LANL software contest

[Cameron Laird]

In article <38832BE8.81EFC77D at mayo.edu>,
Bob Techentin  <techentin.robert at mayo.edu> wrote:
>Harald Kirsch wrote:
>
>[snip]
>
>> But as mentioned in another post, I just don't like it that they don't
>> say: ``we just like Python more than Tcl'' but ``Tcl is easier to
>> learn and read than ..., but not as easy as Python.''
>> 
>> I doubt they have hard facts to prove that, but if the contest will be
>> a success, everybody will later cite their statement as if it were a
>> hard fact. And a new myth will be born like `helmets reduce the danger
>> of bicycle riding' (read rec.bicycle.rides if you think this is a
>> proven fact).
>
>I'm not so sure.  When I read Guido's proposal to DARPA for developing a
>teaching environment for Python
>(http://www.python.org/doc/essays/cp4e.html), I saw the first ever (for
>me) reference to studying the "learnability" of a language.
>
>    We already have some evidence of where changes might be necessary. 
>    Prof. Randy Pausch at Carnegie Mellon University (see below) has
>    conducted some usability studies of Python within their limited 
>    problem domain. Their users seemed most confused by the case 
>    sensitivity of Python's variable names and by the truncation of 
>    integer division. 
>
>I would argue (without any actual physical evidence) that you could
>teach Tcl even more easily to beginners.  Tcl's dead-simple syntax
>should be easy to teach to non-programmers, but has anybody actually
>studied how hard/easy it is to learn?
			.
			.
			.
Briefly, no.

Imagine:  programmers go around arguing about various
qualities pertinent in software engineering (reliability,
efficiency, ...), and very, very few of the claims extant
in our domain ("OO is more scalable"; "HLL coding is
faster"; ...) have been tested (let alone re-tested) with
any pretense of rigor.  There definitely is evolution in
behavior (look at magazine titles--structured programming,
objects everywhere, XML empire, ...), but it's best
modeled as fashion change.

Learnability is an even more extreme case of this.  Guido
worked in the '80s with one of the few groups ever to do
the hard (but illuminating) empirical work of learnability
experiments.

I know of nothing comparable done with Tcl.
-- 

Cameron Laird <claird at NeoSoft.com>
Business:  http://www.Phaseit.net
Personal:  http://starbase.neosoft.com/~claird/home.html