Symposium on Fairness, Equality, and Diversity in Open Source Investigations: Earth Imagery’s Colonial Legacy

[Cris van Eijk is an international lawyer researching what it means to make space ‘common’, and how international law works to effect that. He holds a BA and LLM in International Law from Leiden University, and a BA in Law from the University of Cambridge.]

Introduction

Satellite imagery is one of the most important sources of data in open-source intelligence, and today it is more accessible than ever before. This availability has undoubtedly played a part in enabling the proliferation of open-source intelligence in recent decades. Right now, you or I can load up Google Earth Pro or similar software and begin traipsing across the planet, for free. For the open-source investigator, the implications are profound.

When I grew up, I had strictly limited internet time and a map of the world on my wall. Once my time was up, I would lie on my bed and trace the borders of places I couldn’t even imagine. Then, one day, I discovered Google Earth on my dad’s old laptop. It was just like the map on my wall – except interactive, more accurate, and higher resolution; there was even a flight simulator. Suddenly, from an old desk in the suburbs, I could explore and understand the world around me in a powerful new way.

Satellite imagery is a powerful tool for open-source investigations, from projects digitally verifying human rights abuses, to tracking Russian military operations in Ukraine, to investigating modern slavery operations. But satellite imagery also has a history – the infrastructure and interests that created it have never been neutral. And, as I will argue in this blog post, satellite imagery has also had victims. 

A Brief History of Satellite Imagery

Satellite imagery is one of several types of data that can be gathered by ‘remote sensing’, which refers to electromagnetic sensing of Earth’s surface, “for the purpose of improving natural resources management, land use and the protection of the environment” (Remote Sensing Principles, Principle I). About two-thirds of active satellites in Earth orbit are licensed by the United States, and the vast majority of satellite imagery and related data are derived from operators licensed by the US and subject to its jurisdiction. The US therefore plays an important role both as a historical actor in the historical development of satellite imagery, and as its primary regulator today.

By the time the Soviet satellite Sputnik orbited the Earth in October 1957, the United States had been working to enable and legitimise the collection of satellite imagery for several years. Well before the establishment of the (then-Ad Hoc) United Nations Committee on the Peaceful Uses of Outer Space (UNCOPUOS), the US had determined satellite imaging to be a national priority, albeit for reconnaissance purposes. By UNCOPUOS’ first meeting in 1959, the US considered “one of the most critical aspects of” the Committee to be “the acceptance of the use of reconnaissance satellites as a legitimate outer space activity”, and “agreed that the discussions should be channelled so as not to limit the use by the US of reconnaissance satellites…” (Meeting of the NASC (27 April 1959)). In May 1963, President Kennedy signed National Security Action Memo (NSAM) 156, formalising that objective into policy.

Collecting satellite imagery in the 1960s required technical as well as political infrastructure. At the time, securely collecting satellite images required the US to, in the words of National Reconnaissance Director John McLucas, “build a high-priced camera, throw it into space, and dump it in the Pacific Ocean a few days later” to be manually recovered. This process was so resource-intensive because virtually every aspect of satellite image collection was classified. This included the images themselves, their resolution, the satellites that took them, the data recovery process, and even the existence of the primary federal agency responsible, the National Reconnaissance Organisation (NRO). So while histories of satellite imagery typically begin with certain American projects in the 1970s, the first chapter of that history remained mostly top secret until the 1990s, and remained piecemeal until just a few years ago, thanks to scholars like James David, Aaron Bateman, and Jonathan McDowell. 

Getting “Too Spooky”

For the United States, establishing satellite reconnaissance as legal and legitimate was the single most important objective of space lawmaking. Specifically, the US sought to persuade other states of the imminent economic benefits of what it called ‘Earth observation’ and ‘remote sensing’ programmes (official policy discouraged use of the term ‘reconnaissance’ satellites in public). Making that case to a decolonising world already seized with questions of resource sovereignty was a delicate affair. From the 1960s to the 1980s, the CIA, Pentagon, and later the NRO exerted substantial control over the US approach to space lawmaking, and therefore over its results. These three organizations saw international law as, at best, a potential obstacle, and its making as a risk – the CIA anxiously estimated how the world would respond to public exposure of their reconnaissance programs. Washington had to walk the “[fine] line between economic research photography and economic intelligence photography,” as a 1969 NRO report put it – between accusations of spying and accusations of resource-grabs.

In order to “create wider public acceptance of space observation and photography”, the US resolved in the late 1960s to publicly release a limited fraction of its satellite data and imagery. This publicity was carefully controlled, leading one senior advisor to wonder, using a colloquial term for US intelligence agents, if “we are getting entirely too spooky about our information activities relating to space.” For the US Intelligence and Defence communities, there very much was such a thing as bad publicity. “Future possible NASA applications,” a 1969 NRO report said, “must be planned and controlled very carefully,” as “casual experimentation could trigger challenges to the legitimacy of not only the NASA earth-sensing program but of the National Reconnaissance Program.”

But in this case, the problem was also the solution – publicity was the best way to ensure the widespread acceptance of satellite data collection. In April 1966, the NSAM 156 Committee concluded that “[t]here is great potential political capital in a US program of natural resource surveys and other scientific and economic exploitation of satellite earth observation and sensing,” and proposed “a major political initiative advancing the concept of economic betterment through space activities.” This strategic link between satellite imagery and resource exploitation was quickly taken up by ongoing US modernist international development projects – though, as will be explored below, not always to foreign benefit.

Towards the Promised LandSat

In September 1969, two months after Apollo 11 landed on the Moon, Richard Nixon spoke to the UN General Assembly:

“I feel it is only right that we should share both the adventures and the benefits of space,” he said. “We have determined to take actions with regard to earth resource satellites… this program will be dedicated to produce information not only for the United States, but also for the world community.”

Nixon’s announcement was the indirect result of a rogue press conference in 1966 (p. 195), during which the Department announced (without prior coordination) its plans to operate a resource detection satellite network. Nixon’s speech formally adopted those plans into US policy, at the cost of forcing space imagery collection into the international spotlight. It also marked a public turning point in the use and access to satellite imagery. 

In July 1972, the US launched the first satellite of a programme that would eventually be called LandSat. In contrast to previous top-secret programmes, LandSat data was available for public purchase, albeit at much lower resolutions and restrictions according to US national security interests. LandSat’s design was particularly ideal for detecting of mineral or petroleum resources – which were (by sheer coincidence) exactly what the US desperately needed in the 1970s and 1980s. Still, American representatives insisted, “and often genuinely believed”, that LandSat would be a beneficial tool for Global South development – even as they worked to prevent the UN from publicly debating the programme’s legitimacy. But some in Washington, particularly the military and intelligence agencies, were perfectly willing to use satellite imagery for US economic advantage. And of course, despite having access to superior, classified technology, the CIA, NRO, and Pentagon almost certainly maintained clandestine, privileged access to ‘civilian’ satellite imagery, though the full extent of those relationships have only begun being declassified.

In general, the international response on LandSat was cautious optimism, though some expressed reservations. Global South states like Brazil and Jamaica (p 106), and to lesser extents Non-Aligned States like Sweden, framed LandSat as an issue of permanent sovereignty over natural resources. They argued that collecting resource data required state consent, and that the state should have access to that data at little to no cost. The State Department responded with an all-hands-on-deck diplomatic blitz. USAID promised to award grants to fund developing state access – the programme was discontinued three years later, in 1976. As Megan Black argues (pp. 203-205), its recipients were often friendly with the US and US foreign investment. Even so, these promises helped assuage Global South and Non-Aligned concerns – by 1974, LandSat’s original critics – Sweden, Brazil, and Jamaica – were among the first to buy in.

LandSat data packages broadly fell into two categories: cheaper but less detailed raw data, or partially processed and more detailed data, at much higher cost. From there, the main barrier in practice was the cost of (further) data processing. In 1972, even by NASA estimates that data processing required a $1 million computer system and $11,500 per ‘scene’ to process, after purchase. In this sense, LandSat resource data was far more useable on a systematic basis for Global North corporations than for Global South states. And while technological development slightly lowered these costs over time, this too was felt more quickly in the Global North than the Global South.

As far as we know – this history is still being declassified – just one-third of LandSat images were of non-US territories. However, it was these images which enabled US companies to make substantial profits from Global South resources. After three years of operation, a third to half of LandSat’s global customer base were non-renewable energy companies (Black 2019, 114) – By 1978, oil company profits were estimated at over $1 billion thanks to LandSat data.

LandSat began collecting data on untapped Sudanese petroleum reserves shortly after launch – in its first year alone, it captured enough images to cover Sudanese territory ten times over. This enabled Chevron, one of LandSat’s most loyal clients, to use its superior wealth and processing capacity to prospect for oil using public data (Black 2019, 117). In 1977, Chevron struck oil in the face of a mounting global shortage – and the subsequent extractive frenzy triggered economic and political instability for decades thereafter. 

LandSat’s corporate customers worked hard to obscure their use of satellite imagery, and it is only thanks to recent work (including by Megan Black) that we have begun to understand the wider impacts of these public-private relationships in transnational and decolonial contexts. 

The Global South reacted to earth resource data in different ways, and sometimes positively (Jirout, 137). But for many Global South states, the asymmetrical access to their resource data was further reason for a New International Economic Order. Some states, including Argentina, attempted to use multilateral fora to ensure the equitable use of satellite resource data, including via UNCOPUOS (UN Doc A/AC.105/C.2/L.73) and the ITU (ITU Doc 179-E). This resulted in the nonbinding Remote Sensing Principles in 1986, which attempted to establish norms to ensure equitable data use and access (e.g. Principle XII). However, in the long term, these efforts have arguably failed to establish the hard law they sought. 

Coda: The Current Situation

The history of satellite imagery (and especially LandSat) only got more complex and twisty after 1980 – but that tale has been told elsewhere. Instead, let’s jump to the present. 

In April 2008, the United States announced that all LandSat data would be made free within a year. This, as well as the subsequent rise of non-state space actors, have significantly changed the world of space imagery. Today, companies like Planet Labs sell data that is more accurate, detailed, and affordable than ever. Indeed, Planet’s customers now include open-source media like Bellingcat and even the NRO itself. In this sense, satellite imagery is more open-source than ever.

But herein lies the clincher – that openness is not absolute. The vast majority of satellite imagery is regulated by a limited few states, according to rules that often rhyme or replicate US law. This has advantages. For example, a 2020 revision of US licensing conditions was able to streamline the existing rules for most of the world’s earth imaging systems at once, and has since triggered similar shifts in key licensor jurisdictions, like Canada and India.

Unfortunately, it also presents problems. American remote sensing licenses come with certain strings, depending on their system’s specifications, and justified under the broad banner of national security. As far as we know, the most restrictive such measure, a ‘shutter control’ order to cease data collection, has not been used – yet. But that may be because it has not yet been necessary. Instead, the US (as well as other major licensors like France) have become these systems’ biggest customers, purchasing exclusive rights to certain images and regions. This has very real impacts for us all, as beneficiaries of satellite imagery. It should concern us. Our literal worldview is being changed according to the national security interests of a handful of governments. 

It has recently become trendy again for military and intelligence officials to promise to use this particular power wisely, just like they did in Sudan. That does not change its history of abuse, nor does it repair the decades of resulting (neo-)colonial harms. At present, our most reliable safeguard is not law, custom, or principle, but the risk of foreign market competition. As Brian Weeden put it, now that there are other jurisdictions to flee to, “the United States can’t control what happens anymore.” That may be reassuring for many – that is, for those of us whose resource sovereignty has never been violated. But without a hard look to history and recognition of its harms, our view of the world is far from neutral.

Conclusion

Today, we have more public access to space and satellite imagery than any time in history – but it’s on us to ensure that their histories are similarly accessible. Part of that requires that we, as beneficiaries of satellite imagery, understand where it came from and what that entailed. It means developing a habit, and eventually a practice, of thinking critically about the satellite imagery we use.

Some of the questions that come to mind are the following. What actor(s) collected this data? What legal/regulatory constraints affected that collection? Who are the subjects of the data, including non-state actors and Indigenous peoples? Did these actors consent and/or participate in its collection? What concrete measures were taken to ensure they were able to consent or participate? What power asymmetries affected the relationship(s) between the data collector(s), regulator(s), subject(s), and user(s)? How have they been accounted for?  For what purpose(s) is this data being used? Are its subjects fully informed of those purposes and uses? And finally, but most fundamentally, how have you prioritised ‘fairness in knowing’ in the process?