Existence Precedes Essence - Meaning of the Stored-Program Concept - History of Computing: Learning from the Past - page 173

development of the idea of hierarchical ideas of machine programming, such as
microprogramming with so-called firmware. Bromley saw the development of micro-
programming by Maurice Wilkes in 1952 as the end of the development of the stored
program concept. The storage of data and instructions in the same format occurs in
the middles stages of Bromley's story, distinct from other developments, such as
using separate parts of the machine for storage and processing and the development of
compilers. [11] Like Randell, Bromley saw the stored program concept as more than
mere hardware and more as an attitude towards machine use and design.
Now we come to the other inspiration of my paper: a discussion of the SSEC by
French historian Georges Ifrah. In his Universal History of Computing, Ifrah recog-
nizes the SSEC's stored-program nature, however he declares it to be the first “near-
computer” in history. Ifrah gives two reasons to deny it full computer status. First its
use of mixed electronic and relay technology that he argues was an anachronism and
more importantly that the SSEC embodies “inconsistencies of logic.” Ifrah argues that
the designers of the SSEC lacked a clear theoretical understanding and vision for its
operation and so made it unnecessarily complex. Ifrah gives von Neumann and the
draft report credit for synthesizing the elements of the computer into a complete the-
ory, while maintaining that in the history of technology there is neither a singular first
inventor nor invention but a continuous series. [12]
Ifrah's suggestion that the SSEC design was unnecessarily complex has at least
some justification. The machine used 12 500 expensive and unreliable vacuum tubes,
whereas later machines would obtain more functionality with far fewer tubes. [7][8]
Operators of the SSEC have had mixed opinions. John Backus, father of FORTRAN
and an early SSEC programmer would later state that “I think it's an extreme stretch
to consider it the first `stored program' computer.” [13] However A. Wayne Brooke,
the IBM engineer responsible for maintaining the SSEC, felt that quite the contrary
the SSEC did everything a stored-program machine did just in different ways. [14]
Brooke spent a great deal of time in the early 1980s drafting a description of the
SSEC and arguing for its status as first stored-program computer. The drafts of this
unpublished account reveal many interesting aspects to the machine, including the
creation of a special modification to facilitate conditional branch. The SSEC did not
have a specific operation for branching and worked instead by arithmetic manipula-
tion of the instructions in the memory. The modification Brooke described allowed
the branch to a subroutine to be carried out in a more automatic and straightforward
manner. [15] This suggests that the concepts and means of control of the SSEC were
in flux after its completion.
Another aspect of the SSEC that suggests its “near computer” status is that it lacked
single fixed physical configuration and total control of capabilities by the program.
Certain aspects such as access to relays, tape drives and the like were determined by
plugging. [8] While relatively minor compared to the extensive rewiring required in the
original ENIAC, it suggests the lack of commitment to program control.
3 Essence of the Stored Program Concept
While I remain somewhat skeptical of Ifrah's broad strokes analysis I found its bold-
ness inspiring. In honour of Ifrah's suggestion I take the title of this paper from a