IS CHINA MECHANIZING ITS GROUND FORCES?
A project I worked on at CIA’s Office of Strategic Research (OSR) circa 1977 – a directed research effort – may help illuminate how these elements – basic research, directed research, and current reporting – I’ve been talking about fit together.
The timeframe was 1976-1977. The Soviet Union’s forces based in the Soviet Far East were heavily mechanized (its infantry, artillery, support units, etc. were equipped with tracked vehicles, not just the armor units), and they had been for some time. The Chinese forces remained truck-mobile and were armed with weapons and equipment that were sometime two or three generations older than their Russian counterparts. Most estimates gave the Soviet Union a clear advantage should war break out.
Analysts watched China’s move to mechanize its own forces closely for any change. Imagery analysts, over a period of a year to eighteen months, began to see armored personnel carriers (APCs) appearing in great numbers scattered throughout China. As the trend continued, current analysts began to report to policymakers that the PLA had begun to mechanize its forces, particularly infantry units. This led the intelligence agencies to forecast how long it might take for the PLA to complete its transformation from truck-mobile to a mechanized force, and to speculate on the impact this would have on the PLA’s capabilities.
I didn’t challenge these estimates at the time because CIA was seeing the same thing happening as everyone else. I had reached a point in my career; however, where I believed accepting the conventional wisdom at face value was not always the wisest thing to do. Maybe the Chinese were moving to mechanize their forces, but testing that proposition, providing more evidence, would not hurt. We could end up more confident in the estimate and, if we didn’t, the sooner we recognized that the better. I had the progress that had resulted from the HAT TRICK effort – A multiple country effort to bring China’s ground force order-of-battle to a higher level of confidence. It originally included the US, Great Britain, and Canada thus the name HATTRICK. Soon after, however, Australia and New Zealand joined, but the name HATTRICK was maintained. DIA and CIA/OSR represented the US. I also had the China-wide imagery database for my research : A vast horde of data, but not easy for an all-source analysis to work with in any meaningful way certainly not an all-China APC project.
To study the PLA successfully, analysts depend on a wide range of sources, what I have referred to before as all-source analysis. Because of the secrecy surrounding national security issues, technical collection – primarily imagery and signals – plays an important role in any in-depth investigations. Sometimes, however, in the all the glitz surrounding these technical sources, outsiders lose sight of human and open sources, and the essential role that they play. Open sources contribute far more to intelligence products than most recognize. In the case of the PLA, their role is essential. What all these sources, technical and otherwise, have in common is that they create a data management nightmare for the analyst.
The volume is tsunami-like, and the tools to find what you are looking for in the huge national databases are either primitive or lacking altogether. Imagery, for example, pours in every day, but that is the easy part. Before imagery can become useful intelligence, an enormous amount of labor must go into finding things and recording what the imagery analyst has observed. Day after day, month after month, year after year potentially important pieces of data stack up waiting for someone to use it as part of their analysis. Except for immediately obvious news and order-of-battle research, the vast potential for strategic intelligence and trend analysis that is possible from imagery goes underutilized. Even those that try, quickly find that the recording of events does not follow established rules and standards. Even the simplest machine search cannot return a reliable result.
Two of my colleagues in China Branch, for example, found that there were fifteen different ways a Type-59 tank was recorded in the database, and that was just the tip of the problem. Every piece of ground, air, naval, and missile equipment in the database had multiple names for the same item. There was no dictionary to record and consolidate terms to facilitate accurate searches. Tom Behling and Mike Coyne, just two analysts in a small seven or eight-man cubby at CIA, set out to fix the problem.
They first sought help from the computer specialists at the national database. That got them nowhere. The experts responded that the database was fine and didn’t need fixing. Tom and Mike then turned to CIA’s in-house IBM specialists, again with no luck. Finally, they decided the only thing to do was to fix it themselves. This was a time before personal computers, and neither Tom nor Mike considered themselves computer specialists. Both, however, learned a bit of the “BASIC” programing language somewhere along the way.
Their first task was to compile a complete list of Chinese equipment currently in the database. All the sorting they did was by hand. With both working full time, it took weeks to complete this first step. They next wrote a small “BASIC” program telling the database to recognize Type-59 tank and its fifteen names as all the same item. The resulting printout provided a far more reliable indicator of what, when, and where Type-59 tanks had been observed on the imagery. They did the same for all the other weapons and equipment, including APCs. As a final step, they asked the database to report any term not already on the list for them to assign it as a new item, or to add it to an existing compilation. My project would not have been possible without Tom and Mike’s dictionary.
I took the list of APC’s sightings and correlated them with the types of units where they were found in the newly more accurate order-of-battle. As others had reported, an increasing number of APCs was being observed, and they were scattered countrywide. They were just not appearing in infantry units. Looking closely, I found APCs at headquarters facilities – probably for command and control – some armor units, and a few scout units, but I couldn’t find an APC sighting associated with a single infantry division. If China was mechanizing its forces, it must have been in the very early stages. Until tracked mobility became common in infantry and armor units, it was premature to credit China with much change in the PLA’s capability.
This was not an earth-shattering conclusion by any means. It did, however, demonstrate that an in-depth look China-wide, using a national database, was feasible and could give us a more accurate measure of China’s military modernization than current reporting.
My colleagues and I found the dictionary indispensable to our work. We used it frequently. When OSR disappeared in a reorganization of CIA, China Branch made an effort to interest DIA in keeping the dictionary current. It didn’t take long, however, for it to fall into disuse. Few probably even remember it existed. That’s a shame. For a brief time, analysts could do China-wide detailed research using the vast amount of information contained in the imagery database. To the best of my knowledge, that capability does not exist today, and worse still, there is nobody doing strategic research that would miss it. Little wonder that we made the wrong call on Iraq WMD, Indian nuclear, and Russia’s ability to take Ukraine by force. Unless we get back to doing research the next mistake is just around the corner and is the basis for my conclusion that US national intelligence is broken.