Read about the birth of UNIX in the early 1970s and you see a closely linked
effect on open source software (OSS) as we know it today. Trace the OSS concept
back further through IT history and you find MIT and Bell Labs. That makes sense
as well. But what did GE, IBM, DEC and the founders of the Lakeside Programming
Group have to do with OSS? It's less clear but the outline is strong. Through
the 1970s, what we think of as OSS today was just the way software was done.
In the beginning, software development was dictated by the technology-not culture
or philosophy. Before 1960, technology dictated that computing would be centralized.
I am not talking 'centralized' as in Oracle's single-instance vision of running
an enterprise today; I am talking 'centralized' as in U.S. government economists
predicted there might never be the need for more than a dozen computers. That
prediction was of course based on what computers could do at that time and how
much it cost to build them. It just didn't make sense that people would use
vacuum tubes to replace mechanical calculators.
By 1970, when secretaries were starting to get their own Electronic typewriters
and you could find a growing number of single-purpose minicomputers stuck in
corners of research labs and on factory floors, analysts were admitting that
maybe GM, Ford, McDonnell Douglas and a few government agencies and universities
might actually be able to use their own electronic data processing (EDP) machines.
At that point, economists were predicting a few tens of thousands of discrete
EDP systems-worldwide. Other than that, when it came to business and academic
computing at least, the rest of us could get by with a couple of dozen computer
centers in the United States. All of us would share our computer center's power
in time slices, probably with plenty of time left over to oil the ¾-inch tape
drive mechanisms. One of those potential time-sharing centers was at MIT in
Cambridge, MA. Working with grants and in-kind support from GE and Bell Labs,
in 1965 MIT launched a follow on to its initial time-sharing software studies.
The new project went by the acronym Multics (for Multiplexed Information and
Computing Service). Multics introduced-or perfected commercially-features such
as a segmented ring memory with an integrated file system; a virtual memory
derived from the breakthrough UK Atlas system; operating software mostly written
in a high-level language, PL/1, for the first time; a shared memory multiprocessor;
multi-language support (including support for BASIC, which had been developed
on another time-sharing system at Dartmouth), a relational database that debuted
before Oracle's; and security, on-line reconfiguration, and software engineering
features that took many years to permeate the rest of the industry. Multics
debuted commercially in 1972 and the last system was decommissioned in 2000.