[ale] New website for H1B [Slightly OT]

[John Marasco]

Do you track the cost of doing business with inefficient code from "cheap
programmers" while you identify those portions of the code you want
"corrected" by the guru?  Who identifies which portions of the code need
"correcting"?  What is the cost of that process?  Does the "corrections"
made by the guru change the training you've instituted for the users of your
code, what is that cost?  How about repeat of QA, what is that costs?  Do
you document the new business process, what is that cost?  How about
management of this process, what is that cost?    How informative are
comments made by "cheap programmers", is there a cost associated with
decoding comments by individuals that can't even get the code straight?
What is the cost of having a cheap programmer (who would reasonably be
expected to work slower) do more work by "mandating comments" in the code?
What is the maintenance and feature addition cost of this patchwork code?
I'm certain you can run a business the way you've outlined.  It would be
interesting (at least to me) to see if the costs of running business like
this really outweigh the benefits.  On it's face, it seems redundant, poorly
planned and piecemeal but I'd like more information to consider the idea if
you are finding success with it.

IMO, most people wouldn't find rocket science to be "easy math" (can't speak
for you, don't know you).  It involves control of a likely non-linear plant
model in a non-linear environment (gravity, air pressure, Reynolds Number,
that sort of stuff).  This means digitization of non-linear equations and
some rather complex matrix manipulations most of which I've forgotten.  The
math alone is easy to screw up and the effects are often unpredictable and
difficult to see until you are actually "in the environment".  Compounding
this is the necessity to digitize all your final models.  Throwing a few
extra dollars into the mix for better hardware would change the sampling
rate necessitating very flexible and well organized code or a complete
overhaul of all the digitized equations so that "solution" isn't as helpful.
Real time control is one area where throwing hardware at the problem creates
additional issues or it's own special up front design considerations.

Wouldn't distributed processing systems constitute "tough code"?  Where you
have to design a system to utilize the processing power of multiple machines
through multiple code layers (user, business, data, etc...) to achieve a
scaleable, flexible and least cost solution?  Where the hardware costs are
actually a significant portion of the budget and require careful management.
Isn't this type of coding ubiquitous to many industries in our increasingly
information services oriented marketplace?

-John

-----Original Message-----
From: ale-admin at ale.org [mailto:ale-admin at ale.org]On Behalf Of
To: ale at ale.org
ChangingLINKS.com
Sent: Tuesday, May 27, 2003 9:06 PM
To: ale at ale.org
Subject: Re: [ale] New website for H1B [Slightly OT]


>  I've worked with many domestic and foreign programmers that ran the gamut
> between the two.  The point I was making yesterday was, if you want a
> programer that is going to code like a $75/hr programmer, don't go
shopping
> for someone whos book rate is $20/hr and expect the same results.
>
> I am steak,
> -Grant

I have posted at least 1 job that many $75+ hour programmers would not take
because it was "over their head/not in their coding spectrum/unfamiliar
territory." Still, the most qualified programmer bid at $25/hour, lives
across town and majored in the primary subject!

In my opinion, the majority of the code is "easy." It handles data, or easy
mathmatical calculations - and includes but is not limited to "rocket
science
(probably enough code to get a rocket into orbit and back)."

I will *guess* that tough code would be video game engines, voice
recognition
or maybe drivers for the new wizbang hardware. Maybe software that
graphically designs material products.

Code that saves "processing time" or is more "efficient" can now be offset
with a few extra dollars spent on good hardware. By mandating comments in
the
code (like I do) one can employ cheaper programmers and have the code
"corrected" by a single higher priced guru (saving money) if the need
arises.

Sometimes I eat $28 steak, but I don't always enjoy it more than cheap steak
(especially if I couple an extra beer with the cheap steak). Incidently, I
just came from the "Sirloin Stockade" buffet. $7.99 all you can eat Steak
(it
was more enjoyable than some of the higher priced steak houses I have been
to
- and I am in Texas!)
           I also find that I end up slapping on the same A1 sauce!

*How* does your code work "better enough" to offset it's own price on the
organizations' bottom line?
What does your "steak code" do, Grant?
What does it look like?
How is your code 3 times better?
How are you better?
Did you code in 1/3 the time, or is it code that does something "hard" (like
maybe voice recognition)?
What can you code that the cheaper programmer can't?
Can you show us some "steak code" Grant?

Drew
_______________________________________________
Ale mailing list
Ale at ale.org
http://www.ale.org/mailman/listinfo/ale

_______________________________________________
Ale mailing list
Ale at ale.org
http://www.ale.org/mailman/listinfo/ale