By The Late Will Naeher - DASC Retired
This is the first of what I hope will be a series of articles concerning the history of the Office of Communications (OC) of the Department of State. I realize there is much history that predates these articles. Some have already appeared in the CANDOER. However, I was not a part of that history, so I will leave it to others to add to this.
The following article is about the ATS, which I believe had the most impact on how the Department communicates with its field stations. To my knowledge, there are no written documents which describe the ATS, still available or in existence. Therefore, the following articles will be, to a large part, anecdotal and are derived from my memory of those events. I ask that, if there are any inaccuracies or omissions, you will write to the editor, so that this history will be as accurate as possible.
During the Cuban Missile Crisis, President John F. Kennedy had considerable difficulty communicating with his embassies and other government agencies in Washington. He was forced to use the media to indicate his intentions to other foreign governments. After the crisis was over, he appointed a Committee Chaired by John Orrick, from the Department of Justice, to review and examine the Government's communications networks (Mr. Orrick later became the Deputy Under Secretary of State for Management). One of the Orrick Committees recommendations was that a modern "state-of-the-art communications terminal" be designed and installed in the Department. A task force was formed to accomplish this. A general description of the problem and an announcement was published indicating the Department's intentions, together with the descriptions to solicit interest. Nine companies responded. The task was then to develop a composite of the replies which contained the best solutions to the problem. This composite was then reissued requesting unpriced proposals. All nine of the companies responded.
A committee was then formed composed of elements of the Department, i.e., Security, engineering operations, etc. Representatives of the CIA and the Military were also included. After a review, the committee decided that all companies were responsive and could do the job. This meant that it would be necessary to negotiate with all the companies. This would be very labor intensive and time consuming.
Jack Coffey then appointed another committee consisting of five representatives of OC to again review the replies. This committee eliminated five of the companies for being non-responsive. They reissued the RFP to four companies, ITT, Boroughs, Honeywell, and RCA, with a request for prices. IBM also responded, but their proposal was to perform a study to determine the scope of the project. We found them to be non-responsive and they were eliminated.
After a series of meeting with each vendor, ITT was awarded the contract. ITT had the advantage in that they had just finished a major contract for SAC, called SACDIN, which involved a series of switches based on the ACP 127 protocol and they had completed site construction for several DEW Line sites. Therefore, they had experienced personnel already on board and were prepared to begin the project immediately. Furthermore, they were low-bidder. To eliminate conflict that may arise between two contractors, we decided to award ITT a separate but interrelated contract to construct the site for the ATS.
I was appointed as the Contracting Officer's Representative for the system. Jack Hulbert, who was a member of the various RFP development committees, was appointed the Contracting Officer's Technical Representative for the construction of the site. Our task was to insure contract compliance.
The system design was complicated. A part of it had been done before. However, this system combined the disciplines of system switching with system termination procedures. For example, there were several switches in existence using the ACP 127 protocol. However, there were no systems that combined the discipline of terminal processing.
The requirements of the system were:
1. Up to 50 separate circuits of various speeds.
2. Be capable of automatically switching messages from one circuit to another, using the ACP 127 protocol. This was the standard for all switching systems, either manual or automated in the U.S. Government networks, at that time.
3. Terminate messages addressed to the Department on Video type viewing screens, where analysts could review the text of the telegram and determine distribution. The analysts could also correct garbles and spelling or forward the telegram to others for service message action. If the message was addressed to other stations, the switch would relay the message to the addressees indicated.
4. The system should automatically examine certain captions contained in the heading to determine suggested distribution and indicate this distribution on the bottom of the video screen. The analyst could then change this distribution, or add to it, after a careful review of the text. The distribution system also indicated the number of copies which were to be made for each individual distribution addresses, and would indicate the total number of copies to be reproduced and whether it was to be reproduced on pink paper for outgoing or white paper for incoming. The system would also automatically change the format for distribution by stripping off element of the ACP 127 heading and replace it with a designated distribution format.
5. Indicate in the Logo whether the message was Incoming or Outgoing.
6. Store the message for a period of 30 days, so that it could be retrieved by message reference number so that the analysts could review the distribution, or by the message channel number, so that the Systems personnel could make a retransmission in response to a service message.
7. Automatically assign the appropriate JANAP 117 routing indicator to outgoing messages based on the designated addressee(s).
8. Design the system in such a way that no message could be lost. The system was required to be fully redundant.
9. The system should automatically issue channel numbers for each transmission to each circuit to ensure circuit continuity and prevent the loss of messages en route, which could be caused by circuit difficulties. This was accomplished by sending, automatically, a service message addressed to RUEH which was the Department's routing indicator. When the service message was received, the channel number was checked. If there was a discrepancy, a message was printed on a monitoring Teletype receiver for action. The system also assigned a Message Reference Number to every outgoing message. It also assigned a Message Continuity Number to each message between the Department and addressees to ensure that no messages were lost en route.
10. Accept messages from a high speed tape reader art the rate of 1200 words per minute. Later, this was replaced by an optical character reader (OCR), which was interfaced directly to the system. This was a first in communications operations and affected many of the telegram development mores in the Department. It required that a distinctive type font, "OCR A," be used when preparing telegrams, so that the OCR could read the message. To accomplish this, the Department procured the IBM "golf gall" typewriter as a standard so that the OCR A font could be used. This new procedure eliminated more than 20 positions in the communications center.
11. Maintain an accounting of all messages entering the system to determine that all appropriate actions were completed. In the event of an outage, the system would determine which messages had not had all actions completed. These messages would be retrieved automatically and reprocessed. This action resulted in many messages being retransmitted to recipients who had already received then once. However, as the system became more accurate and the outages became less frequent, this problem disappeared. Later a program was written to determine which actions were yet completed and the system would not repeat those actions again.
12. The system would consist of two main frames which were duplicates of each other and would run in a parallel mode. They were called ADX A and B. They were installed in such a way that if one system went down the other system would automatically take over. Each main frame consisted of 64k of core memory. This automatic take over system was eliminated when we decided that we would take a quick program dump to determine the cause of the failure, so that the programmers could make corrective patches to the operating system. As a result, the software became very accurate and failures became very infrequent.
The system consisted of about 13 separate programs controlled by an "Executive Program." Experts who reviewed it after cut over declared it to be 15 years ahead of its time and at the cutting edge of communications technology.
Many of the operational concepts were taken from a system being designed by the Military. However, the ATS was the first system of its kind on line. The military system did not come on line until over a year later.
The contract specified that the system was to be built in 16 months. The contract also stipulated that liquidated damages would be charged for each day the cut over from the old to system to the new system was delayed. The system developed unexpected difficulties particularly in the software and the cut over was delayed. At the time, the contractor was charged about $2,000 per day. It was finally agreed that the system would be cut over on July 25, 1967.
The contracting officer and his representatives met monthly with the representatives of ITT. We alternated locations for the meetings, one month we met at the Department and the next month at ITT in Paramus, NJ, where the system was being assembled and tested.
The contract also stipulated that the system would run "error free" for 30 consecutive days. If a failure occurred, the system was shut down, the problem was fixed and the 30-day countdown started all over.
We had a "cut back" plan to go back to our old system on the sixth floor, but the plan turned out to be flawed because the ATS was so incompatible with our old system that a cut back was not possible.
Furthermore, the ATS with all its faults was infinitely better than the old system. There were 25 cogent deficiencies in the ATS and I refused to accept the system. Finally, a meeting was held with the ITT Defense Communications Division President, Rand Aroskog and the Deputy Under Secretary of State for Management, Idar Rimstead. It was decided that we would accept the system with contingencies and we began to pay rent. The liquidated damage charges were negotiated and were suspended. ITT agreed to make a maximum effort to eliminate the deficiencies.
Other elements of the contact were that ITT would train Department personnel as programmers.
We also contracted with ITT to maintain the hardware.
Authors Note: I will write separate sections in the coming months to explain the training and other aspects of the ATS.
First Communications Center to use video screens to analyze telegrams for distribution
First Communications Center to use a high speed tape reader to input data to a communications switch
First Communications Center to use an OCR to input telegrams into an automated switching system
First Communications Center to interface directly into an automated reproduction and collating system
First Communications Center to use Random Access Computer Equipment (RACE) to store and retrieve telegrams
First Communications Center to automatically format ACP 127 protocol using JANAP routing indicators
First Communications Center to automatically scan text of telegrams to determine distribution
One of the requirements of the ATS contract was that both incoming and outgoing telegrams would be displayed on a video screen equipped with a keyboard to permit text editing. Software was developed to electronically scan the text and determine the suggested distribution to be made of the telegram to the various elements and Bureau's of the Department of State. Telegram analysts would also review the substance of the text and retrieve any references to determine what additional distribution should be made.
The Analysts ability to retrieve references was limited to thirty days. Almost all references to previously sent or received telegrams were less than thirty days old. In this way, the storage capacity was reserved for the most current traffic. The distribution of the reference was displayed so that the analyst could apply this data into their distribution decision of the current telegram.
Software was written to determine the appropriate number of copies of each telegram. This determination was listed on the telegram together with the distribution symbol indicating the organization to which the telegram was to be distributed. The total number of copies, 'R' for outgoing telegrams, and 'W' for incoming telegrams, was also shown. This indicated to the reproduction center how many copies and what color paper was to be used for reproduction. 'R' for red or pink paper and 'W' for white paper.
The Analysis Section was unique in Communications Center operations. It was the first Center to come "on line" using this concept. It proved to be extremely successful in expediting and processing telegrams. Traffic volume varied from day to day, starting out slowly in the beginning of the week and ending heavy, about 1,000 messages. Messages were displayed by order of precedence, FLASH, IMMEDIATE, PRIORITY, and ROUTINE. Before the ATS, the Friday volume was frequently not completely distributed until early Monday morning. After the ATS, the volume was cleared by mid-Saturday morning.
The Analysis Section was located in a room within the enclosure and separated from the rest of the Center. Because of the nature of the Analysis work, I believed that the Analysis room should have an atmosphere of a library. So, I requested that wood paneling be installed on the walls. Also, because no one had done this before, we did not know what the appropriate ambient lighting levels should be. We installed rheostats to control the levels and zoned the lighting so that the analysts could adjust the ambiance to any level that was comfortable for them.
When the room was being constructed, one of Jack Coffey's assistants informed him that I was using wood paneling in the analysis room and he believed Congressman John Rooney, who sat on our appropriate committee, would object to the apparent opulence. Jack asked me to explain my reason for the use of the paneling. I told him that installing acoustic tile, which was in the rest of the Center, would be more expensive. Jack said, "I want an atmosphere of austerity and I don't care what it costs." So, the wood paneling was changed. When the Congressman visited the site, he noticed that we had a wood railing leading to an emergency exit. He suggested that we use metal. We removed the wood railing and installed metal at a price increase of 10% to 15%.
We had another problem with the Analysis Section. After several months, the Analysts were complaining that they were experiencing pink vision. They said the signs on the walls of the Department, which were gray with white letters, were now gray with pink lettering. At a meeting with the Analysts concerning their problem, I remarked that the building was changing the signs from gray with white letters to gray with pink letters. One analyst commented, "well then how do you explain the pink line running down the center of Shirley Highway." Many of the Analysts went to their Ophthalmologists. They were assured that this phenomena would not affect their eye sight permanently nor their ability to distinguish colors. We were very sensitive to this issue and to avoid publicity in the Department, we asked our friends in CIA to investigate. The investigation revealed that the "Pink Vision" was the result of the type of phosphor used on the video display screens we were using in the Analysis Section. The phosphor was gray and yellow, because this combination had less fatiguing affects on the eyes for long viewing. This made the screen appear a light green with yellow letters. This was discovered by the Air Traffic Controllers and was noted in the early TV sets. After a while, the complaints disappeared as the Analysts became accustomed to the screen and the "pink Vision" disappeared.
During the distribution process, the ACP 127 routing heading together with the JANAP routing indicators, which were used to route telegrams through the various relay centers in the network, were stripped off and a different distribution format was inserted by the software.
The Analysis section was very successful and after several months the personnel became more comfortable with the equipment. At first, they were reluctant to release the telegrams because upon release they were transmitted to high speed printers in the Reproduction Section, outside the enclosure. It was inconvenient to go the distance to correct the distribution if a second though occurred to them. As the Analysts gained confidence, the dwell time on the screens was rapidly.
One of the requirements of the ATS contract was that any telegram previously transmitted or received could be retrieved within three seconds. The telegram could be retrieved either by message reference number (MRN), message continuity number (MCN), or channel number. The analysts retrieved a message received or transmitted during the last 30 days by MRN. The system supervisor retrieved by channel number or by MCN to respond to service messages requesting a retransmission of the message that did not reach its destination or arrived garbled.
At the time of the development of the ATS, other mass memory devices were slow or very expensive. The RCA 3488 RACE units were used by the California State Police and were reputed to be reliable. Therefore, the Department stipulated that these units would be provided for the ATS by any contractor successful in the procurement.
The units consisted of 4" by 13 ½" plastic electronic cards. The cards were installed in an edgewise position in trays. Each card had a distinctive index tab. The RACE software was written to accommodate three indexes for retrieval by the MRN, MCN or channel number.
These cards, when selected, were quickly transferred along a channel and wrapped around a capstan to become in effect a small magnetic drum. The data was transferred quickly to a large magnetic drum, which was used for intermediate storage of the telegram during processing. From there the data was sent to the source of the retrieval request.
The RACE did not fully comply with the terms of the specification concerning the three second retrieval time. The best time was about nine seconds. However, it was faster than anything on the market at that time and was very reliable, so we accepted them.
There were two RACE units for redundancy. The requirements for retrieval placed considerable stress on the equipment and they were used far beyond their design capacity and capability.
All telegrams while being processed were stored on very large magnetic Drums. I don't recall the access time, but I believe it was in nano seconds. Software was written to protect the drums from crashing due to overload. Therefore, it was necessary to monitor the volumes very closely. To do this, a journal was printed out each minute to a Model 28 Teletype Roll-Around located at the system supervisor's position. The journal reported the time, messages in process, and drum cells in use during that minute. If the drum cells in use exceeded certain levels (8000), service messages were sent to all units and circuits to hold traffic. If the drum cells continued to rise to a level of 8125, the system was set down for all but FLASH traffic to prevent loss of data. This action stressed the relay stations and required them to store the traffic for the Department, until the drum cells in use were reduced to what was considered a safe level (below 6000). Some of the supervisors were overly protective. The late Rod Bishop, a very competent supervisor on the evening shift, had a practice of setting down the internal operations, particularly the analysis section, as soon as he arrived, so that drum cells could be purged, regardless of the drum cells in use. The Communications Center had an extensive intercom system to coordinate the centers activities. One station was on my desk.
After several evenings of this action, I inquired of Rod as to why he stopped the Analysts from working. His rationale was that if he purged the system for about and hour, the Analysis could then work uninterrupted for perhaps the rest of the evening. I disagreed with this approach because not only did he stop the Analysts from working, but the reproduction center was also shut down and the traffic became backlogged in that section as the outgoing traffic was sent to the circuits. I believed it was a better approach to permit the Analysts to operate for short periods, as necessary, which permitted the reproduction center to keep up with traffic volumes.
The ATS was delicate in many respects and in time we became more experienced with it. Programs were also written to relive some of the sensitivity.
As stated previously, the ATS was fully redundant. Therefore, we had two very large magnetic drums. After about a year of operations, ITT recommended that the drums be resurfaced. A spare drum was installed on the system and one of the drums was shipped to Los Angeles for resurfacing. Because the drum contained highly classified data, a program was written to over write the drum many times with X's and O's repeatedly on the surface of the drum. At the insistence of NSA, the drum was moved under armed guard and must remain under constant guard. We bought two first class tickets on American Airlines non-stop to Los Angeles. Ernie Field accompanied the drum. Ernie reserved first class seats for him and the drum and give the airlines my phone number. I received a call from American Airlines asking me how Mr. Drum would like his steak?
We transported the drum to National Airport by armored car. He was met by an armored car in L.A. He accompanied the drum to the factory where the drum was placed on a lathe and the surface was stripped. The chips were collected and put in a plastic bag. Ernie brought the bag back and we sent it to NSA. I guess they may still have them stored someplace.
In the spring, in Washington, D.C., as warmer weather sets in, the air gets quite humid. After a short period of full operations of the ATS, we noticed that although the air-conditioning plenum was under the floor, the return ducts in the ceiling began dripping water. Fortunately, there was no equipment under the vents. We placed buckets to catch the water, while we studied the problem. A known factor is that warm air retains more moisture than cold air. I asked that the temperature in the enclosure be increased gradually until the dripping stopped. It did at 74 degrees. However, shortly thereafter our high speed circuit to the Pentagon became erratic. I called the Chief at the military switch and asked him if they had changed anything. He denied that they had. Our programmers met with them to sort out the problem. It suddenly occurred to me that I had changed something, the temperature! I was also reminded, when the air conditioning failed during a previous water tower failure, the first thing that went down was the high speed circuit to the Pentagon. It was supposed to operate at temperatures of up to 90 degrees and that this failure at a lower temperature could possibly be a latent defect. ITT acknowledged the requirement of the specification, but stated that after operating in a constant environment of 70 degrees, the computers and peripheral equipment had become acclimated and the only way to restore this was to go through the bays of equipment with a hair dryer to check for those components which would fail and replace them. This would be a long and laborious process and would leave the ATS without redundancy. The temperature sensitivity would then return as a result of a constant environment once again. We set the temperature at 72 degrees, the high speed circuit was restored and, to my knowledge this did not happen again.
A lesson learned!