Re: [OT] Re: Offshore Outsourcing
From: Edward G. Nilges (spinoza1111_at_yahoo.com)
Date: 06/06/04
- Next message: Otto Wyss: "Tutorial and guidelines: Coding standards (long message)"
- Previous message: Brian: "What computer language is used a lot in the IT industry?"
- In reply to: Nick Landsberg: "Re: [OT] Re: Offshore Outsourcing"
- Next in thread: Mike: "Re: [OT] Re: Offshore Outsourcing"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: 6 Jun 2004 04:02:42 -0700
Nick Landsberg <hukolau@NOSPAM.att.net> wrote in message news:<rZwwc.39278$_k3.955535@bgtnsc05-news.ops.worldnet.att.net>...
> Edward G. Nilges wrote:
>
> > Nick Landsberg <hukolau@NOSPAM.att.net> wrote in message news:<lMawc.35438$_k3.838455@bgtnsc05-news.ops.worldnet.att.net>...
> >
> >>Edward G. Nilges wrote:
> >>
> >>[SNIP]
> >>
> >>
> >>>Modern PCs, what with fans blowing in all directions, represent a
> >>>series of social compromises and market pressures.
> >>>
> >>>But if Moore's law is repealed, then Ancient Programmers will perhaps
> >>>be called from retirement to once again write "tight" code in assembly
> >>>language.
> >>
> >>You mean there's hope for me yet? Seriously though,
> >>the only non-trivial program I ever wrote that worked
> >>the first time out of the box was a disk bootstrap
> >>written in assembler. This may either be a failing of mine,
> >>or an indication that reliance on HLL's and such
> >>leads to slovenliness. I don't know since I only have
> >>one data point. (Note that I consider any compile
> >>time error in a a higher level language, even a typo,
> >>as equivalent to the program "not working,"
> >>for the purposes of comparison.)
> >
> >
> > Actually it is more likely that compilers will have to improve their
> > optimizers if CPU speeds stop obeying Moore's law.
>
> Not altogether, I don't think. The choice of algorithm
> is still up to the "coder." Optimizing a bubble sort
> is generally a waste of time. Unless you mean that
> optimizers should somehow use the most effective
> sorting algorithm for a given situation. That leap
> of faith is still beyond what I would expect given the
> current state of the art.
It is true that nothing can save code which uses a bubble sort for
large data sets. It is also true that nothing can detect the equal
mistake, which is not to use the O(n*n) exchange sort (kin to the
bubble sort) when the size of the data set is sufficiently small.
However, the applications programmer may in future have no reason to
code a sort per se. For one thing, much sorting in MIS programming is
now replaced by the ORDER BY clause of SQL.
>
> >
> >>>We should never rule this out. Y2K called forth, from under the Rhine,
> >>>troops of old Cobol programmers to rise up and slay the Y2K bug.
> >>
> >>This conjures up images of Valkyries (to the music of
> >>Wagner) riding forth and doing their thing. I just hope
> >>the Valkyries are not coming for my remains :)
> >>
> >>
> >>>More recently, even more ancient analog programmers have been summoned
> >>>from beyond the River Styx by wireless companies to once again mutter
> >>>their dark incantations.
> >>>
> >>>Last seen in 1960, dragging their oscilloscopes and slide-rules to
> >>>their Ford Pinto in the parking lot at IBM accompanied by Security,
> >>>they have, we may imagine, been long moldering in California on early
> >>>retirement while building differential analyzers out of mere water,
> >>>but today they are being called back to renumerative employment.
> >>
> >>I don't live in CA, and I don't work for IBM, nor do I build
> >>differential analyzers out of mere water (deuterium oxide is
> >>a better alternative:), and I'm not yet retired, but the analogy
> >>is apt. And yes, I still trust my slide rule more than I
> >>trust the "calculator" thing on the screen. The slide-rule
> >>forces you to think. The calculator does not.
> >>
> >
> > The slide rule also forces you to round your measurements down to two
> > places which according to my physics prof was about the limit of
> > precision of ordinary measurements in the laboratory.
>
> Absolutely true, although when I had keener eyes, I could
> judge to 3 places sometimes, if it was at the lower end
> of the scale.
>
> >
> > Dan McCracken was sufficiently of the slide-rule generation to write
> > Numerical Methods and Fortran Programming which warned the tyro not to
> > be beguiled into actual imprecision by the apparent precision of
> > computer math.
>
> I have never read that particular book, but I have read other
> books on numerical methods. The point being (as far as I am
> concerned) that there is a difference between accuracy
> and precision which is lost on most folks. The slide-rule
> generation (as you put it) learned not to be obsessed with
> precision because they had no choice in the matter. Some (most?)
> of the current generation seems to want to present numbers
> (not data) to 6 or more decimal places just because their
> calculator (or Excel) presents it that way.
>
> This is probably a failing of an educational system which
> puts a premium on the "right answer" whereas in real-life
> there is usually no one "right answer" but an approximation
> based on assumptions and variables over which one has no
> control. But that's a subject of another post, if ever.
>
> >
> > Unfortunately I bought my first slide rule in college two weeks before
> > I bought my HP 65.
> >
>
> Then you missed out on learning about essential estimation
> techniques while using the slide rule. But so have
> the folks which grew up after you. A pity.
Actually, I was fortunate in that my first Fortran compiler (one that
I debugged and deployed myself in the absence of IBM support) had a
memory-saving option to limit decimal places and, having full access
to the mainframe late at night, I was able to get a feel for precision
and false precision.
One can learn responsible numerical analysis using digital as well as
analog computers; Dan's book proved so. The problem is that most
teachers of computer science lack the background to do so.
Indeed I feel that a small subset of post-computer students including
my kids are more and not less mathematically aware owing to their
access from middle school to advanced machines, affordable by the
parent and justifiable as an educational expense, such as the HP48. As
long as the kid is motivated these devices teach math far better than
some abusive gym teacher assigned to monitor the math class.
I was in fact outraged when one of my kid's teachers told me at PTA
that he was "wasting time" by sitting in the back of class hacking his
HP-48, for I knew from my own experience that my kid was doing a
better job teaching himself, in his case with the aid of the 48, than
the teacher was. Part of the sadism of education is the overstress on
"paying attention".
I have long felt that all computation inside a digital computer should
be performed with real number "objects" with an unlimited mantissa and
an unlimited exponent. However, not being a numerical analyst, I can't
prove that this will avoid errors in all cases. Furthermore, it only
works for surds and when the result of a calculation is either a
continuing fraction or an irrational, the object would have to switch
into the suitable representation of the number such as 1/3 (which
wouldn't be representable in a base 10 floating point number).
Of course, the base could be put in play as well in a true object
oriented implementation.
I've just reinvented Mathematica, of course...which was in fact
invented by a genius who is a surd in an irrational world, Steven
Wolfram, and who also probably had "attention deficit disorder". This
is because Mathematica encapsulates the insight that computer
limitations on numerical precision, while a reality, don't have to
exist.
John Nash was in fact busy inventing unlimited precision when I
assisted him at one point, but Nash (another surd) switched to
Mathematica in later years.
I conclude that there is a level of mathematical insight far beyond
what high school or society wants people to have. It is more
convenient for people to worship reified numbers.
Another fascinating detail I learned recently is that Don Knuth, yet
another of my surds save for his rather endearing interest in the
Bible, an unusual interest for a surd, is still plugging away at
completing The Art of Computer Programming and has taken retirement
from Stanford to do so. Thus we can yet hope to have all volumes on
our shelves at some future date.
>
> Hmm... that's a leap of faith I wan't expecting.
> One can enforce security down at the ASM level, if
> it has been specified as a requirement. I was not
> talking about the cell phones themselves, but about
> the base stations and other equipment which make
> cell phones possible. Please expand on your point
> if you would.
I am a layperson as regards the technology, but aren't radio signals
both wireless and interceptable, by definition? And, if they are in
code, can't all codes be cracked?
- Next message: Otto Wyss: "Tutorial and guidelines: Coding standards (long message)"
- Previous message: Brian: "What computer language is used a lot in the IT industry?"
- In reply to: Nick Landsberg: "Re: [OT] Re: Offshore Outsourcing"
- Next in thread: Mike: "Re: [OT] Re: Offshore Outsourcing"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
