Death by analogy Part 1 (was Re: is ada dead?)

["Michael P. Card" <michael.p.card@lmco.com>]

[-- Attachment #1: Type: text/plain, Size: 3733 bytes --]

Hello CLA-

This post reminds me of a post I read a couple of years ago by Bertrand Meyer regarding the use of analogies and the erroneous assumptions that
often underlie them.

In this case, the hypothetical wire cutters offer a 2% productivity boost and cost 100X as much as a competitive set of wire cutters. The obvious
implication is that Ada is similar, i.e. single-digit productivity boost for 100X the cost.

When I read things like this, and many of the other posts in this thread, I wonder if the "Ada offers no benefits worthy of its cost/risk of
vendors going out of business/etc etc" viewpoints are the result of the types of work being done by their posters.

In the sort of work I do, I find C/C++ to be very backward by comparison to Ada, especially in the areas of type safety and concurrent programming
for real-time.

Rather than saying Ada is like a pair of wire cutters that offer a 2% productivity boost for 100X the cost, I would say the situation is more like
a case last year where I did some of my own landscaping. The tools at my disposal were a shovel, a mattock and a 5-gallon pail, i.e. the stuff I
had in my garage. The job took me a few days to complete, but I didn't have to spend a nickel on tools.

Now, I could have gone to NationsRent and rented a Bobcat (mini-backhoe) for a few hundred dollars and done the job in less than a half hour, but
to me it wasn't worth the cost. Does that mean that a Bobcat is a tool that offers a mediocre performance boost for thousands of times the cost of
a shovel?

It really depends on whether you are doing some minor home landscaping or building a highway.

The posts I have read here imply to me that many of the Java/C/C++ devotees are building small-scale projects where the interaction of 1, 2 or 3
programmers is sufficient for the job. At my place of employment, interaction is required between tens of programmers at least as they develop
thousands and thousands of lines of code, and from my experience Ada is *VASTLY* superior for such jobs. In these environments, the cost/benefit
of using Ada is **NOTHING** like a 2% productivity boost for 100X the cost.

Indeed, using C++ for these kinds of jobs is more like building a highway with a mattock and shovel, all the while praising oneself for being
frugal by avoiding the cost of the backhoe, to put forth a counter-analogy ;-) I have been on big projects done in C++ and this experience has
only reinforced my perceptions about the benefits of Ada.

Just because I wouldn't buy my own Bobcat to tinker around my yard doesn't mean a Bobcat isn't a great tool. Likewise, saying that Ada offers
minimal benefits for small-scale/Web-applet type jobs does not therefore imply that it offers no worthwhile benefits to anyone.

- Mike

Al Christians wrote:

> Jerry Petrey wrote:
> >
> > This is certainly not nonsense.  But don't feel bad.  Many people in
> > the industry are unable to understand the true cost of developing
> > software and only look at the up-front coding costs, tool costs, etc.  > That is one of the main reasons most software is over budget and of
> > poor quality or not even ever delivered.
> >
>
> Suppose you are an electrician and you hear about a new kind of
> wirecutter.  There are studies that say this wirecutter improves
> average productivity by 2%.  If you do the math, you can figure that
> this is worth $2,000 to you over the expected 5 year life of the
> wirecutters.  You go to the store and see $1,295 wirecutter on sale
> next to all the others at $11.  Which pair do you buy?  Which toolmaker
> has biggest market share and good cash flow to finance ways to improve
> their product?
>
> For $1,284 most can think up a reason why they are not average.
>
> Al

[-- Attachment #2: Card for Michael P. Card --]
[-- Type: text/x-vcard, Size: 344 bytes --]

begin:vcard 
n:Card;Michael
tel;fax:315-456-0441
tel;work:315-456-3022
x-mozilla-html:TRUE
org:Lockheed Martin ;Ocean, Radar, and Sensor Systems
version:2.1
email;internet:michael.p.card@lmco.com
title:Principal Software Engineer
adr;quoted-printable:;;Electronics Park=0D=0ABuilding 6, Room 201;Syracuse;NY;13221;USA
fn:Michael Card
end:vcard