We will adopt an approach suggested by the Trace center as a framework
for cross-modal display optimized to each client (user profile). We
will make a reeasonable assumption -- at least for future education and
training -- that all the information is web-based and described by the
emerging W3C document object model (DOM). We will use "active pages"
gotten today by capturing all events and properties from a web-page in
version 4 browsers. As dynamic HTML gains acceptance and browsers
improve, we expect this support to improve. However it is already good
enough today for the limited support needed for common authoring tools
for educational material. We now note that this model is already that
used by TangoInteractive to support shared dynamic HTML in the form of
messages passed between clients sharing displays and expressing content
in the W3C DOM. Although one usually considers TangoInteractive as
linking different machines one can use this machanism internally to a
given browser to allow dynamic interpretation of documents. The events
received by Tango on each client will be passed through a software
agent that will use a user profile to map modalities optimally for the
user. Here the NeatTools custom support for blind and
motor-impaired users resides. This approach can be applied in three
useful ways. First, one can map modalities internally to a single
client. Secondly, note that todays reducing cost of PC's implies that
one can consider it quite practical for a given user to access multiple
PC's and their displays concurrently. TangoInteractive can then allow
users to view simultaneously different modality mapping of a given page
and choose that most appropriate.  This mode also elegantly supports
comparitive evaluation of particular display strageies. Finally one can
naturally use TangoInteractive with multiple PC's and multiple users so
that a set of collaborating users (teachers and students in our
education example) can choose very different displays -- each optimized by the SoftBot to a different profile..
TangoInteractive's WebWisdom substem already does this in a simple case
when it allows each client to separately choose the resolution of a
shared image so as to make best use of network bandwidth.