#2 Fog of War
DARPA wannabes, Autonomous Weapons, Influence Operations,and Japan's Siliconpolitik move
Competitive Advantage: Everyone Wants to DARPA
— Pranay Kotasthane
Japan. The UK. Germany. Even the US. These countries are now attempting their own versions of the original Defence Advanced Research Projects Agency (DARPA). The success of DARPA’s 2013 grant to Moderna for using m-RNA to develop vaccines seems to have further fuelled the FOMO.
Will they succeed? Answering that question requires a reimagination of what ‘success’ implies in this context. Most of DARPA’s initiatives fail, by design. Had most initiatives been marketable, it would’ve only meant one thing: the agency wasn’t betting on the groundbreaking ones. Secondly, even the successful ones such as the ARPANET require long gestation periods. In essence, DARPA replicas need to be set up with the apriori acknowledgement — and requirement — that it should fail most of the time and prepare for long periods with zero successes.
That seems to be a difficult act to accomplish. The Economist (June 5, 2021) edition describes a few principles that made DARPA tick:
An anti-bureaucracy setup. From The Economist:
Whereas most (R&D agencies) focus on basic research, DARPA builds things. Whereas most use peer review and carefully selected measurements of progress, DARPA strips bureaucracy to the bones (the conversation in 1965 which led the agency to give out $1m for the first cross-country computer network, a forerunner to the internet, took just 15 minutes). All work is contracted out. DARPA has a boss, a small number of office directors and fewer than 100 programme managers, hired on fixed short-term contracts, who act in a manner akin to venture capitalists, albeit with the aim of generating specific outcomes rather than private returns.
Freedom to try and fail. This often means no ministerial oversight and more crucially, a common consensus amongst political actors that such agencies will be given a long rope.
An assured customer from within the government. Some of US’ own DARPA copies haven’t met similar successes partly because they don’t have an assured customer like the US Department of Defence ready to deploy products of grantees.
Apart from these three elements, there’s another underappreciated factor in my view: a powerful national adversary. DARPA was given the freedoms it got because of the threat the USSR posed. The narrative aspect (democracy vs communism) was no less important in getting scientists onboard on dual-use inventions.
What about India’s chances at replicating DARPA? I would wager that factors #2 and #3 are not difficult for India to manage. There is precedence for India’s national security agencies being left out of parliamentary oversight and financial audits. What’s more difficult is #1. For a government to create an anti-bureaucratic setup that pursues excellence requires immense state capacity of the kind that Indian governments lack.
Of course, Indian governments will have much less money to spare than their western counterparts. But that shouldn’t directly mean fewer risks. It only means that the areas that India chooses to focus on should be different from the ones that the US does. As economists would say, focus on the comparative advantages.
More importantly, India’s revealed preferences show that in its collective imagination, Pakistan was, until now, the most significant adversary. Managing such an adversary didn’t require cutting-edge technology. It just required us to be marginally better than Pakistan. But a much stronger and advanced PRC poses a challenge that requires India to come out strong on all fronts, including technology. Herein lies the impetus for India to be audacious.
Game of Drones #1: Autonomous Weapons Need Babysitting
— Aditya Ramanathan
In July 2020, Elon Musk, the CEO of Tesla, claimed the company would make its first fully autonomous car in a few months. Notwithstanding Musk’s ambitions, driverless vehicles still have key challenges to overcome. These include developing sensors that can operate in difficult environments and artificial intelligence that isn’t easily confused.
Autonomous weapon systems share these challenges with self-driving cars. A new study by Arthur Holland Michel, a researcher at the United Nations Institute for Disarmament Research (UNIDIR), argues that since the conditions of combat are “harsh, dynamic and adversarial,” autonomous systems will be prone to failures of various kinds.
Michel identifies four underlying data issues that increase the risk of failure. One is incomplete data, which can cause an autonomous system to “misclassify objects and activities or fail to recognize its progress towards a given goal.”
The second is low-quality data - essentially raw information that is beset by a low signal to noise ratio.
The third is incorrect or false data, which can arise out of either faulty sensors or acts of deception.
The fourth is discrepant data. These could include “edge cases” or “corner cases” as well as other anomalies that “do not fit neatly within the structured categories that human designers code AI systems to recognize or respond to.”
Much like self-driving cars that struggle to navigate poor weather conditions or fatally strike pedestrians, autonomous weapons are bound to make mistakes. Michel argues these failures are made more likely because of harsh and variable wartime conditions and the actions of adversaries. The persistent fog of war and well-known instances of military deception and concealment through history suggest autonomous weapons will have to go through much further development as well as being tested in the rigours of combat before commanders have the confidence to delegate greater responsibilities to them.
Game of Drones #2: War, Not a Game
— Suchir Kalra
While autonomous weapons need supervision, meaningful human control has become an emerging theme in the discussion around autonomous weapons. As per a paper on the operational risks of autonomous weapons, "Without a human in the loop to act as a fail-safe, the consequences of failure with an autonomous weapon could be far more severe than an equivalent semi-autonomous weapon."
Campaigns such as 'Stop Killer Robots' are increasingly arguing against the use of fully autonomous weapons and pushing for 'Meaningful human control and intervention'. However, according to research and the UN Convention on Certain Conventional Weapons, the meaning of' Meaningful' remains blurred.
While keeping Humans-in-the-loop would be necessary to maintain accountability and reduce the chances of civilian casualties, it is also crucial to realise the psychological harms faced by the operators of semi-autonomous weapons such as the MQ-1 Predator remotely piloted aircraft.
It is no coincidence that the Predator's control room in Italy and a video game room in Japan look similar, as shown above. But the outcomes of each of these are drastically different. The outcome of the Predator control room is potentially death, while the video game room's outcome is entertainment for the player. And it is this aspect of 'push-button assassination' that could potentially trigger emotional and mental distress. To quote from a scientific study conducted by Irina Komaskaya and team on Police Officers,
"The association with killing or seriously injuring someone in the line of duty approached significance in predicting depression severity, suggesting a potential contribution of those experiences to the development of depressive symptoms."
In the case of Human-in-the-loop Semi-Autonomous Weapons, the frequency of exposure to the killing, injury and death in a short span of time is likely to be far greater than in the case of police officers. The book Wired for War by P.W. Singer describes the impact of Predator control rooms on the controllers' minds. Singer mentions that such combat operations happen "24/7, over long periods of time". He describes the experiences of Colonel Downs, who was an operations director leading high altitude unmanned operations.
Besides being stressful, the repetitive action of pointing and shooting potentially also leads to a sense of emotional guilt and moral injury (also outcomes of PTSD), which makes the job even more difficult for control room operators. In an NYT article, I.S.R. wing members describe how they see more death than the 'normal' things in life. Such instances reflect the severity of mental harm in these control rooms.
In my perspective, vast exposure to death and PTSD symptoms itself makes the accountability aspect questionable. Being in such a mentally and emotionally draining environment, controllers are very likely to lose focus, potentially leading to unintended casualties. Moreover, being in a closed room far from physical combat could easily lead to the controller failing to realise and reflect upon the consequences of taking a shot.
Therefore, due to the weapon controller's environment of work and the mental/emotional impacts, there is a potential for higher human errors. Ethically, this makes it difficult to hold the controller accountable in the first place.
At first glimpse, the weapon controller room could resemble a video game room. However, the intricacies, mental and emotional effects of the events that the controllers have to witness are unlike any video game. Human out-of-the-loop autonomous weapons have an accountability and "lack of human judgement" problem. On the other hand, Humans-in-the-loop autonomous weapons can potentially have a severe mental and emotional impact on the people operating them. This is where the main challenge lies with 'Humans-in-the-loop' semi-autonomous weapons.
Cyberpolitik: How Influential are Influence Operations?
— Prateek Waghre
In its May 2021 Coordinated Inauthentic Behaviour Report, Facebook disclosed that it had taken down a network that originated in Pakistan and targeted both domestic and global audiences. The content was in English, Arabic and Pashto. An accompanying report by Graphika Labs identified five kinds of narratives, one of which consisted of content that was ‘anti-India’.
The network’s various assets/accounts on Facebook and Instagram, had approximately 800,000 followers across 40 accounts and 25 pages. There were 1,200 Facebook group members across six groups, and 2,400 followers across 28 Instagram accounts. While interesting, these numbers don’t tell us much about how effective the activities of such networks are. Commenting on the era of disinformation operations since the 2010s in his book Active Measures, Thomas Rid categorised them as more active, less measured, high tempo, disjointed, low-skilled and remote. In contrast, he said, earlier generations were slow-moving, highly skilled, labour intensive and close-range [Aside: It is worth clarifying that Graphika Labs attributed the network to a Public Relations firm and not to a state actor].
Coming back to the question of effectiveness, in his book Hype Machine, Sinan Aral refers to the concept of “lift” or the change in behaviour caused by a message/series of messages. The word ‘change’ is crucial since it implies the necessity of determining causality, not just establishing correlation. For this reason, assessments based just on the number of impressions, followers or engagement are incomplete and ignore the ‘selection effect’ of targeting messages to a user who was predisposed towards a certain course of action already. And while lift has not yet been quantified in the context of influence operations due to the complexities of reconciling offline behavioural change with online information consumption, Aral suggests such targeting is most effective when directed towards ‘new and infrequent’ recipients. Or in the electoral context, at undecided voters or those unfamiliar with a certain political issue. In other words, as we at Takshashila like to say, ‘change happens at the margin’.
At least some operations appear to be adapting. Facebook highlights a shift from ‘wholesale’ (high volume operations that broadcast messages at a large scale) to ‘retail’ (fewer assets to focus on a narrow set of targets) operations in a report on the State of Influence Operations 2017-2020. We should expect to see both kinds of operations by different actors based on their capabilities. There appears to be, at some level, a convergence with an earlier era of disinformation operations.
We have an All Things Policy episode on this topic as well.
Siliconpolitik: US and Japan pick up the Slack
— Aditya Pareek
Competition with China has put a lot of critical supply chain issues in focus. Japan and the US have been central to much of this effort. As mentioned in the previous edition, the four Quad states (the US, Japan, India and Australia) have discussed issues around cybersecurity and semiconductors supply chains, in a recent meeting.
US President Biden has spoken about increased government investment, incentives and efforts to protect and diversify the supply of vital semiconductors, pharmaceuticals and their active ingredients, as well as rare earth elements and other critical minerals since at least March-April 2021.
The Biden administration recently published an exhaustive two hundred page review document that reviews progress on securing and diversifying vulnerable supply chains in the aforementioned sectors. The document goes into great detail on the ongoing global shortage of semiconductor chips and the ways in which the US seeks to mitigate it with its own and allied efforts.
The initial round of funding that the Biden administration recommended in the document amounts to $50 billion “to advance domestic manufacturing of leading-edge semiconductors”. The US Senate seemingly also accepted the recommendation, as they have passed a $52 billion bill meant for the same purpose.
According to this article, semiconductor fabrication plants are an inherently expensive and resource-intensive investment, and a single facility may cost upward of $10 billion. While a good start, the approved $50-52 billion would be a mere fraction of the total investment needed in the sector by the US government and private entities.
For reference according to the same article, China has envisioned under its fourteenth five-year plan an investment of “$1.4 trillion over six years to secure a lead in sectors like chips, artificial intelligence and autonomous driving”. Likewise, according to this article, the EU aims to invest at least $160 billion in similar “tech investments” over a similar timeline.
Japan has been asking local company Sony and Taiwan’s TSMC to establish semiconductor fabrication facilities that can churn out chips with 20nm process nodes. According to this report in Nikkei, the plant will be able to switch between producing 28nm and 16nm processes as most of the manufacturing equipment are the same. The government of Japan is ready to invest half the required cost of the project, which is said to be worth $338 million in total.
Why Nanometres go a long way
The smaller the fabrication process node, the higher the density of semiconductors which can be packed onto a single silicon wafer. The shrinking of fabrication processes is indicative of a proportionate increase in performance per watt for the resultant chips, something everyone from hobbyist PC hardware enthusiasts to automobile giants is well acquainted with.
The general purpose computing hardware industry may rely on relatively advanced small fabrication processes as small as 2-5nm(TSMC/AMD) which are relatively less mature and susceptible to higher wafer wastage and costs. Some other industries like the global automobile sector largely rely on 28-40nm chips, the fabrication process for which is mature, decades-old and has minimal wafer wastage, thus offering higher yields and lower costs.
Japan mulls ways to expand local chip production
According to another report in Nikkei, the Japanese cabinet is reviewing a strategy document that will address the semiconductors sector.
The Japanese government and the ruling Liberal Democratic Party(LDP) are also debating the creation of a much larger fund to finance this push for increased indigenous chip production. At present, Japan has a minuscule fund of $1.82 billion for semiconductors compared to the aforementioned EU and US Funds.
With the concentrated effort between the Quad countries, it is possible that dependence on a single or few sources in various critical sectors including semiconductors can be reduced. However, unforeseen factors like a recent fire at a key fabrication facility in Japan, the drought in Taiwan, and extremely cold weather conditions in certain southern parts of the US can easily spoil much of these plans. Only a cautious approach that relies on redundancies will create long term dependable solutions.
Our Reading Menu
Wired for War by P.W. Singer
Instruments of Darkness: The History of Electronic Warfare, 1939-1945 by Alfred Price
The Undersea Network by Nicole Starosielski - A book about submarine cable systems
A Study of the ‘US Darpa Model and its Applicability to the Indian Defence Research and Development System’ by Centre for Land Warfare Studies
Autonomous Weapons and Operational Risk by Paul Scharre