.br
.ul
Documentation:
.pg
Fundamental to all branches
of research and development is
the need to keep records of all
activity as well as details of
all procedures and methods.
The system of record keeping is
called
.ul
Documentation.
In many organizations, personnel
trained in technical writing
are employed as documenters,
and in a scientific
computation center may be a
part of the library section or
as will be discussed later,
a part of the applications
programming groups.
Good computation center documentation
whether it be for management progress
reports, for summary of
technical analysis, data
and report requirements, or for
the recording of program analyst
development of
.ul
macros,
subroutines, subprograms and programs
for the project is 
.ul
disclosure.
Disclosure to persons authorized
to receive information is
highly desired|-|but disclosure
to others is certainly a
violation of the rights of
the client to privacy.
When the client's project
includes features that many
be copyrighted, patented,
or fall into the "trade secret"
category, unwarranted disclosure
is not only unethical
but may cause economic loss to
the client organization.
Sometimes the clever use of
the computer in manufacturing
or exploitation of
national resources falls into the
classification of "know-how"
and has real economic value to the
client organization.
The scientific computation
center in carrying out
its documentation function is
very strict about control of
these documents and their
distribution.
.pg
Every substantial project
must be documented at
least at three levels,
.in7
.ti-4
.ul
1.||Scientific:
.ti-4
.ul
2.||Computation program:
.ti-4
3.||Computing.
.in0
Without this kind of record
keeping any loss of
key personnel can be
disastrous.
.br
For Example:||In one agency of a
.m4 1
major governmental research
institute, two and a
half years work in developing a
computerized procedure was
lost when the program analyst
leader quit and took his personal
notes with him leaving
a set of computer programs almost
finished and completely undocumented.
Recovery was economically impossible,
it was cheaper to start over.
.br
.ul
Scientific documentation:
.pg
Scientific documentation specifies
what the client
organization wants done;
what scientific rules it
is using and where in the
analytic procedure each rule
is to be applied.
This permits the client to plan
modification, improvement, radical
.m4 3
alteration as well as to
publish what has been done.
It must be done jointly with
computation center personnel but
is the responsibility of the client.
.br
.ul
Computation program documentation:
.pg
This is a highly technical document
which will permit
expert computation analysts
to evaluate the machine
procedures, or to plan to meet
new analytical requirements by the client.
Its form is often affected by the
program language, associated compiler,
and operating system used in the center.
Every macro, every subroutine,
every variable must be described
and its function in the program made explicit.
This holds even for those algorithms provided
in packages by the manufacturer.
Since the overall operating system
used by the center is dynamic
and subject to modification it has
and will happen that changes in
the operating environment of
the computing system can adversely
affect a proven program.
With good documentation it is
possible to detect and
correct programming which
would otherwise be negated by known,
scheduled modifications
to the system.
.br
.ul
Computing documentation:
.pg
This documentation is
prepared for projects whose
computation is of a recurring nature
and where the responsibility for
repeated runs will be turned
over to a center operating group formed
from the input and results clerk force
and a similar group from the project.
This document contains a section often called the
Run Book.
This section documents the computational
requirements in complete
and systematic detail.
It lays out the job request sheet
which will specify each input
and each output element as well
as the amount of running time,
the amount and type of printed output,
and for each error indicator will
describe in minute
detail the required action.
If loading an improper input
tape or any other type of
malfunction will cause the
run to abort, the documentation
will spell out the RESTART procedures
in detail which will
be operable by non prime shift crews.
This is important to do well,
because otherwise the run will be cancelled,
a day lost for the project,
and a substantail use of the
computer may be wasted.
In very important work,
it is common to list emergency
telephone numbers so that in
the event of an abort or
serious malfunction a project
representative can
make immediate decisions or
corrections.
.br
.ul
Leadership and Liability:
.pg
The project leader may be
a member of the computation center
although it is preferable that he
come from the customer's staff.
When it is desirable that
the defacto leader must come
from the center staff, it is
often possible and certainly
wise to find a respected
member of the customer staff who
accepts the title and delegates informally
to the defacto leader the necessary
authority.
Someone with authority must accept
the responsibility of coordinating all
of the activities of the
project for the duration of its
initial planning and development and of making
available people and information
needed for the project.
Occasionally he finds that inter-department
rivalry within the customer group is 
so strong that it will threaten
the success of the project.
He must have the authority 
or have immediate access to authority to
override inter-department conflicts and
to permit analysts to approach
any internal department
associated with the project
for information required
by the project.
It will be assumed in this monograph that
our project leader has been
given all necessary authority to carry
out his responsibility.
The project leader role may continue
to exist after the project becomes
operational if the project
if of the regularly
recurring type or if further
substantial development
is anticipated.
.ce
.ul
ADDENDA
.br
.ul
Remote Operations
.pg
It will be noticed that this
monograph has ignored the
developments of remote
use of the computer
system.
This has been deliberate.
Except for those regions where
the telephone system has
.ul
spare
capacity of transmission
facilities of data grade,
it is electronically
not feasible for interlocation
remote operation.
For most nations of the
world there is an acute
shortage of 2400 baud,
50 kilo-bit,
or 1 mega-bit transmission facilities.
As fast as new circuits are
built governmental agencies and industry
absorb them.
Furthermore, these circuits are expensive.
Aside from the availability of
adequate communications support,
there is a further negative aspect about
remote operation.
Viewed from an economic point
of view, remote operation
in research and development is
a trade-off of professional
and technical manpower
for money, communication
and computer resources.
This mode of operation is
not an efficient as a good batch
processor system
for a national resource (which implies
scarcity).  It is usually
not in the national interest
to make such a trade-off.
Sociologists are beginning
to suspect another negative
feature of dialogue remote operation.
The use of a typewriter-like
terminal with computer response in
fractions of a second
seems to diminish the
state of reflection and though
and encourage the
substitution of action
based on first impression
for action based on careful analysis.
There is some indication that slower
response and controlled access reduces
the ratio of number of attempts to success.
Whatever the value of this technology,
the addition of a remote
operations capability will require
all the technical and
organizational support described
in this monograph plus the technical
supervision to control this new
type of facility, plus
a well organized sub-group in the
results section to deliver
by conventional methods the
generated output.
.br
.ul
Plotters, paper and microfilm
.pg
These are machines that can be
.hc %
very valuable to the scientific
computation center.
The paper plotters are
used for automated engineering
drawing and meet reasonable
drafting standards.
Their price run from 20,000-120,000\$U.S.
The microfilm plotters are
more versatile and can produce
text, as well as
mathematical and
engineering drawings with
the basic unit a 35 mm frame and with
a large number of frames
per job permitted.
.hc $
The quality of text can
with great care be almost to
graphic arts standards.
Some machines mow claim to
meet electronic drafting standards.
.hc %
Although these machines have a high initial
costs 200,000-250,000\$U.S. the
unit cost when the demand is high
can be very competitive.
For text, one manufacturer claims 4
pages per second.
In multi-lingual nations and
especially for those languages which have
a different character structure and should be
written rather than printed
this type of output can
be very valuable.
The following areas can profit from
this facility.
Catalogs of national collections,
inventories in vernacular language
of major national programs,
passport control lists,
telephone directories in the local
languages.
Both types of machine are usually used
.hc $
in the off-line mode, but the
microfilm devices are also stand-alone
and with proper interface elements
are computer independent.
This means that this machine can serve a
much greater demand than
will be generated by the
center's computer system.
Hence demand from other areas in
the region, and in neighboring
regions should be examined for
possible load.
Sample is shown as
Figure 3 which was completely done
by a microfilm printer,
a master produced, and then copied
.m4 1
electrostatically.
Each character was traced on
the face of a cathode ray tube
by an electron beam
and captured on an
exposed 35mm film.
In this figure two
frames were exposed and then joined
.hc $
manually before a master print
was made.
.bp
.nx mt58.n
