MathGroup Archive 2010

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

Search the Archive

Re: On the foundation of Mathematica, was Re: Foo /: Plus[b_Foo] := b

On 10 Dec 2010, at 08:30, kj wrote:

>> In other languages, there is always an "official Language reference
>> manual" that would contain in it _everything_ about the language.
> The important thing to know is that this omission is not an isolated
> error, but rather a matter of policy.  The entire Mathematica
> documentation is an exercise in revealing the bare minimum that
> they feel they can get away with.  WRI has always flouted the
> standards of documentation you allude to.
> This policy has long been inexplicable to me, to put it mildly.
> I hope that some day someone with first-hand knowledge of what this
> official policy is will divulge it to the world (WikiLeaks, anyone?).
> If that ever happens, I'd be surprised if it didn't turn out to be
> a huge embarrassment for WRI.  It's hard for me to imagine any
> plausible defense for the grossly incomplete documentation that
> WRI churns out, I don't care how pretty it looks.
> In the meantime, WRI staff find it funny that their customers waste
> hours figuring out bugs caused by undocumented details such as this
> one.  Ha-ha.  Hil-a-rious.
> ~kj

My impression is rather different. Different people at Wolfram work on different aspects of the program. I very much doubt that there is anyone there today who understands the "whole thing". So what happens is that individual Wolfram's employees are experts in certain aspects of Mathematica but possess much more limited knowledge of others. I can easily imagine that there some aspects which are not completely understood by anyone. This is certainty true in some areas of mathematics and it seems to me that Mathematica is complex enough for the same situation to occur.

My personal experience suggests that when a user is able to trace and contact the person at WRI who is really expert at those things that the user is interested in, there is almost no limit to how much information may be divulged. Certainly, this includes full details of algorithms and their implementations and, I am speaking from experience, even source code - or at least parts of source code that are written in Mathematica.  This has been my experience and I do not have any reason to think that it is exceptional.

This suggests to me that there is probably no "official policy" of the kind that is conjectured above, or if there is one, it leaves a great deal to the judgement and choice of individual WRI employees. The fact of life is, however, that only some of them seem to welcome this sort of approaches from users and an even smaller subset can be reached through this forum. And, of course, even Wolfram employees will be usually reluctant to answer questions about those aspects of Mathematica they themselves do not fully understand. It seems to me that thing like the details of the relationship between Attributes and pattern matching may well belong to this latter category, although I think I can think of  one or two persons who almost certainly understand this topic quite thoroughly.

I think if my speculations are correct than this also explains the state of the documentation. In particular, it is obviously going to be easier for people to document fully things that they are themselves currently working on than things developed by someone else 20 years ago that are often only peripherally relevant to the newer developments.


  • Prev by Date: Re: Replacement Rule with Sqrt in denominator. Also Bug in Series
  • Next by Date: Re: Replacement Rule with Sqrt in denominator
  • Previous by thread: Re: On the foundation of Mathematica, was Re: Foo /: Plus[b_Foo] := b
  • Next by thread: few things would be nice to have in next version of Mathematica