A Short History of Systems Development

"If they do not have an appreciation of whence we came, I doubt
they will have an appreciation of where we should be going."
- Bryce's Law

INTRODUCTION

I always find it amusing when I tell a young person in this industry that I
worked with punch cards and plastic templates years ago. Its kind of the
same dumbfounded look I get from my kids when I tell them we used to
watch black and white television with three channels, no remote control,
and station signoffs at midnight. It has been my observation that our younger
workers do not have a sense of history; this is particularly apparent in the
systems world. If they do not have an appreciation of whence we came,
I doubt they will have an appreciation of where we should be going. Consequently,
I have assembled the following chronology of events in the hopes this will provide
some insight as to how the systems industry has evolved to its current state.

I'm sure I could turn this into a lengthy dissertation but, instead, I will try
to be brief and to the point. Further, the following will have little concern
for academic developments but rather how systems have been implemented
in practice in the corporate world.

PRE-1950'S - "SYSTEMS AND PROCEDURES"

Perhaps the biggest revelation to our younger readers regarding this period
will be that there was any form of systems prior to the advent of the
computer. In fact, "Systems and Procedures" Departments predated the
computer by several years. Such departments would be concerned with the
design of major business processes using "work measurement" and "work
simplification" techniques as derived from Industrial Engineering. Such
processes were carefully designed using grid diagrams and flowcharts. There
was great precision in the design of forms to record data, filing systems to
manage paperwork, and the use of summary reports to act as control
points in systems. For example, spreadsheets have been extensively used
for many years prior to the introduction of Lotus 1-2-3 or MS Excel. There was
also considerable attention given to human behavior during the business
process (the precursor to "ergonomics").

Systems were initially implemented by paper and pencil using ledgers, journals
(logs), indexes, and spreadsheets. We have always had some interesting filing
systems, everything from cards and folders, to storage cabinets.

Perhaps the earliest mechanical device was the ancient abacus used for simple
math (which is still used even to this day). The late 1800's saw the advent of cash
registers and adding machines as popularized by such companies as NCR in
Dayton, Ohio under John Patterson who also introduced sweeping changes in
terms of dress and business conduct. This was adopted by Thomas Watson, Sr.
who worked for many years at NCR and carried forward these practices to IBM
and the rest of the corporate world. Also, Burroughs was a major player in
the early adding machine industry.

The first typewriters were also introduced in the late 1800's which had a tremendous
effect on correspondence and order processing. This was led primarily by Remington
Arms (later to become Remington Rand).

In the early 1900's, tabulating equipment was introduced to support such things
as census counting. This was then widely adopted by corporate America. Occasionally
you will run into old-timers who can describe how they could program such machines
using plug boards. Punch card sorters were added as an adjunct to tabulating
equipment.

As a footnote, most of what IBM's Watson learned about business was from
his early days at NCR. However, he had a falling out with Patterson who fired
him. As a small bit of trivia, after Watson died, he was buried in Dayton on a
hilltop overlooking NCR headquarters, the company he couldn't conquer.

During World War II, both the U.S. military and industrial complex relied heavily
on manually implemented systems. We did it so well that many people, including
the Japanese, contend it gave the Allies a competitive edge during the war.

The lesson here, therefore, is that manually implemented systems have been with
us long before the computer and are still with us today. To give you a sense of
history in this regard, consider one of our more popular Bryce's Laws:

"The first on-line, real-time, interactive, data base system was double-entry
bookkeeping which was developed by the merchants of Venice in 1200 A.D."

One major development in this area was the work of Leslie "Les" Matthies, the
legendary Dean of Systems. Les graduated from the University of California at
Berkeley during the Depression with a degree in Journalism. Being a writer, he
tried his hand at writing Broadway plays. But work was hard to come by during
this period and when World War II broke out, Les was recruited by an aircraft
manufacturer in the midwest to systematize the production of aircraft. Relying
on his experience as a writer, he devised the "Playscript" technique for writing
procedures. Basically, Les wrote a procedure like a script to a play; there was a
section to identify the procedure along with its purpose; a "Setup" section to identify
the forms and files to be used during it; and an "Operations/Instructions" section
which described the "actors" to perform the tasks using verbs and nouns
to properly state each operation. He even went so far as to devise rules for
writing "If" statements.

For details on "Playscript," see "PRIDE" Special Subject Bulletin
No. 38 - "The Language of Systems" - Aug. 22, 2005
http://www.phmainstreet.com/mba/ss050822.pdf

"Playscript" became a powerful procedure writing language and was used
extensively throughout the world. It is still an excellent way to write procedures
today. Ironically, Les did not know what a profound effect his technique would
have later on in the development of computer programs.

1950'S - INTRODUCTION OF THE COMPUTER

Yes, I am aware that the ENIAC was developed for the military at the end
of World War II. More importantly, the UNIVAC I (UNIVversal Automatic Computer)
was introduced in 1951 by J. Presper Eckert and John Mauchly. The UNIVAC I
was a mammoth machine that was originally developed for the U.S. Bureau of
the Census. Corporate America took notice of the computer and companies
such as DuPont in Wilmington, Delaware began to lineup to experiment with
it for commercial purposes. The Remington Rand Corporation sponsored the
project, but the company's focus and name eventually changed to "UNIVAC"
(today it is referred to as "UNISYS," representing a merger of UNIVAC with
Burroughs).

The UNIVAC I offered a sophistication unmatched by other manufacturers,
most notably IBM's Mach I tabulating equipment. This caused IBM to invent
the 701 and its 700 series. Other manufacturers quickly joined the fray and
computing began to proliferate. Although UNIVAC was the pioneer in this
regard, they quickly lost market share due to the marketing muscle of
IBM. For quite some time the industry was referred to as "IBM & the
BUNCH" (Burroughs, UNIVAC, NCR, CDC, and Honeywell).

Programming the early machines was difficult as it was performed in a
seemingly cryptic Machine Language (the first generation language). This
eventually gave way to the Assembly Language (the second generation
language) which was easier to read and understand. Regardless, many of
the utilities we take for granted today (e.g., sorts and merges) simply were
not available and had to be developed. In other words, programming was a
laborious task during this period.

Recognizing both the limitations and potential of the computer, the 1950's
represented the age of experimentation for corporate America. Here, the
emphasis was not on implementing major systems through the computer,
but rather to develop an assortment of programs to test the machine as a
viable product. As such, programmers were considered odd characters who
maintained "the black box," and were not yet considered a part of the
mainstream of systems development. The "Systems and Procedures
Departments" still represented the lion's share of systems work in corporate
America, with an occasional foray to investigate the use of the computer.
The computer people were segregated into "computer departments"
(later to be known as "EDP" or "Data Processing" departments).

1960's - MANAGEMENT INFORMATION SYSTEMS

Competition between computer manufacturers heated up during this
decade, resulting in improvements in speed, capacity, and capabilities.
Of importance here was the introduction of the much touted IBM 360
(the number was selected to denote it was a comprehensive solution -
360 degrees). Other computer vendors offered products with comparable
performance, if not more so, but the IBM 360 was widely adopted by
corporate America.

The programming of computers was still a difficult task and, consequentially,
Procedural Languages were introduced (the third generation languages). In
actuality, these languages got their start in the late 1950's, but the proliferation
of computers in the 1960's triggered the adoption of procedural languages such
as COBOL, FORTRAN, and PL/1. Interestingly, these languages were patterned
after Les Matthies' "Playscript" technique which made active use of verbs,
nouns, and "if" statements.

The intent of the Procedural Languages was twofold: to simplify programming
by using more English-like languages, and; to create universal languages
that would cross hardware boundaries. The first goal was achieved, the
second was not. If the languages were truly universal, it would mean that
software would be portable across all hardware configurations. Manufacturers
saw this as a threat; making software truly portable made the selection of
hardware irrelevant and, conceivably, customers could migrate away from
computer vendors. In order to avoid this, small nuances were introduced
to the compilers for the Procedural Languages thereby negating the concept
of portability. This issue would be ignored for many years until the advent
of the Java programming language.

The 1960's also saw the introduction of the Data Base Management System
(DBMS). Such products were originally designed as file access methods for
Bill of Materials Processing (BOMP) as used in manufacturing. The "DBMS"
designation actually came afterwards. Early pioneers in this area included
Charlie Bachman of G.E. with his Integrated Data Store (IDS) which
primarily operated under Honeywell GCOS configurations; Tom Richley
of Cincom Systems developed TOTAL for Champion Paper, and; IBM's BOMP
and DBOMP products. In 1969, IBM introduced IMS which became their
flagship DBMS product for several years.

With the exception of IMS, the early DBMS offerings were based on a
network model which performed chain-processing. IMS, on the other
hand, was a hierarchical model involving tree-processing.

Realizing that programming and data access was becoming easier and
computer performance being enhanced, companies now wanted to capitalize
on this technology. As a result, corporate America embarked on the era
of "Management Information Systems" (MIS) which were large systems
aimed at automating business processes across the enterprise. These were
major system development efforts that challenged both management and
technical expertise.

It was the MIS that married "Systems and Procedures" departments with
computing/EDP departments and transformed the combined organization into
the "MIS" department. This was a major milestone in the history of systems. The
systems people had to learn about computer technology and the programmers
had to learn about business systems.

Recognizing that common data elements were used to produce the various
reports produced from an MIS, it started to become obvious that data should
be shared and reused in order to eliminate redundancy, and to promote
system integration and consistent data results. Consequently, Data Management
(DM) organizations were started, the first being the Quaker Oats Company in
Chicago, Illinois in 1965. The original DM organizations were patterned after
Inventory Control Departments where the various components were uniquely
identified, shared and cross-referenced. To assist in this regard, such organizations
made use of the emerging DBMS technology. Unfortunately, many DM
organizations lost sight of their original charter and, instead, became obsessed
with the DBMS. Data as used and maintained outside of the computer was
erroneously considered irrelevant. Even worse, the DBMS was used as
nothing more than an elegant access method by programmers. Consequently,
data redundancy plagued systems almost immediately and the opportunity to
share and reuse data was lost. This is a serious problem that persists in
companies to this day.

1970's - AWAKENING

Although the MIS movement was noble and ambitious in intent, it floundered
due to the size and complexity of the task at hand. Many MIS projects suffered
from false starts and botched implementations. This resulted in a period where
a series of new methods, tools and techniques were introduced to reign in these
huge development efforts.

The first was the introduction of the "methodology" which provided a road map
or handbook on how to successfully implement systems development
projects. This was pioneered by MBA with its "PRIDE" methodology in
1971. Although the forte of "PRIDE" was how to build systems, it was initially
used for nothing more than documentation and as a means to manage projects.
Following "PRIDE" was John Toellner's Spectrum I methodology and SDM/70
from Atlantic Software. Several CPA based methodologies followed thereafter.

Also during this time, mainframe based Project Management Systems were
coming into vogue including Nichols N5500, PAC from International Systems,
and PC/70 from Atlantic Software.

The early methodologies and Project Management Systems give evidence of
the orientation of systems departments of that time: a heavy emphasis on Project
Management. Unfortunately, it was a fallacy that Project Management was
the problem; instead people simply didn't know how to design and build
systems in a uniform manner. As companies eventually learned, Project
Management is useless without a clear road map for how to build something.

In the mid-to-late 1970's several papers and books were published on how
to productively design software thus marking the beginning of the "Structured
Programming" movement. This was a large body of work that included such
programming luminaries as Barry Boehm, Frederick P. Brooks, Larry Constantine,
Tom DeMarco, Edsger Dijkstra, Chris Gane, Michael A. Jackson, Donald E. Knuth,
Glenford J. Myers , Trish Sarson, Jean Dominique Warnier, Generald M. Weinberg,
Ed Yourdon, as well as many others. Although their techniques were found
useful for developing software, it led to confusion in the field differentiating
between systems and software. To many, they were synonymous. In reality,
they are not. Software is subordinate to systems, but the growing emphasis
on programming was causing a change in perspective.

The only way systems communicate internally or externally to other systems
is through shared data; it is the cohesive bond that holds systems (and software)
together. This resulted in the introduction of Data Dictionary technology. Again,
this was pioneered by MBA with its "PRIDE" methodology (which included a manually
implemented Data Dictionary) and later with its "PRIDE"-LOGIK product in
1974. This was followed by Synergetics' Data Catalogue, Data Manager from
Management Software Products (MSP), and Lexicon by Arthur Andersen & Company.

The intent of the Data Dictionaries was to uniquely identify and track where
data was used in a company's systems. They included features for maintaining
documentation, impact analysis (to allow the studying of a proposed change),
and redundancy checks. "PRIDE"-LOGIK had the added nuance of cataloging
all of the systems components, thereby making it an invaluable aid for
design and documentation purposes.

The Data Dictionary was also a valuable tool for controlling DBMS products
and, as such, several adjunct products were introduced, such as UCC-10,
DB/DC Data Dictionary, and the Integrated Data Dictionary (IDD) from
Cullinet. Unlike the other general purpose Data Dictionaries, these products
were limited to the confines of the DBMS and didn't effectively track data
outside of their scope.

DBMS packages proliferated during this period with many new products
being introduced including ADABAS, Image, Model 204, and IDMS from
Cullinet (which was originally produced at BF Goodrich). All were based
on the network-model for file access which was finally adopted as an
industry standard (CODASYl).

There were a few other notable innovations introduced, including IBM's
Business Systems Planning (BSP) which attempted to devise a plan for
the types of systems a company needed to operate. Several other comparable
offerings were introduced shortly thereafter. Interestingly, many companies
invested heavily in developing such systems plans, yet very few actually
implemented them.

Program Generators were also introduced during this period. This included
report writers that could interpret data and became a natural part of the
repertoire of DBMS products. It also included products that could generate
program source code (COBOL predominantly) from specifications. This
included such products as System-80 (Phoenix Systems), GENASYS (Generation
Sciences), and JASPOL (J-Sys of Japan), to mention but a few.

MBA also introduced a generator of its own in 1979 - a Systems generator
initially named ADF (Automated Design Facility) which could automatically
design whole systems, complete with an integrated data base. Based on
information requirements submitted by a Systems Analyst, ADF interacted
with the "PRIDE"-LOGIK Data Dictionary to design new systems and, where
appropriate modify existing systems. Because of its link to LOGIK, ADF
emphasized the need to share and reuse information resources. Not only
was it useful as a design tool but it was a convenient tool for documenting
existing systems. The only drawback to ADF was that the mindset of the
industry was shifting from systems to software. Consequently, program
generators captured the imagination of the industry as opposed to ADF.

The increase in computer horsepower, coupled with new programming
tools and techniques, caused a shift in perspective in MIS organizations. Now,
such departments became dominated by programmers, not systems people. It
was here that the job titles "Systems Analyst" and "Programmer" were married
to form a new title of "Programmer/Analyst" with the emphasis being on
programming and not on front-end systems design. Many managers falsely
believed that developers were not being productive unless they were
programming. Instead of "Ready, Aim, Fire," the trend became "Fire, Aim,
Ready."

Data Management organizations floundered during this period with the
exception of Data Base Administrators (DBA's) who were considered the
handmaidens of the DBMS.

The proliferation of software during this decade was so great that it
gave rise to the packaged software industry. This went far beyond
computer utilities and programming tools. It included whole systems
for banking, insurance and manufacturing. As a result, companies were
inclined to purchase and install these systems as opposed to reinventing
the wheel. Among their drawbacks though was that they normally required
tailoring to satisfy the customer's needs which represented modification to the
program source code. Further, the customer's data requirements had to
be considered to assure there were no conflicts in how the customer
used and assigned data. After the package had been installed, the
customer was faced with the ongoing problem of modifying and enhancing
the system to suit their ever-changing needs.

1980's - THE TOOL-ORIENTED APPROACH

As big iron grew during the 1960's and 1970's, computer manufacturers
identified the need for smaller computers to be used by small to medium-sized
businesses. In the 1970's, people were skeptical of their usefulness but
by the 1980's their power and sophistication caused the "mini" computer
to gain in popularity as either a general purpose business machine or dedicated
to a specific system. Among the most popular of the "mini" computers were:

• IBM's System 36/38 series (which led to the AS/400)
• DEC PDP Series (which gave way to the DEC VAX/VMS)
• Hewlett-Packard's HP-3000 series with MPE
• Data General Eclipse series with AOS
• PRIME

The competition was fierce in the "mini" market which resulted in
considerable product improvements and better value to the customer.
Instrumental to the success of the mini was the adoption of UNIX as
developed by Bell Labs, a powerful multi-user, multitasking operating system
that eventually was adopted by most, if not all, mini manufacturers.

But the major development in computer hardware was not the mainframe,
nor the mini; it was the "micro" computer which was first popularized by
Apple in the late 1970's. IBM countered with the its Personal Computer (PC)
in the early 1980's. At first, the micro was considered nothing more than
a curiosity but it quickly gained in popularity due to its inexpensive cost,
and a variety of "apps" for word processing, spreadsheets, graphics, and
desktop publishing. This caught on like wildfire as micros spread through
corporate desktops like the plague.

By the mid-1980's the "micro" (most notably the PC) had gained in power
and sophistication. So much so, that a series of graphical based products
were used for software development in support of the Structured Programming
movement of the 1970's. Such tools were dubbed "CASE" (Computer Aided
Software Engineering) which allowed developers to draw their favorite software
diagramming technique without pencil and paper. Early CASE pioneers
included Index Technology, Knowledgeware, Visible Systems, Texas
Instruments, and Nastec, as well as many others. CASE tools took the industry
by storm with just about every MIS organization purchasing a copy either for
experimental use or for full application development. As popular as the tools
were initially, there is little evidence they produced any major systems but,
instead, helped in the design of a single program.

Recognizing the potential of the various CASE tools, IBM in the late
1980's devised an integrated development environment that included IBM's
products as well as third parties, and entitled it "AD/Cycle." However, IBM
quickly ran into problems with the third party vendors in terms of agreeing
on technical standards that would enable an integrated environment. Consequently,
the product ran aground not long after it was launched. In fact, the
prosperity of the CASE market was short-lived as customers failed to realize
the savings and productivity benefits as touted by the vendors. By the
early 1990's, the CASE market was in sharp decline.

Instead, companies turned to Programmer Workbenches which included
an all-in-one set of basic tools for programming, such as editing, testing,
and debugging. Microsoft and Micro Focus did particularly well in offering
such products.

Data Base Management Systems also took a noticeable turn in the 1980's
with the advent of "relational" products involving tables and keys. The
concept of the "relational" model was originally developed by IBM
Fellow and mathematician Edgar (Ted) Codd in a paper from 1970. The concept
of a relational DBMS was superior to the earlier network and hierarchical
models in terms of ease of use. The problem resided in the amount of
computer horsepower needed to make it work, a problem that was
overcome by the 1980's. As a result. new DBMS products such as Oracle and
Ingres were introduced which quickly overtook their older competitors. There
was an initial effort to convert DBMS mainstays such as TOTAL, ADABAS, and
IDMS into relational products, but it was too little, too late. As for IBM,
they simply re-labeled their flagship product, IMS, as a "transaction processor"
and introduced a totally new offering, DB2, which quickly dominated the
DBMS mainframe market.

Program generators continued to do well during the 1980's but it was during
this period that 4GL's (fourth generation languages) were introduced to
expedite programming. The 4GL was a natural extension of the DBMS and
provided a convenient means to develop programs to interpret data in the
data base.

Another development worth noting is the evolution of the Data Dictionary
into "Repositories" (also referred to as "Encyclopedias") used to store the
descriptions of all of an organization's information resources. One of the
motivating factors behind this was IBM (for AD/Cycle) who realized they
needed some sort of cohesive bond for the various CASE tools to interface. This
is another area pioneered by MBA who introduced their "PRIDE"-Enterprise
Engineering Methodology (EEM) to study a business and formulate an
Enterprise Information Strategy, and their "PRIDE"-Data Base Engineering
Methodology (DBEM) to develop the corporate data base, both logically
and physically. To implement these new methodologies, their
"PRIDE"-LOGIK Dictionary was expanded to include business models, and
data models. By doing so, MBA renamed "PRIDE"-LOGIK the "PRIDE"-IRM
(Information Resource Manager) which complemented their concept of
Information Resource Management.

In terms of the MIS infrastructure, two noteworthy changes occurred;
first was the introduction of the Chief Information Officer (CIO) as first
described in the popular book, "Information Systems Management In Practice"
(McNurlin, Sprague) in January 1986. Basically, the MIS Director is elevated
to a higher management level where, theoretically, he/she is operating on the
same level as the Chief Operating Officer (COO), and Chief Financial Officer
(CFO) for a company. In reality, this has never truly happened and, in many
cases, the title "CIO" is nothing more than a change in name, not in stature.
The second change is the change in job title of "Programmer" to "Software
Engineer." Again, we are primarily talking about semantics. True, many
of the programmers of the 1980's studied Structured Programming, but
very few truly understood the nature of engineering as it applies to
software, most are just glorified coders. Nonetheless, the "Software
Engineer" title is still actively used today. In contrast, the last of the
true "Systems Analysts" slowly disappeared. Here too is evidence of
the change of focus from systems to software.

During the 1980's we also saw the emergence of MBA's graduating from
the business schools and working their way into the corporate landscape. Although
they didn't have an immediate impact on the systems world, they had a dramatic
effect on the corporate psyche. Their work resulted in severe corporate cutbacks,
downsizing, and outsourcing. This changed the corporate mindset to think
short-term as opposed to long-term. Following this, companies shied away from
major systems projects (such as the MIS projects of the 1960's) and were content
tackling smaller programmer assignments, thus the term "app" was coined to
describe a single program application.

Interestingly, a "quality" movement flourished in the 1980's based on the
works of W. Edwards Deming and Joseph M. Juran who pioneered quality
control principles in the early part of the 20th century. Unfortunately, their
early work was unappreciated in America and, consequently, they applied
their talents to help rebuild the industrial complex of postwar Japan. It was
only late in their lives did they receive the recognition of their work in the
United States (after Japan became an economic powerhouse). Another
influential factor was the introduction of the ISO 9000 standard for quality
management which was originally devised by the British and later adopted as
an international standard. Little attention would probably have been paid to
ISO 9000 if it weren't for the fact that European businesses started to demand
compliance in order to conduct business with their companies.

Nevertheless, these factors resulted in a reorientation of American
businesses to think in terms of developing quality products which,
inevitably, affected how systems and software were produced. The real impact
of the quality movement though wouldn't be felt in the systems world until
the next decade.

To summarize the 1980's from a systems development perspective, the focus
shifted away from major systems to smaller programming assignments which
were implemented using newly devised CASE tools. This fostered a "tool-oriented
approach" to development whereby companies spent considerably on the
latest programming tools but little on management and upfront systems
work. In other words, they bought into the vendor's claims of improved
programmer productivity through the use of tools. Unfortunately, it resulted
in patchwork systems that required more time in maintenance as opposed to
modifying or improving systems. "Fire fighting" thereby became the normal
mode of operation in development.

1990's - REDISCOVERY

As the PC gained in stature, networking became very important to
companies so that workers could collaborate and communicate on a
common level. Local Area Networks (LAN) and Wide Area Networks (WAN)
seemed to spring-up overnight. As the PC's power and capacity grew, it
became obvious that companies no longer needed the burden of mainframes
and minis. Instead, dedicated machines were developed to control and
share computer files, hence the birth of "client/server computing" where
client computers on a network interacted with file servers. This did not
completely negate the need for mainframes and minis (which were also
used as file servers), but it did have a noticeable impact on sales. Companies
still needed mainframes to process voluminous transactions and extensive
number-crunching, but the trend was to move away from big iron.

Thanks to the small size of the PC, companies no longer required a big
room to maintain the computer. Instead, computers were kept
in closets and under desks. This became so pervasive that companies no
longer knew where their computer rooms were anymore. In a way, the
spread of computers and networks closely resembled the nervous system
of the human body.

One of the key elements that made this all possible was the introduction
of Intel's 30386 (or "386") chip which allowed 32-bit processing. To effectively
use this new technology, new operating systems had to be introduced, the
first being IBM's OS/2 in the late 1980's. OS/2 provided such things as
virtual memory, multitasking and multithreading, network connectivity,
crash-protection, a new High Performance File System, and a slick object
oriented desktop. Frankly, there was nothing else out there that could
match it. Unfortunately, Microsoft bullied its way past OS/2 with
Windows 95 & NT. By the end of the 1990's, OS/2 was all but forgotten
by its vendor, IBM. Nevertheless, it was the advent of 32-bit computing
that truly made client/server computing a reality.

Another major milestone during this decade was the adoption of the
Internet by corporate America. The Internet actually began in the late
1960's under the Department of Defense and was later opened to other
government and academic bodies. But it wasn't until the 1990's that
companies started to appreciate the Internet as a communications and
marketing medium.

The first web browser was developed by Tim Berners-Lee in 1990 which
led to the World Wide Web protocol on the Internet. Early web browsers
included Mosaic, Netscape Navigator, and Microsoft's Internet Explorer,
among others. The beauty of the Internet was that all computers could
now access the Internet regardless of the operating system, making it a
truly universal approach to accessing data. To write a web page, a simple
tag language was devised, Hyper Text Markup Language (HTML), which was
compiled at time of request to display the web page. HTML was nice for
developing simple static web pages (not much interaction, just simply view the
web page). Developers then invented new techniques to make a web page
more dynamic thereby allowing people to input data and interact with files,
which ultimately allowed for the merchandising of products over the Internet.

Wanting to do something more sophisticated through the web browser,
Sun Microsystems developed the Java programming language in 1995. Java
was a universal programming language that could run under any operating
system. Their mantra was "Write once, run anywhere." This was a radical
departure from programming in the past where it was necessary to recompile
programs to suit the peculiarities of a particular operating system. Basically,
Java made the operating system irrelevant, much to Microsoft's chagrin. Further,
Java could be used in small pocket devices as well as in the new generation
of computers powering automobiles. This did not sit well with Microsoft who
ultimately fought the propagation of Java.

By the 1990's the Structured Programming movement had fizzled out. Instead,
"Object Oriented Programming" (OOP) gained in popularity. The concept of OOP
was to develop bundles of code to model real-world entities such as customers,
products, and transactions. OOP had a profound effect on Java as well as the
C++ programming language.

During this time, source code generators faded from view. True, companies were
still using report writers and 4GL's, but the emphasis turned to "Visual Programming"
which were programming workbenches with screen painting tools to layout
inputs and outputs.

The Relational DBMS movement was still in high gear, but the use of Repositories
and Data Dictionaries dropped off noticeably. Of interest though was the
introduction of "Object Oriented Data Base Management System" (OODBMS)
technology. Like OOP, data was organized in a DBMS according to real-world
entities. Regardless, Relational DBMS dominated the field.

Also during this decade "Data Mining" became popular whereby companies were
provided tools to harvest data from their DBMS. This effort was basically an admission
that companies should learn to live with data redundancy and not be concerned
with developing a managed data base environment.

Because of the radical changes in computer hardware and software, companies became
concerned with their aging "legacy" systems as developed over the last thirty years. To
migrate to this new technology, a movement was created called "Business Process
Re-engineering" (BPR). This was encouraging in the sense that companies were
starting to think again in terms of overall business systems as opposed to just
programs. I'm not sure I agree with the use of the term "Re-engineering" though;
this assumes that something was engineered in the first place (which was hardly the
case in these older systems).

Nonetheless, CASE-like tools were introduced to define business processes. Suddenly,
companies were talking about such things as "work flows," "ergonomics," and
"flowcharts," topics that had not been discussed for twenty years during the frenzy of
the Structured Programming movement. Ultimately, this all led to the rediscovery of
systems analysis; that there was more to systems than just software. But by this time,
all of the older corporate Systems Analysts had either retired or been put out to pasture,
leaving a void in systems knowledge. Consequently, the industry started to relearn
the systems theory, with a lot of missteps along the way.

Companies at this time were still struggling with devising a suitable development
environment. Most were content with just maintaining their current systems
in anticipation of the pending Y2K (Year 2000) problem (where date fields were to
change from 19XX to 20XX which could potentially shutdown companies). However,
a few companies began to consider how to apply more scientific principles to the
production of systems. Since people were already talking about "Software
Engineering," why not apply engineering/manufacturing principles to the
development of total systems?

Back in the early 1980's, Japan's Ministry of International Trade & Industry (MITI)
coordinated a handful of Japanese computer manufacturers in establishing a
special environment for producing system software, such as operating systems
and compilers. This effort came to be known as Japanese "Software Factories"
which captured the imagination of the industry. Although the experiment
ended with mixed results, they discovered organization and discipline
could dramatically improve productivity.

Why the experiment? Primarily because the Japanese recognized there are
fundamentally two approaches to manufacturing anything: "one at a time" or
mass production. Both are consistent approaches that can produce a high
quality product. The difference resides in the fact that mass production offers
increased volume at lower costs. In addition, workers can be easily trained
and put into production. On the other hand, the "one at a time" approach is
slower and usually has higher costs. It requires workers to be intimate with
all aspects of the product.

MBA took it a step further by introducing their concept of an "Information
Factory" in the early 1990's. The Information Factory was a comprehensive
development environment which implemented MBA's concept of Information
Resource Management. Basically, they drew an analogy between developing
systems to an engineering/manufacturing facility, complete with assembly lines,
materials management and production control. These concepts were proven
effective in companies throughout Japan, most notably Japan's BEST project,
which was sponsored by the Ministry of Finance. As background, the ministry
wanted to leapfrog the west in terms of banking systems. To do so, they
assembled a team of over 200 analysts and programmers from four of the top
trust banks in Japan; Yasuda Trust & Banking, Mitsubishi Trust & Banking,
Nippon Trust & Banking, and Chuo Trust & Banking. By implementing MBA's
concepts they were able to deliver over 70 major integrated systems in less
than three years. Further, because they had control over their information
resources using a materials management philosophy, the Y2K problem
never surfaced.

In terms of infrastructure, development organizations essentially went
unchanged with a CIO at the top of the pyramid and supported by
Software Engineers and DBA's. But there was one slight difference,
instead of being called an MIS or IS department, the organization was
now referred to as "IT" (Information Technology). Here again, the name hints
at the direction most organizations were taking.

Finally, the 1990's marked a change in the physical appearance of
the work force. Formal suit and ties gave way to casual Polo shirts and
Docker pants. At first, casual attire was only allowed on certain days
(such as Fridays), but it eventually became the normal mode of dress. Unfortunately,
many people abused the privilege and dressed slovenly for work. This had
a subtle but noticeable effect on work habits, including how we build systems.

2000's - GADGETS

We are now past the halfway point in this decade and there is nothing of
substance to report in terms of computer hardware, other than our machines
have gotten faster, smaller, with even more capacity. Perhaps the biggest innovation
in this regard is the wide variety of "gadgets" that have been introduced, all of
which interface with the PC, including: Personal Digital Assistants (PDA's), iPods,
MP3 players, digital cameras, portable CD/DVD players (and burners), cell phones,
PS2 and XBox game players. These devices are aimed at either communications or
entertainment, giving us greater mobility, yet making us a bit dysfunctional
socially. All of this means the computer has become an integral part of
our lives, not just at work but at home as well.

Shortly after taking the reigns of IBM in 2003, CEO Sam Palmisano introduced
"On-Demand Computing" as the company's thrust for the years ahead and, inevitably,
it will mark his legacy. The concept as described by Palmisano was simple,
treat computing like a public utility whereby a company can draw upon IBM for
computing resources as required. "On-Demand Computing" made a nice
catch-phrase and was quickly picked up by the press, but many people
were at a loss as to what it was all about. Some of the early developments
resulting from IBM's "e-Business On Demand" research included balancing the load
on file servers, which makes sense. But IBM is carrying the analogy perhaps
too far by stressing that "on demand" is the manner by which companies should
run in the future. Basically, the theory suggests we abandon capacity planning
and rely on outside vendors to save the day. Further, it implies computers
supersede the business systems they are suppose to serve. Instead of
understanding the systems which runs a business, just throw as much computer
resources as you need to solve a problem. This is like putting the cart before
the horse.

The "on-demand" movement has evolved into "Service Oriented Architectures"
(SOA) where vendors are introducing "on-demand" applications that will take
care of such tasks as payroll, marketing, etc. through the Internet. Again, it all
sounds nice, but as far as I can see, this is essentially no different than service
bureaus like ADP who for years provided such processing facilities. Now,
companies are being asked to swap out their internal programs for third party
products. I fail to see how this is different than buying any other packaged
solution, other than an outsider will be taking care of your software.

The need to build software faster has reached a feverish pitch. So much so,
full-bodied development methodologies have been abandoned in favor of
what is called "Agile" or "Extreme Programming" which are basically
quick and dirty methods for writing software using power programming
tools. To their credit, those touting such approaches recognize this is limited
to software (not total systems) and is not a substitute for a comprehensive
methodology. Agile/Extreme Programming is gaining considerable attention
in the press.

Next, we come to "Enterprise Architecture" which is derived from a paper
written by IBM's John A. Zachman who observed that it was possible to apply
architectural principles to the development of systems. This is closely
related to consultants who extoll the virtues of capturing "business rules"
which is essentially a refinement of the Entity Relationship (ER) Diagramming
techniques popularized a decade earlier using CASE tools.

As in the 1990's, concepts such as "Enterprise Architecture" and "business rules"
is indicative of the industry trying to reinvent systems theory.

CONCLUSIONS

Like computer hardware, the trend over the last fifty years in systems
development is to think smaller. Developers operate in a mad frenzy to
write programs within a 90 day time frame. Interestingly, they all know that
their corporate systems are large, yet they are content to attack them one
program at a time. Further, there seems to be little concern that their work
be compatible with others and that systems integration is someone else's
problem. Often you hear the excuse, "We don't have time to do things
right." Translation: "We have plenty of time to do things wrong." Any shortcut
to get through a project is rationalized and any new tool promising improved
productivity is purchased. When companies attempt to tackle large systems
(which is becoming rare) it is usually met with disaster. Consequently, companies
are less confident in their abilities and shy away from large system development
projects.

Corporate management is naive in terms of comprehending the value of
information and have not learned how to use it for competitive advantage
(unlike their foreign competitors). Further, they are oblivious to the problems in
systems development. They believe their systems are being developed with a high
degree of craftsmanship, that they are integrated, and that they are easy to maintain
and update. Executives are shocked when they discover this is not the case.

The problems with systems today are no different than fifty years ago:

• End-user information requirements are not satisfied.
• Systems lack documentation, making maintenance and upgrades difficult.
• Systems lack integration.
• Data redundancy plaques corporate data bases.
• Projects are rarely delivered on time and within budget.
• Quality suffers.
• Development personnel are constantly fighting fires.
• The backlog of improvements never seems to diminish, but rather increases.

Although the computer provides mechanical leverage for implementing
systems, it has also fostered a tool-oriented approach to systems development. Instead
of standing back and looking at our systems from an engineering/manufacturing
perspective, it is seemingly easier and less painful to purchase a tool to solve a
problem. This is like taking a pill when surgery is really required. What is
needed is less tools and more management. If we built bridges the same way
we build systems in this country, this would be a nation run by ferryboats.

The impact of the computer was so great on the systems industry that it
elevated the stature of programmers and forced systems people to near
extinction. Fortunately, the industry has discovered that there is more
to systems than just programming and, as a result, is in the process of
rediscovering basic systems theory. Some of the ideas being put forth are truly
imaginative, others are nothing more than extensions of programming theory,
and others are just plain humbug. In other words, the systems world is still going
through growing pains much like an adolescent who questions things and learns
to experiment.

I have been very fortunate to see a lot of this history first hand. I have
observed changes not just in terms of systems and computers, but also
how the trade press has evolved and the profession in general. It
has been an interesting ride.

Throughout all of this, there have been some very intelligent people who
have impacted the industry, there have also been quite a few charlatans,
but there has only been a handful of true geniuses, one of which was
Robert W. Beamer who passed away just a couple of years ago. Bob
was the father of ASCII code, without which we wouldn't have the computers
of today, the Internet, the billions of dollars owned by Bill Gates, or
this document.