[Axiom-developer] SAGE, Axiom, and usage

[C Y]

--- "Page, Bill" <address@hidden> wrote:

> The reason why I asked: "Are Axiom developers and users really
> motivated to use this sort of thing?" is because no one has
> shown very much motivation so far. :( I am beginning to
> seriously wonder if investing more time in better tools is
> really worth the effort. It seems that we are still as stuck
> in the old situation where (almost) the only people who use
> the tools are the people who built them, e.g. MathAction and
> ALLPROSE. And I guess what I really want to do is just that...
> which is probably the reason I have been feeling so conflicted
> lately.

There has been some usage Bill - I really like what MathAction was able
to do with that Emacs file I worked on a while back.  The primary
reason the units work isn't on MathAction yet is there isn't any code
to speak of yet.  There is other interesting code on MathAction -
Guess, for example.

I guess my first question is "What usage had you envisioned?"  A
broader audience for Axiom is going to take many years.  Most people
want to use the tool rather than work on it, and we are nowhere near
that stage as a project.  Axiom can be used to do other work, but it
doesn't look or feel like a modern program and that's going to have an
impact.  I think SAGE is getting a fair bit of attention because it
looks more "familiar" to people.  When I look at this: 
http://modular.math.washington.edu/sage/screen_shots/.html/37a.html or
http://modular.math.washington.edu/sage/screen_shots/.html/ec-soya.html
my first thought is "that looks a lot like Maple."  Now it probably
isn't all that similar when you get down to it, but that's my first
thought.  I think the SAGE project looks like it is doing some very
interesting work, but it has not yet withstood the test of time.

> I am amazed (and pleased) to see the amount of effort going
> into the development of Sage. I am disappointed at the rate
> of progress in Axiom.

At the bottom of the SAGE website is this:
"Work on SAGE is partially supported by the National Science Foundation
under Grant No. 0555776."

That helps, and being centered at a university also helps - after all,
in one sense Axiom has very few links to academia in terms of
sponsering/patronage.  Most of us have something else as our primary
responsibility, so there is less intense, focused effort on Axiom. 
Also, SAGE doesn't have to go through the process of bringing code from
thirty years ago up to modern standards, which is a lot of work and is
less "sexy".

> I worry about the delay in the open source status of Aldor.

I think this is probably the biggest single holdup.  We don't have much
serious work on SPAD code because we all think and hope it will
eventually have to become Aldor code.  We don't want to start the
SPAD->Aldor (plus actual literate documents with academic research and
theory background as opposed to source code in a pamphlet file) because
if for whatever reason Aldor DOESN'T become open source the way we need
it to be then we are back to square one, and all the Aldor based work
becomes marginalized.  Then we have to turn the SPAD compiler into
Aldor and add the improvements we need.  Statements made earlier on the
list seem to indicate that such a project might not be workable, which
would leave Axiom in a rather precarious position.  All of this
uncertainty does not help Axiom in becoming a major "player" in either
open source or academia.

> I am disappointed that in the Sage
> project I see some things that Axiom does very well being
> re-invented badly. I am thinking about the best strategy for
> the survival of a legacy project like Axiom when something
> old made new (Sage) can attract so much interest.

It's easier to start a new project than understand an old one.  At
least in the beginning.

> I am tempted to suggest that hooking Axiom to the Sage bandwagon
> as soon as possible might be the best idea.

I don't think so, personally.  SAGE is new, and has some impressive
pieces, but it's long term success is not decided.  Axiom has the hard
parts - the core mathmatical logic, the type system, the focus on
combining research and code.  We don't have the sexy GUIs and notebook
interfaces, but they will come in time.

> But that makes
> Axiom seem subordinate when I really believe that it does
> many things better even though it was designed nearly 30 years
> earlier. Which rather leads off-topic, I have to admit...

It's a long standarding problem in computers generally that newer and
prettier is always seen as better.  That's I think been due largely to
the increases in physical hardware power that people have seen, and the
benefits that provides in terms of usability.  The core designs of
software, algorithms, and systems are more timeless than that.  TeX has
been and continues to be the best typesetting solution available for
scientific papers, decades after it was written.  Despite this, it is
not as widely used as one would expect.  I think that rather than say
Axiom was designed 30 years earlier, we should say that it takes
decades to make the kind of program Axiom is and wants to become.  This
isn't Yet Another Editor - this is rocket science, the TeX of
mathematics software, if we do it right.  The space program didn't look
very sexy either early on (except a) people knew the goal was the moon
and b) rockets blowing up is more interesting than software crashes)
but everybody paid attention when it starting doing really amazing
stuff.  I think Axiom is like that - we're in the conference room
working on blackboards right now.  Drawing blueprints and looking at
the technologies we want to use, what we have available, and how it all
works together.  We'll start getting press when we start doing stuff no
one else can do, and everyone WANTS to do.  Things like being able to
answer the question "why should I trust this result" by saying "here is
a proof generated for this result that was checked in ACL2/COQ/Isabelle
and you can also either check by hand or in any of a dozen other
industrial strength proof checkers."  In the end, if we do what we are
trying to do, the question people will be asking is why would you use
anything BUT Axiom.  The problem is, that's never easy.  

Cheers,
CY

__________________________________________________
Do You Yahoo!?
Tired of spam?  Yahoo! Mail has the best spam protection around 
http://mail.yahoo.com