As a retired engineer living in South Florida and a successful
international career focused on highly reliable software systems I am
depressed by the direction of computer science. In 1968 when I graduated
competition in computer architecture was not only possible it was the
only way to go. Nothing was off the shelf. Consequently, innovation was
rampant as individual companies rolled-their-own computer hardware. For
example, Plessey Telecommunication in the UK designed a first
fault-tolerant, Capability-Based Multiprocessor. The PP250 was an
innovation and remains so to this day as the only pure object-oriented,
capability-based computer. The goals were scalability throughout a
growing network of users, software reliability guaranteed at run time,
and deep and detailed digital security.
Since 1972, the PP250 design team made many presentations that led to a
key debate at SOSP'77 with Butler Lampson and other experts on the
future of Capability-Based Addressing. The mainstream had no need for
reliable software with guarantees built into the hardware. The
telecommunication industry had bigger problems beyond the
batch-processing. The IBM 360 created a mainstream obsession with
monolithic, centralized operating systems, page-based virtual memory,
and shared off-line, static compilations. Global telecommunication is
very different. It defines a living global network of living
applications as varied as found in nature. As the smartphone drove the
internet to expand into mobile communications, computing requirements
changed. Global networking exposed the flaws in the shared, centralized,
batch-processing architecture.
Malware and hacking are the direct consequence of the misapplication of
the batch processing architecture copied by the microprocessor from the
1960s mainframe. The threats that drove capability limited computers in
the 1970s remains true. Networked software is exposed to remote attacks
not possible when mainframes were isolated in locked rooms. While
software remains flexible and rapidly improves, the hardware is
inflexible and microprocessors remain stuck in the past. The static,
stand-alone, monolithic compilations using a hub and spoke
communications all share privileged services. This dictatorial
architecture is medieval and like the middle ages malware and hackers
pillage the kingdom robbing the software citizens of independence and
prosperity.
The solution is unchanged since capability-based computers were first
invented. It is the same problem with the same answered found in nature
and expressed by the dual-sided science of the Church-Turing Thesis, but
that is another story.
international career focused on highly reliable software systems I am
depressed by the direction of computer science. In 1968 when I graduated
competition in computer architecture was not only possible it was the
only way to go. Nothing was off the shelf. Consequently, innovation was
rampant as individual companies rolled-their-own computer hardware. For
example, Plessey Telecommunication in the UK designed a first
fault-tolerant, Capability-Based Multiprocessor. The PP250 was an
innovation and remains so to this day as the only pure object-oriented,
capability-based computer. The goals were scalability throughout a
growing network of users, software reliability guaranteed at run time,
and deep and detailed digital security.
Since 1972, the PP250 design team made many presentations that led to a
key debate at SOSP'77 with Butler Lampson and other experts on the
future of Capability-Based Addressing. The mainstream had no need for
reliable software with guarantees built into the hardware. The
telecommunication industry had bigger problems beyond the
batch-processing. The IBM 360 created a mainstream obsession with
monolithic, centralized operating systems, page-based virtual memory,
and shared off-line, static compilations. Global telecommunication is
very different. It defines a living global network of living
applications as varied as found in nature. As the smartphone drove the
internet to expand into mobile communications, computing requirements
changed. Global networking exposed the flaws in the shared, centralized,
batch-processing architecture.
Malware and hacking are the direct consequence of the misapplication of
the batch processing architecture copied by the microprocessor from the
1960s mainframe. The threats that drove capability limited computers in
the 1970s remains true. Networked software is exposed to remote attacks
not possible when mainframes were isolated in locked rooms. While
software remains flexible and rapidly improves, the hardware is
inflexible and microprocessors remain stuck in the past. The static,
stand-alone, monolithic compilations using a hub and spoke
communications all share privileged services. This dictatorial
architecture is medieval and like the middle ages malware and hackers
pillage the kingdom robbing the software citizens of independence and
prosperity.
The solution is unchanged since capability-based computers were first
invented. It is the same problem with the same answered found in nature
and expressed by the dual-sided science of the Church-Turing Thesis, but
that is another story.
Comments