Re: Languages don't matter. A mathematical refutation

[brbarkstrom@gmail.com]

> > > To change the subject a bit, is there any possibility of using
> > > the probabilistic approach to estimating the long-term cost of
> > > maintenance given a history of errors with a particular language?
> > 
> > since errors made in C are different from
> > errors made in Ada. You change the language, you get *other* errors, no
> > statistics across to compare.
> > 
> > Or taking another path, considering "probabilities" of same error made in C
> > and Ada. Again,
> > ...
> > Apples and oranges all the way.
> 
> You can create a criteria valid for any error or language
> 
> A) Severity
>  1) Data lost
>  2) Crash
>  3) ...
>  4) Inconvenience
> B) Frequency
>  1) each day
>  2) One a week
>  3)...
> C) Time needed to fix it (hours)
> 
> Score= ((Severity * K1) * (Frequency * K2)) + Time * K3
> 
> Or whatever formula you want to come with. Calculate the errors of several similar projects and you have an statistic.
> 
> Obviously the trick is in "similar projects". There are many factors, experience of programmers, experience of management, the domain of the project, the difficulty of the project.....
> 
> In fact there are a lot of  off-the-shelf packages (accounting, antivirus...), probably many of them programmed in the same language, and the quality ranges from crap to very good. That's a good clue: Language is not the main factor.

Both of these responses are interesting, although I think there's a
need for deeper exploration.

For example, consider a stream whose bed contains flecks of gold.  A given
area of undisturbed bed certainly has a fixed set of flecks.  Thus, if we
put a screen with fine enough mesh across the stream bed, each mesh element
either contains a gold fleck or not, as Dr. Kazakov postulates about errors.
Now, we consider a mining operation that scoops up stream bed and puts the
material into a sluice that collects the flecks.  The mining company wants
to know which of two sluice designs is more efficient.  There are many random
variables that could influence the fleck fraction: volume of scoop, area
of stream bed covered, as well as the sluice design.  The question is whether
the variations in design are detectable in the presence of the other 
variabilities.  I expect that our colleagues in electrical engineering
and in the newer enthusiasm for uncertainty propagation have some experience
dealing with this kind of problem.

As to whether language is the main factor in determining quality, I'm curious
as to whether there's empirical evidence about its relative importance in
the cost.  jm.ta's response suggests the issue is cut and dried.  Perhaps
he's correct - and language choice is a negligible factor.  However, a
quantitative answer needs more evidence.

Bruce B.