#6 National Power and its Technopolitikal Domains
Tech and National Power, India's Semiconductor Ecosystem,Space Stations, Outrage against the machine, AI for protein folding, 5G Standards
Matsyanyaaya: Technology and National Power
— Pranay Kotasthane
To say that technology impacts geopolitics is to state the obvious. There are new books and articles written every day about how technology X might impact State Y’s politics. Yet, there are surprisingly few frameworks that precisely articulate the ways in which technologies can increase or decrease national power.
It is for this reason that I found a recent CSET Report National Power after AI interesting. The central argument of the report is that major, widely used technologies — such as AI — alter the power of states and societies in a fundamental and non-linear manner. This means even previously disadvantageous factors that held a state back may become advantageous in the new setting. For example, the application of AI would require large and varied data points, favouring authoritarian states which already extract such information from their citizens.
Keeping the AI angle aside, the report identifies three ways in which major innovations impact national power:
“First, innovations introduce new elements of power. Major innovations, in changing how states generate power, can create new factors that must be considered in characterizing power. For example, the advent of railroads, internal combustion engines, and nuclear weapons dramatically increased the importance of a state’s access to steel, oil, and uranium, respectively. New factors, however, are not only limited to materials. They may also encompass characteristics of a society’s culture, organizations, or economic activities.”
When we talk about technology and geopolitics, it is usually this first level of impact that’s being referred to. An example that comes to my mind is how ASML’s mastery over Extreme Ultraviolet Lithography (EUV) increased Netherlands’ national power and made it an important — even if unwilling — player in the ongoing US-China confrontation over semiconductors.
“Second, innovations change the importance of existing elements of power. Major innovations also change the “coefficients” of existing elements of power, causing them to matter more or less than before. For example, Mongol light cavalry, modern navies, and ballistic missiles all changed how geographic barriers affected one’s balance of power with geographic neighbors, eroding the effectiveness of simple remoteness, oceans, and armies still in the field, respectively, as shields against coercive power. Industrialization meant the inventiveness of a nation’s scientists and engineers became more important.”
At this level, the example that comes to mind is how the technology to blind a state’s satellite can reduce the military effectiveness of that state.
“Finally, innovations alter states’ intermediate goals. Perhaps least obviously, major innovations sometimes broadly alter what policies states pursue, by making certain kinds of behavior more valuable or less costly. While states retain the same ultimate ends, such as securing survival and prosperity, the intermediate, instrumental goals they pursue to reach those ends may shift. This can drive dramatic changes in state goals and policies. For example, before the Industrial Revolution, potential productivity gains in areas like agriculture and manufacturing were small and stable; this made conquering territory a primary means by which one group could increase its wealth and security. During and after the Industrial Revolution, modern states could also pursue substantial military and economic growth by applying new technologies to increase productivity.”
This is perhaps the least explored impact of technology. An example: in the current context, engaging in information warfare — both domestically and internationally — is one behaviour that states find tempting.
This three-level framework is a good starting point for High-tech geopolitics. In subsequent editions, I’ll explore different facets of this argument.
SiliconPolitik: What’s India Good at, Really?
— Pranay Kotasthane
With so much talk around the semiconductor shortage, it is useful to analyse where India stands in this sector. With that intention, four of us got together to do a Strengths-weaknesses-opportunities-threats (SWOT) analysis of India’s semiconductor ecosystem. The result is a discussion SlideDoc that we think should be useful for policymakers and foreign policy analysts alike.
Specifically, we find that:
“India’s primary strength lies in its vibrant integrated circuit (IC) design ecosystem with a highly experienced talent pool. However, weak research & development (R&D) focus, prohibitive costs of acquiring intellectual property (IP), and limited start-up capital have inhibited the potential of local design houses.
In semiconductor manufacturing, misplaced policies prioritising capital intensive leading-edge nodes have led to several false starts. The real opportunity for India lies in trailing edge node fabs and speciality fabs.
Finally, in the absence of backward linkages with fabrication plants or forward linkages with Original Device Manufacturers (ODMs) or Original Equipment Manufacturers (OEMs), doing business in the Assembly, Testing, Marking & Packaging (ATMP) segment in India becomes prohibitively expensive.
We recommend that India should strive to create a world-class fabless ecosystem by facilitating domestic design IP creation. The ATMP market is gradually becoming R&D intensive and the demand for product conceptualisation skills is increasing. India will have to align its skilling policies in alignment with the industry. Further, we suggest that India “looks outward” and leverages consortiums like the Quad to pool in resources, jointly invest, and conduct trade to obtain critical access to materials, technological know-how, and markets for semiconductors.”
Do give the full document a read and send in your comments to us.
If the content in this newsletter interests you, consider taking up the Takshashila GCPP in Technology Policy. It is designed for technologists who want to explore public policy. By the end of this course, you will be able to use a #ResponsibleTech framework to systematically understand the ethical dimensions of technology advancements.
Intake for the 30th cohort ends on 28th August. To know more, click here.
Antariksh Matters: Space Stations Edition
- Aditya Pareek
Cosmopolitik: Russia’s “Nauka” rocks the ISS
Russia’s newest module for the International Space Station(ISS), Nauka(“Наука” in Cyrillic) was launched on 21st July. Nauka successfully docked with the Russian section of the ISS on 29th July but not without some high-octane series of events. An unexpected firing of Nauka’s thrusters, according to NASA, caused the ISS to “lose altitude control” or “tilt” to 540’. The situation was serious but not catastrophic or posed a danger to the ISS’s inhabitants.
In simpler terms, the entire ISS spun “one and a half times” and ended up upside down, but with mitigation measures, in the end, it returned to its correct position.
Other thrusters on Russia’s older Zvezda Module and Progress Cargo ship, which are also docked to ISS, tried to correct the spinning ISS until Nauka’s thrusters stopped firing -eventually setting things right without any loss to life or material.
ROSCOSMOS, the Russian state space company, clarified that the unexpected, unintended firing of Nauka’s thrusters was due to a software glitch.
The saga is interesting because earlier this year, Russia and ROSCOSMOS had publicly announced plans for their own separate space station and were contemplating quitting ISS by 2025. With Nauka in the picture, the latter is unlikely at least in the near future. Russia’s concern that the ISS is fairly old and may pose a risk to its crew, in the long run, is not unfounded. The risk can be gauged from the increased amount of upkeep and maintenance work on their modules in recent times.
Nauka, true to its name, which in Russian literally means Science is a research module. According to this TASS report, the module will also soon house greenhouse facilities able “to grow plants on an industrial scale.”
The Russian version of the same report naturally has more details and is worth checking out with even a machine translation.
Dragon in Orbit is a Soviet Imitator
It is also important to note that, earlier this year, China put into orbit its first module called Tianhe for its separate space station dubbed Tiangong. Tiangong is significantly smaller than the ISS, and much more comparable to the former Soviet space station Mir, both in form and function - especially the module by module assembly in orbit aspect.
Cyberpolitik #1: Outrage against the machine
‘Outrage as a business model’ that’s the headline from a recent article investigating The Daily Wire’s use of Facebook ‘to build an empire’ [Miles Parks - NPR]. The article demonstrates that The Daily Wires receives significantly higher engagement per Facebook post than mainstream outlets like The New York Times, The Washington Post, NBC News, CNN and Fox News, as well as conservative non-mainstream counterparts like Breitbart News, The Blaze and The Western Journal (in general, as per the article, conservative non-mainstream counterparts receive higher engagement on Facebook than mainstream sources)
One quote, in particular, got my attention:
“ (The Daily Wire) has turned anger into an art form and recycled content into a business model.”
Some insights on outrage as a business model can be gleaned from The Outrage Industry by Jeffrey M. Berry and Sarah Sobieraj. They define outrage as a genre (it was first published in 2013, so the focus, understandably, is on talk radio, tv news and blogs).
They distinguish it from emotion (emphasis added):
“What distinguishes this type of discourse is not that it seeks to evoke emotion in the political arena. On the contrary, emotional speech has an important place in political life, and many emotional appeals are not outrageous. What makes outrage distinctive are the tactics used in an effort to provoke the emotion.”
“outrage is incivility writ large. It is by definition uncivil but not all incivility is outrage. Rude behavior such as eye-rolling, sighing, and the like are not outrageous because they do not incorporate the elements of malfeasant inaccuracy and intent to diminish that characterize outrage.”
So, what it is then? They identify some attributes:
Has a discursive style to elicit reactions through ‘overgeneralisation, sensationalism, misleading/inaccurate information, ad hominem attacks, and ridicule’. It favours “melodrama, misrepresentative exaggeration, mockery, and hyperbolic forecasts of impending doom” over nuance.
Personality centred where the voice of other participants take a back seat to a single dominant voice whose worldview drives things forward.
Reactive in the sense that such content often starts out as a ‘response’ to events that need to be ‘unpacked’ or ‘reinterpreted’.
Ideologically selectiveness follows from 3 in that the dominant actors from 2 can choose/define what they react to.
Engaging since it is essentially a performance.
Marked by ‘internal intertextuality’, i.e. outrage content producers frequently refer to one another.
Rely on oversimplification to communicate.
Pause here for a second and think about how much content we come across today checks many of these boxes, even the things we agree with.
What’s changed between the past and now? The increase in the number of ‘venues’ where one can express outrage, speed of circulation and the interplay between mainstream news and outrage venues which react and respond to each others’ reporting.
They also stress the supply-side of outrage-driven content, the changed dynamics of which are attributed to the fragmentation of audiences. In a public sphere with few venues, the incentives of content producers are (generally) to offend the least amount of people possible. In a fragmented public sphere where the aim is to reach ‘niche’ audiences, that may no longer apply (bullets added).
“structural changes we describe have rendered outrage politically and financially profitable, whether those profits appear in the form of
increased advertising revenues (linked directly to ratings and traffic)
political support, coming in the form of votes, increased support for policy positions
increased membership in advocacy groups.”
Put another way, tribalism (of a certain kind) seems to bring profits for outrage-driven content producers. Note: they do clarify that fragmentation is not the only factor - social, cultural and political forces also shape what kind of content is ultimately financially profitable.
But, there is a demand-side to it as well (that doesn’t absolve supply-side actors). In Angrynomics, Eric Lonergan contends that there are 2 sides to ‘public anger’ - moral outrage and tribal rage :
Moral outrage: The positive form which seeks to draw attention to a problem that needs to be fixed.
Tribal rage: Negative form that wants to dominate, suppress or destroy.
(the fascinating bit about this distinction is that even the acts classifying displays of public anger as moral outrage or tribal rage are not going to be independent of tribal or partisan considerations)
Here too, there is a stress on (a subset of) supply-side actors:
“cynical politicians effortlessly play on both forms of anger to garner support.”
One can reasonably argue that it is no longer just politicians who do this.
On why outrage works, Berry and Sobieraj say:
“It works because its coarseness and emotional pull offer the “pop” that breaks through the competitive information environment, and it works because it draws on so many of our existing cultural touchstones: celebrity culture, reality television, a two-party system, as well as the conventional news and opinion to which those in the United States have become accustomed”
They also refer to the collapse of local news, which is fairly common in any literature that tries to make sense of our fraying social norms [Sample: Murtaza Hussain - The Intercept, or the U.S. Antitrust Subcommittee Report]. Yet, not all of these will make sense in every context. In India, we’re certainly not a two-party polity, and while there is concern about the viability of news media business models, the specific role of ‘local news outlets’ seems to be underexplored (also, how does one define local, city-level? state-level? based on language?). So while I’m not sure I agree with all the attributes listed in the last quote, I do agree with this:
“Recognizing the economic underpinnings of the genre is vital for a more complete understanding of its prevalence, as these insights advance our ability to recognize the phenomena as culturally and politically dependent, but not reducible to culture or politics. Without this lens, the repetition of outrage discourse across media platforms can be read erroneously as an indicator of a landmark shift in political orientation on the part of the audience or of profound cultural intolerance and insularity.”
They do list 2 caveats of sorts, though:
It isn’t necessary that advertisers will dictate content choices. In fact, they frame this as a ‘narrow view’ (while it probably holds true for most advertisers, but there is scope to consider how it can impact choices/incentives when there is a heavy dependency on a subset of advertisers).
Commercially driven media will not always lead to adverse outcomes for democracy. Rather they are indifferent to it (a recent paper advocating for the study of collective behaviour affected by digital communication networks [see Technopolitik 5, Cyberpolitik #2] to be considered a “crisis discipline” made a similar point about the indifference of business models).
This post is adapted from MisDisMal-Information 46
Biotechpolitik: AlphaFold - AI for protein folding
DeepMind, the AI arm of Google, made codes for its neural network ‘AlphaFold’ available to the public in July 2021. By doing this, a powerful AI tool to predict protein folding has become available to the global scientific community. DeepMind was a British tech startup that Google acquired. It shot to fame when one of their neural networks, AlphaGo, beat the European champion at a strategy-based board game called ‘Go’. AlphaFold, another product of DeepMind, is a computational proteomic toolset to complement the genomic revolution of the last two decades, during which DNA sequencing had become cheap, fast, and accurate.
However, the determination of protein folding from the corresponding DNA sequence using conventional methods continues to be a laborious and expensive process. Various computational methods were being used to address this issue. DeepMind, using its expertise in neural networks, has provided the global scientific community with one of the fastest protein folding prediction tools. It can predict protein folding in a matter of hours to days, a significant improvement from the years it took previously using physical methods.
Solving the protein folding problem
Proteins are the molecules responsible for all biological activities, from lending structure to a cell to performing biochemical reactions. A protein is made up of several linear chains of amino acids called polypeptides. This sequence of amino acids can be easily inferred from the corresponding DNA sequence of its gene. But this data is of relatively little value as what determines the function of a protein is its unique 3D structure. Predicting protein structure is challenging as a polypeptide chain can theoretically fold onto itself in 10300 ways. Moreover, two polypeptides can interact with each other to form even more complex structures. Therefore, predicting protein structure from a linear amino-acid sequence is a computationally challenging problem. Given the lack of advanced computational tools, researchers have resorted to experimental techniques like X-ray crystallography, cryo-electron microscopy, and Nuclear Magnetic Resonance (NMR). These techniques require specialized equipment and are time-consuming.
AlphaFold is an algorithm that predicts the final protein structure using deep learning models. It uses a two-step approach. In the first step, two deep neural networks were trained on roughly 100,000 proteins whose structures are already known. One neural network was trained to predict inter-amino-acid distances, while the other was trained to predict the angle of joints between consecutive amino acids. In the second step, a gradient descent algorithm was used to optimize these parameters to best match the results from the first step.
The hardware running the two neural networks uses approximately 16 TPUc3s that is equivalent to 100-200 GPUs, a relatively modest hardware requirement.
Along with the source code, DeepMind has also released structures of ~350,000 proteins predicted using AlphaFold. This includes all proteins encoded by the human genome and proteins of model organisms used in research. The protein database is being maintained in conjunction with the European Bioinformatics Institute (EMBL).
Availability of the source code will enable researchers to develop the algorithm further, resulting in a reduction of the time required to predict protein folding. The cost of such research will also reduce significantly as sole reliance on physical techniques currently being used will fall. However, physical and experimental techniques will still be required to validate and confirm structures that AI programs like AlphaFold will predict.
This development has more significance during the ongoing COVID-19 pandemic. Global genomic surveillance efforts have led to different strains of the coronavirus being identified. AlphaFold can aid in identifying the corresponding changes in protein structure in these new strains. This can assist in the development of better vaccines and treatment protocols to specifically target these proteins. Some US universities have started using AlphaFold for such research.
The other applications of computational protein predictions are to use this technique to develop novel proteins such as oil-degrading enzymes, heavy metal absorbing proteins etc., that are not produced by any living organisms. Such proteins can have applications in clearing out oil spills and efficient waste management.
Cyberpolitik #2: Why 5G Standards Matter
The process of setting and influencing the industry standards for emerging technologies across the globe has become a strong geopolitical tool for aspiring global powers. With the ability to completely control future prospects in a specific technology, states lobby hard for the acceptance of their backed standards in order to have the upper hand in the global supply chains and development projects of the respective sector. The arrival of 5G technology coincided with the US-China trade war, which made 5G effectively the fulcrum of geopolitical and geoeconomic rivalry between the erstwhile Trump administration and the Chinese government. With China consistently increasing its presence in the leadership positions of the 3rd Generation Partnership Project (3GPP) subgroups, the organization responsible for setting international communication standards, its influence is clearly seen with the already confirmed Release 15 and Release 16 standards. This is most likely to continue into the Release 17 standards, which are expected to be out in early 2022. But the question is, does it actually matter to a country like India if its geopolitical rival, China, has a major say in setting these global standards or is it just hollow talk?
One of the main reasons that China wants to assert itself in the race to set and influence the international standards in 5G technology is that the telecommunication industry in China has had to spend huge sums of money in royalty payments towards major Western technology companies who had patent rights to critical technologies in the 3G and 4G/LTE era. China’s telecommunication industry has effectively managed to gain the first-mover advantage in the 5G race. They are keen for their homegrown companies to set the next standards and essentially get the bulk of the Standard Essential Patents (SEPs) with respect to 5G technology. This will result in Chinese companies and the state generating a large sum of revenue through patent licensing and royalty payments which can then be used to fund R&D in critical technologies. The China Standards 2035 project being undertaken by the Chinese government is just an affirmation of the argument that this is indeed a priority and a fast-track route into achieving global superpower status.
Should this concern India and other geopolitical rivals of China? If it should, what is the possible recourse we have? India has had to rely on standards in major technologies, mainly from Western countries. There has been a consistent effort by India to make its presence felt on the global technological standards stage, with 2020 showing a major breakthrough for the Indian telecommunication industry when it got the approval for a locally developed 5G standard from the International Telecommunications Union (ITU) named 5Gi. Now, the extent to which the 5Gi economically and geopolitically benefits India can be extensively debated. Still, the fact that India now views international standards in emerging technologies as a concrete area of geopolitical leverage shows the importance of the 5G (and other technology) standards.
The hegemony of any country controlling critical technology standards like 5G will have major repercussions on the geopolitical stage. This might result in global supply chains concentrating in favour of whichever country has the necessary patents and has set the standards for using said technology. It is imperative that the discussion of international standards in critical technologies finds a place in every state's foreign policy.