never seemed to impart
any sense of what was a reasonable answer or not. Nothing new really, but
so much more obvious when doing math was just an equals-button away. Bringing
math to the masses in this way clearly lacked something more profound.
A clear example of precision fallacy from my youth was a plywood factory
that churned out yield-analysis reports. To make plywood, you take logs and
sort of "peel" off the wood in a kind of lathe. Cut into rough squares and
trimmed, these sheets were then stacked before being dried, further processed,
glued and pressed into plywood. At various steps along the way someone would
measure the stacks. The long production lists of stack dimensions eventually
reached the main office, where numbers were crunched and yield reports generated
for management.
As summer trainee during my engineering studies, I was assigned to evaluate
this procedure. Rather early I suspected that things were not what they seemed,
because I saw yield figures with 6 to 8 significant digits (e.g. 21.1234%),
and some that simply had to be wrong in terms of accuracy. It was fairly
simple to conclude that the average precision was at best two digits at point
of measurement, and that a lot of the detailed analysis in the office was
utter nonsense. Stack measurement was a bit like getting a precise value
of material volume in a stack of potato chips. On the input side, the logs
were entered into the factory statistics as cu.ft. with around 6 significant
digits, but this was clearly only order-of-magnitude correct. The entire
"precision" thing ultimately rested on mechanical calculators and the number
of digits they displayed.
The best figures in terms of analysis could in fact be had with a slideruler
and callibrated eyeball -- and I believe some of the older foremen ran the
factory on the basis of experience and rule-of-thumb estimates alone. That
is up to then.
New management had decided to make things more on-paper stringent,
hence the office analysis, but clearly those in charge of this were more
accountant types than people with any real understanding of what the figures
represented. The plan was to build a new modern plywood line in part based
on the analysis. Interestingly, the factory went out of business only a few
years later, and some years after that became a parking lot.
Under this wreckage lies Samuel Joint,
Somebody dropped a decimal point. -- old school saw
I'm not saying that flawed math was the cause, but it can't have helped any.
The interest in my mathematical analysis was in any case minimal, and after
a few initial weeks of this, I was instead put to work repainting old machinery
("industry strength" paint where the foremen tell you to ignore the dust
and grease, just paint over it), and with chisel chopping out old glue from
clogged up pipes. The painting was interesting in a perverse way, because
the forman was right: you could paint over the dust and grease.)
Follow-up thread on wiki about
depreciated web pages. And another page
summarizing issues
accessing other filesystems.
|
The
update-link (above) points to where I last added
some text, which should simplify your keeping up to speed. Of course, this
assumes I remember to move it, and you may still have to scroll back a bit
and see if I've updated more than once since you last visited, but that is
easily done.
LeufNet
-- picking 5-bit ascii
codes :)
-- the IconShop window