Re: WSJ article on software liability

From: Traveler (traveler_at_nospam.com)
Date: 02/28/05 

Date: Mon, 28 Feb 2005 07:29:27 -0500

In article <slrnd25l9u.1aij.willem@toad.stack.nl>, Willem
<willem@stack.nl> wrote:

>Traveler wrote:
>) In article <slrnd24oqf.2vph.willem@toad.stack.nl>, Willem
>)<willem@stack.nl> wrote:
>)>Another unfounded assertion. A simple metric to test this would be the
>)>size and complexity of a software emulator of a chip.
>)>
>)>As you may know, software emulation is used in the design and testing of
>)>new hardware.
>)
>) So? What is your point? Does that make hardware any less complex? Most
>) hardware systems usually involve many ICs interacting with one another
>) in very complex ways. Besides, you are making my point for me without
>) realizing it. A software emulation of hardware is a lot more stable
>) than most other types of software. Why? Because the software emulation
>) follows the synchronous, signal-driven model of hardware. That's why.
>) Most conventional software don't. COSA aims to change all that.
>
>No. I'm making my own point.

Your point is worthless.

> Software emulation of hardware is so much
>more stable because it is so much more *simple* than other software.

Nonsense. A computer system consists of many integrated chips that
communicate with each other in highly complex ways. Yet the system
remains very stable throughout its life. It is stable because of the
synchronous, deterministic nature of hardware, your feeble
protestations to the contrary notwithstanding.

>Do you have an actual argument supporting your assertion that hardware
>complexity is even in the same ballpark as software ?

It does not have to be in the same ballpark. Algorithmic software is
unstable because it is algorithmic, period. You take an algorithmic
system of comparable complexity to a hardware system and the
algorithmic system will be less reliable by orders of magnitude. This
is regardless of the amount of testing that goes into them.

>)> This is partly because hardware is so much simple
>)>and therefore easier to test.
>)
>) Well, if it is simpler to test (which I don't deny), why test it so
>) much? A few test runs should do the trick.
>
>Because when hardware fails you have to recall all your chips, and rework
>them. Hardware failure is so much more expensive than software failure.

So you don't think it's expensive to lose an airplane full of people
or a half a billion dollar martian probe because of software failure?
Do you know how expensive it is to recall millions of cars because
there is something wrong with the embedded software in the cruise
control system?

>)>I refuted your reasons. Your assumption remains unfounded.
>)
>) Yeah, right. In your own mind, that is.
>
>Your attitude does not help your cause.

What cause? This is not MY cause. Software reliability and safety are
extremely important to the entire world. Your only interest is to show
off how intelligent and knowledgeable you are. My interest is to solve
a critical problem that is hurting a lot of people. It is your
attitude that sucks. Big time. You are a know-it-all nerd. I've seen a
lot of nerds like you in my career. You are a dime a dozen. You love
to shoot from the hip at the slightest opportunity not realizing that
the target is out of your range. You're fucking pathetic. Fuck you.

Louis Savain

The Silver Bullet: Why Software Is Bad and What We Can Do to Fix it
http://users.adelphia.net/~lilavois/Cosas/Reliability.htm

Loading