How the Soviets Beat the American's Computer Game

I recently learnt that the US military had advanced information technologies for 15 years before they made any organisational changes to their approach to combat and decision-making. When I say 'information technologies' I do not just mean computers, but (military) technologies enabled by computers. Things like precision guided missiles, advanced spy planes and satellites, improved communication technologies, and so on. These all seem like things we imagine when we picture a modern military--enough movies and dramas have demonstrated the capabilities of reaper drones--but few, I think, realise how profoundly different these technologies make warfare, compared to--say--the Second World War. As above, for 15 years (from the late 1970s to just after the first Gulf War) this included the US military itself.

When the US military did decide to reorganise itself to exploit these technologies (drawing on ideas like targeting forces at key enemy infrastructure to destablise enemy organisation, rather than at weak points in lines to overrun enemies in a more tank-and-infantry orientated approach to battle), most of the organisational changes were actually taken from ideas generated by the Soviet military. This is perhaps surprising, in no small part because the Soviets did not have the advanced technologies that the US did. Somehow, despite not actually having these technologies, the Soviet military were able to conceive of a military organisational structure which was better than whatever the US could organically determine in 15 years of experimentation.

One intuitive explanation is that the tendency for central planning which defined the Soviet economy also found a home in military planning. But unlike in the economy, where it struggled to provide essential goods to Soviet citizens, this mindset flourished in the military. I do not think this is correct. It would certainly be nice to say that the US military failed to conceive of superior organisational structures because they were all just too damn individual and freedom-loving. But I'm not convinced this is anything other than cope mixed in with some Cold War prejudice.

The better explanation, I think, is this. It is precisely because the Soviet's lacked the advanced information technologies of the US that they had to think much more seriously about how information technology could be used--in particular, used against them. Without doing so, the Soviets could neither develop an action plan for how to (eventually) reform their own military (the Soviet Union was moving into its late reform period under Gorbachev), nor develop countermeasures to the US technological advantage. It was not viable for the Soviets to pin all their hopes on catching up with the US technologically, or even to hope that their conventional superiority would be enough to counter an increasingly electronic US military. They needed to predict how the US was likely to deploy their newfound technological advantages, identify the weaknesses within those deployments, and develop strategies (within their existing capabilities) to exploit those weaknesses.

By contrast, the US did not need to think so seriously about how their new information technologies might be used. In part, the comfort of knowing the Soviets lacked such technologies likely dampened anxieties to develop new organisational structures quickly. But this belies the wider point that the US did not need to worry about being a victim of these new technologies, and thus did not need to think about how to fight an enemy with these technologies. As above, this problem is--in my opinion--what enabled the Soviets to develop better theories about how information technology could be used despite not actually having these technologies available to them.

Now, I am quite sure this narrative is an over-simplification of actual events (I am using Dmitry Adamsky's chapter The Two Generals in the New Maker's of Modern Strategy as my point of reference). Militaries are extremely complex organisations attached to even more complex organisations, governments. But I find this example interesting because it highlights how simply having technology does not mean good things come of it. It also highlights, interestingly, how understanding how technology can be used is a quite different problem from simple experience with technology. The key modern example I could cite would be DeepSeek, the Chinese AI company which has been able to innovate various efficiencies in large language models despite having only limited access to advance graphical processing units (GPUs). As above, the story of DeepSeek is more complicated--we do not know how many GPUs they actually have, nor what involvement state enterprises may have had in supporting the initiative. But that's largely besides the point. We do know that DeepSeek had more limitations than rivals like OpenAI and Google, and yet they created a system of comparable performance. So, necessity is the mother of invention?

Well, not quite. The problem with the DeepSeek example is it was about recreating US AI products under constrained circumstances. I'm sure such an exercise will benefit the company in ways which are right now difficult to determine. As above, the Soviet example highlights how understanding how a technology can be used is different from experience with a technology. DeepSeek demonstrated comparable experience. They have not come forth with striking new AI products, or a radically different approach to AI, or a blueprint for how AI can most effectively be integrated into companies and societies. To be fair, no AI company has. That's the major problem faced by the industry. And this all leads me to my point in this post.

Right now, I see so many people, businesses, and governments racing to introduce AI into their activities. The genius business move by OpenAI was to make LLMs so easily accessible to people. But this ease of access is taking resources away from the critical questions of how should this technology actually be used? I have no doubts that had the Soviets had access to comparable technologies, they--like the US--would have spent more time kitting out their army than asking how should the army change? But their lack of access forced them to confront this question in the search for a strategy given their technological disadvantage.

Today, businesses are in a panic. AI is there. We are all told it is transformative. Revolutionary. We must adopt AI, and if we do not, tell me why you want our business to fail. Etc. The inevitable result is a whole lot of technology is and will be deployed to do wholly wasteful things. Shareholders will seek justifications which will prompt strategic reorganisations whereby people are forced to use AI to manufacture evidence that an executive's ill-informed decision to buy the tech was actually a brilliant business move.

In a previous post, I noted the idea of the technology J-curve. This concept argues that technology takes a while to become useful because we need to experiment with it to figure out what is best. I'm quite sympathetic to this argument. But what the Soviet example prompts me to wonder is whether the idea of experimentation is a bit over-emphasised. Would the Soviets have developed their theories of reorganisation even faster if they could have gotten their hands on US information technology? Maybe. But this matters less than the apparent reality--that even without the technology, the Soviets developed better ideas than those American who were, seemingly, 'experimenting' with the tech. Thus, I think want matters when it comes to 'experimentation' and technology adoption is less that you have the technology, or even understand it, but that you have a clear grasp of what your organisation is trying to do, what capabilities it has, what pressures it will face, and so on.

It is telling how many 'wonderful' AI applications seem to have been wholly discovered on the fly. How many are based on feeling, or how few seem to have been undertaken with a clear grasp of the productivity impact or value-added involved. It is distressing how so few people seem to grasp that productivity is not about doing something faster or cheaper, but about what you do with the resources that you save. This is all to say, the benefits of technology flow from a clear, effective understanding of the organisation. Free from the blinding light of new technology, I think the Soviet military could see more clearly their organisation's objectives, capabilities, and pressures. I am wholly cynically that those who are selling AI training courses, or more damningly, their 'expertise', can see these things clearer than a canny manager or coal-faced employee.