The Goals of

The Distributed Statistical Computing '99 conference held in Vienna was the first occassion for those involved in to meet face to face and discuss the project and its goals. The following hopes to clarify the nature of the project and explains each members (current) indvidual perspective.

Douglas Bates

For me the Omega project is an opportunity to experiment with distributed algorithms for complex statistical calculations, such as those involved in fitting linear and nonlinear mixed-effects models. It also allows easy access to SQL databases and other common data storage protocols. The use of Java as the implementation language and the ability to incorporate compiled Java classes as top-level Omega objects means that we can move from purely interactive use to compiled classes very easily. I think this will be important in many statistical computing uses. Finally, I am enthused about developing freely-available (or ``open-source'') software. I have benefitted tremendously from the GNU tools, from the Linux kernel, from TeX and LaTeX, and from many other freely-available software components. This is an opportunity to return something to the open software community.

John Chambers

Peter Dalgaard

Robert Gentleman

Kurt Hornik

Ross Ihaka

Friedrich Leisch

Martin Maechler

Paul Murrel

Balasubramanian Narasimhan

Brian Ripley

Duncan Temple Lang

The Omega project is an opportunity to build a very flexible set of software components from the ground up which can be used in all aspects of ``computing with data''(see JMC). It will provide the components needed for different styles of statistical environments than can be created for different users and tasks such as a general statistical environment, GUIs for specific statistical methods, embedded scripting languages, etc. The form of the tools allows extensibility in many different ways not possible in other systems. This approach allows researchers to investigate new methodology, tools and perspectives such as model fitting, web based reporting, parallel and distributed algorithms, etc. based on the basic building blocks as well as introducing entirely new components. Additionally, its use of Java will allow us to explore and exploit facilities for distributing new methodology (i.e. technology transfer) in interesting ways.

Luke Tierney