Many of today's operating systems are bloated, overly complex and insecure. Security shouldn't be an afterthought: it should be an integral part of any high-quality piece of software. Typically, we find that mainstream operating systems employ design decisions that may have been wise 30 years ago, but today are no longer appropriate.
Reliability of software is important to me as a developer; there is simply no excuse for systems software that has residual defects going unnoticed. Traditional monolithic architectures are no longer acceptable these days; we need simplicity for the Cloud, not increasing bug-ridden complexity.
The Orchid Project is a radical step toward achieving this goal. The aim is provide security, reliability and performance by employing a 'Capability' architecture. System resources are treated as 'objects' that can be accessed or modified by the Kernel by issuing 'tickets' known as Capabilities. The use of Capabilities provides security at the Kernel level, not as an afterthought.
Another proposed feature of Orchid is its support of Cloud or Grid based computing. The is to be no concept of a single 'machine' in Orchid; we have instead an internetwork based, self-healing operating system that provides the 'effect' of operating like a single, powerful virtual machine.
To find out more about Orchid and view it's code, take a look at:
Gavin R Brewer.