WebMath: iMath - why it is important that we have open math infrastructure

[Stephen M. Hunt]

These notes are in part a response to a mail by Ron Fitzgerald.

---

Would you call the W3C a proprietary organization?  In one sense they are,
they hold defensive licenses and trademarks to their specifications.  But, I
think there are significant differences between the way they leverage that
proprietary interest and the approach of a typical vendor.  The same is true
of iMath.  In the case of the W3C it is a small group of caretaker
institutions that own the underlying IP associated with the Web.  In the
case of iMath it will be a global collection of institutions with a
commitment to math.

Another example of a cooperative group within the education sector which
derives fees through affiliation for their work is Educause.  In particular
take a look at their IMS project.  There are affiliation and other fees to
underwrite the work of IMS, but the specification is open, and implemented
by various vendors.

Now, both these examples, the W3C and IMS, have severe issues associated
with implementations.  In the case of the Web it is estimated that 25% of
the cost of Web site development is associated with browser
incompatibilities.  From day one we have seen the various vendors try and
leverage off the Web to create their own proprietary space, not just at the
level of tools, but also at the level of the underlying protocols.  IMS is
similar, vendors may claim to implement IMS, but they also implement closed
extensions that result in your data not being portable.  So, when iMath is
asked to reimplement a Blackboard course we have to implement various parts
of the course from scratch.  If Blackboard adhered only to open
specifications, then we would merely need to import the data.  The cost
instructors face to move their data out of the current Web course management
systems into open specifications is much higher than 25%.

In math, interoperability is key.  At all levels, notation, computation and
visualization.  Math is the only universal language.  iMath takes a stronger
position than the W3C on both specifications and derivative implementations.
An approach similar to Linux.  You must maintain interoperability.

The specifications iMath evolves either independently or in cooperation with
other groups are all open licensed specifications.  The principal
specification iMath provides an implementation of at the moment is MathML, a
W3C open specification.  An iMath distinctive is that it doesn't try to
leverage an open specification into a proprietary framework.  Instead we
abstract any extension as an open specification and make these available
free to any vendor to implement, subject to the interoperability condition.

Here are some reasons iMath provides an implementation of open math
specifications:
1. developing an implementation is part of the process of developing a
specification
3. our affiliates need implementations
3. we want to make math freely available for any individual

The reason our implementations at the moment require a modest fee for
institutions is that we don't have the funding to execute our mission
otherwise at this stage.  iMath seeks to operate profitably.  This
requirement ensures iMath is efficient in its use of income and that it is
conscious of market conditions.  iMath as an entity has been designed for
sustainable growth and evolution.  We look forward to a time when our
implementations will not be linked to a fee, only a license, but this
requires broad institutional buy-in to our mission.

We want to see the users of math and math educators and students to have
access to appropriate solutions.  We would like to see the math vendors
business plans evolve from featuring proprietary infrastructure to content
and services on top of open math infrastructure (Windows cf. Linux).

Why do institutions want open math infrastructure?  There are lots of
reasons, but here are some that relate to computation.  If your rocket
crashes, was it your model or the underlying implementation at fault?
Without access to the source of the implementation this can be hard to
resolve.  You can ask the vendor and they may look at the issue for you.
But, if they find a bug, they typically do not document it publicly so
others can become aware of the issue.  And, in a way this knowledge comes to
late, your rocket crashed.  Here is an example from the Web, when a protocol
bug was identified in HTTP, the next day a corporate Apache developer posted
a patch, it took Microsoft 3 months to post a patch for IIS.  A typical math
engine is a fickle beast to maintain.  While hundreds of new algorithms of
merit are developed each year within academic and research organizations, it
takes about 5 years for only the most commercially viable algorithms to find
their way into an engine.  Implementations of math algorithms can be
'published' and 'maintained' just like Web pages within the iMath framework,
by an open community of experts, not a handful of developers that usually do
not understand what the algorithm is trying to model.

The Web architecture created the opportunity of the Web.  The challenge we
have is to add open math infrastructure so that a similar benefit might
accrue to math.  Where is the math in today's Web or in online courses?  If
math pervades our world and underpins our technical age, why can't you
calculate and visualize math within popular applications?  Why is math
literacy and numeracy so poor?  If everyone could just communicate math,
wouldn't that be a good place to start?  We believe that math should be as
available as text.  That math should be a first class element of the
Internet and Web.  Today anyone can communicate math freely within the
Internet/Web using iMath Communicator (http://www.imath.net).  Soon, anyone
will be able to freely compute and visualize math.  iMath does not create
the content or algorithms, it enables them.  The math can run on any product
that implements open math infrastructure.

I know these are difficult and complex issues to resolve.  Thank-you for
your time in wrestling with them.  We value your feedback and insights.

Regards

Steve

Dr Stephen M. Hunt
Director
Internet Math Consortium
steve at imath.org
http://www.imath.org



-----------------------------------------------------------------
WebMath at mail.math.ca - WebMath Mailing List
To unsubscribe:
via Web:     http://camel.math.ca/cgi-bin/wcms/webmath.pl
via e-mail:  send message a to majordomo at mail.math.ca with
"unsubscribe webmath" in the BODY of message
List Archives: http://camel.math.ca/mail/webmath/
-----------------------------------------------------------------