MathGroup Archive 2009

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: "Do What I Mean" - a suggestion for improving

  • To: mathgroup at
  • Subject: [mg97135] Re: "Do What I Mean" - a suggestion for improving
  • From: AES <siegman at>
  • Date: Fri, 6 Mar 2009 04:24:31 -0500 (EST)
  • Organization: Stanford University
  • References: <goo7l7$shc$>

In article <goo7l7$shc$1 at>,
 Bill Rowe <readnews at> wrote:

> I do not think it is even a good
> idea to attempt to make Mathematica accessible to users with
> minimal computer/mathematics experience/knowledge assuming this
> is even possible.

I guess we'll just have to disagree -- vehemently! -- on this one (and 
also with great sadness on my part, if this should represent Wolfram's 
anything like Wolfram's actual views or objectives).

By sheer coincidence, a few minutes after seeing the above post I read 
the following post in another newsgroup (it's a big long, but just skim 
down to the end):

This thread reminded me of one of my favorite
published papers (because of its sheer readability) and I could not
resist bringing it to the attention of others, old and dated though it
may be.  Scrounge through the stacks of your local engineering

Test yourself: how much do you know about international 
communications? [International numbering systems]

Robrock, A. (Italtel, Milan)
IEEE Communications Magazine, December 1989
Volume: 27,  Issue: 12


We like to think of international telephone communications as
`transparent', the successful outcome of 100 years of technical
progress and standards setting, but the author shows us that it is
not. The user still has to be something of an expert to understand how
to make international calls, and there are chaotically differing
numbering systems for telephony, telex, and electronic mail. We should
be reminded that usability of services, not just their usefulness, is
a critical component of communications. Simplicity, consistency, and
rationality of service features and the `human interface' that allows
users to invoke them should be a high priority for communications
engineers as they work toward the integrated services networks of the

Besides the "chaotically differing" phraseology, it's the final two 
sentences that catch my eye.  Should Mathematica interface designers 
maybe be reminded that

   "it's the _usability_ of software, not just its _usefulness_,
   that's a critical component of software interface design"

and even better

   "Simplicity, consistency, and rationality of software features 
    ** and the `human interface' that allows users to invoke them **
    should be a high priority for software designers as they work toward
    the integrated services networks -- sorry, integrated software 
    packages -- of the future."

Interesting -- "_integrated_ software packages?" -- don't I recall that 
that's one of the big selling points for Mathematica? (although one that 
I personally believe can really only be effectively achieved -- for 
software that is, not necessarily for networks -- using a much more 
modular approach.

> There are a great many things in mathematics that work in
> specialized cases. For example, a user with little experience in
> mathematics likely would expect Sqrt[x^2] to simplify to x. But
> that transformation is only valid when x is real and positive.
> If Mathematica were to automatically do this simplification (or
> many others of a similar nature) it would not be an adequate
> tool for me or many other users since it would be creating
> erroneous output. Worse, even for those users where this
> happened to be the correct output, the issue gets hidden and
> they would learn to trust Mathematica only to lose trust when
> things were more complex.
> The point is mathematics is complex. A tool designed to
> implement mathematics can hardly be less complex. Attempts to
> reduce the complexity invariably mean some aspects (typically
> special cases) of the mathematics are being ignored or hidden.
> Ignoring or hiding such special cases limits the usefulness of Mathematica.

Don't really disagree on the facts here -- just the operational 

1)  Are you really saying that the whole series of superb hp calculators 
out of which I got so many useful scientific and engineering results in 
earlier years -- and which, incidentally and painlessly, also gave me at 
least an introduction to the concepts of Reverse Polish notation and 
stacks as an aside -- should not have had a "Sqrt[x]" key?

[And incidentally:  Would you not like to see Mathematica be as widely 
used, and useful, as were those superb tools?]

2)  I'm an engineer and physicist; other potential Mathematica users 
might be from innumerable other practical fields (econ, stat, business, 
etc etc).  We know some math; varying amounts for different fields and 
levels within fields.  We know there are complexities in math that we 
may not understand.  But we also have the protection that when we 
calculate results using some, we can (and do!) look at them and apply 
"physically reasonable" criteria (or "realistic results" in other 
fields) as part of our criteria.  We don't denigrate rigor, or fail to 
take care about the possibility of unanticipated special cases.  But we 
have _other_ tools that mathematicians don't have, to help us cope with 
the possibility of those.

  • Prev by Date: RE: Epilog doesn't work with loglinearplot
  • Next by Date: Mathematica and some General Comments
  • Previous by thread: Re: "Do What I Mean" - a suggestion for improving
  • Next by thread: Re: Re: "Do What I Mean" - a suggestion for improving