I will explain why security didn’t get included in the beginning, and what it would take to add it now.

Moving most of the computers to a secure OS is not more complicated than the changes made to utilize the new communications systems.

But, we don’t have secure systems today for two reasons.

  1. We started writing OSs in about 1960. Shared computers were non-existent; no one had communications; most of the hardware was not designed with the features needed by the software to build a secure OS. Most of today’s OSs are just extensions of this early architecture.
  2. By the late 1960s there were commercial systems offering shared access to computers, and they experienced the first hacking. These shared systems lasted about 20 years, before they were replaced by mini computers. Minis were only shared locally so hacking was not a problem. Systems available over a communications system like airline reservations or banking were as secure as possible with the standard OSs, and the costs incurred by the lack of security were considered normal business operation expenses.

About ten years ago hacking of systems became more pervasive, but still the costs to corporations were manageable. We are finally at the point where it is clear that a secure system is worth the trouble to install.

The good news is that projects started developing secure OSs in the 1970s. Much of this original work was done at Tymshare, for obvious reasons; they were operating shared computers over a world-wide communications network. Although various architectures were tried, the one that has consistently been the most successful at security is known as capabilities. Various groups, beginning in the 1980s, have further developed, enhanced and tested this technology, and at this time reliable designs are well known in this community.

Once you decide on a project a system can be implemented in 5-10 man years over a period of about 2 years. As Fred Brooks pointed out in The Mythical Man Month back in the 1960s, it is the design that takes all the time. If you have that right, the programming is straight forward.

The modern systems when completed will require little or no changes to the vast majority of applications, while being easier to maintain. The critical part of the new OS is the kernel which is much smaller than any of the systems popular today: Windows, iOS, Unix or Linux. That OS will be replaced with the new one. This was done in about 1978 at Tymshare. The company ran a credit card accounting application on the 370/67 for several years with no changes to the application as it had been running on the VM/CMS operating system. Systems today are even more developed and able to handle this challenge.