The promise and perils of produced waters: intelligent trial and error as an anticipatory framework for enabling responsible innovation

ABSTRACT

Human population increases alongside climatic changes has spurred interest in technologies exploiting unconventional water resources. However, analysis of the hazards entailed by these technologies has lagged behind research on their scientific and technical facets. This paper considers one unconventional resource: the waters resulting from oil and gas extraction processes-‘produced waters.’ This paper redeploys the Intelligent Trial and Error perspective on technological risk as an anticipatory framework, examining the question ‘How amenable is the sociotechnical landscape of produced waters to learning of, averting, and responding to negative unintended consequences?’ Controversy over recycling oil and gas wastewater for irrigation in Kern County, CA is analyzed as emblematic of the cultural, technical, and political barriers to responsible produced water innovation more broadly. This article shows how Intelligent Trial and Error can enhance the Responsible Research and Innovation framework with regard to addressing the material and political obstacles to more responsible R&D.

KEYWORDS:

• Produced water
• intelligent trial and error
• responsible research and innovation
• environmental controversy
• anticipatory governance

Introduction

Growing scarcity and increasing demand drives interest in innovations to extract unconventional sources of valued resources. Water is no exception – especially in arid regions. Climate change threatens diminished flows in western rivers while desert cities expand, further taxing already challenged water supplies. Such pressures have spurred technological developments to exploit brackish, deep, and non-potable water resources overlooked by previous generations – namely waters ‘produced’ by oil and gas extraction. Produced water recycling innovations promise to help recharge depleted aquifers and stabilize supplies during droughts but also entail considerable environmental and human health risks.¹ How might efforts to tap this unconventional resource be intelligently managed so as to minimize the likelihood and scale of undesirable consequences and maximize the degree to which they are used in ways compatible with the interests of a broad range of stakeholders?

This paper examines the controversy surrounding recycling produced water for ‘beneficial’ or human uses, namely the case of Kern County, California. Applying a reconceptualized intelligent trial-and-error (ITE) framework (Morone and Woodhouse 1986; Collingridge 1992; Genus 2000), I analyze produced waters’ sociotechnical landscape: What cultural, technical, and political strategies would help ensure that produced water recycling innovation proceeds responsibly?

This paper shows how ITE fits into responsible research and innovation (RRI) (Stilgoe, Owen, and Macnaghten 2013). Strategic ITE guidance for coping with the uncertain consequences of emerging innovations is paltry, because most ITE research focuses on established technologies (see Dotson 2017). In contrast, this paper uses ITE as a forward-looking strategic framework for enabling RRI, focusing incisively on the material and political barriers to more responsible R&D (see De Hoop, Pols, and Romijn 2016; Genus and Stirling 2018).

The intelligent steering of technoscience

ITE is a framework for better managing the risks of innovation, developed via case studies of both failure and instances when innovators proceeded more responsibly (Morone and Woodhouse 1986; Collingridge 1992). Early research focused on mistakes made in developing large-scale technologies like nuclear energy and the space shuttle program. More recently, the framework has been extended to explain ‘natural’ disasters, such as Hurricane Katrina’s impact on New Orleans (Woodhouse 2007), and more slow moving tragedies, like the continued development of sprawling suburbs (Dotson 2016).

The ITE framework is rooted in political incrementalism, a 60 year old political theory founded on recognizing the limits of analytic rationality in the face of unpredictably complex policy changes² (Lindblom 1959; Woodhouse and Collingridge 1993; Genus 2000). Because analysts lack the necessary knowledge to predict the results of a policy change and are handicapped by biases, their own partisanship, and other shortcomings, incrementalism posits that policy should be – and very often is – developed via gradual, decentralized, and pluralistic processes. Evolutionary mutual adjustments among partisan groups better include diverse interests and produce more prudent and gradual policy change than expert-led decisions. In short, pluralist democracy is more ‘intelligent’ or learning-enhancing than technocratic politics. Elite decision makers in pre-Katrina New Orleans, for instance, eschewed adequate precautions, viewing flooding risks as acceptable or as less important than supporting the construction industry; enabling constituent groups to advocate for their own interests would have forced compromises leading to more prudent infrastructural policies (see Woodhouse 2007). In any case, ITE is the extension of incrementalism to technological development.

ITE differs from technology assessment, though both seek to avoid undesirable unintended consequences (see Genus 2000). Complex technologies are highly unpredictable, best thought of as experiments whose consequences become clear only after becoming entrenched (Collingridge 1980; Genus and Stirling 2018). Analytical forecasts of future risks are thus insufficient; research is not quick, complete, or unbiased enough to reduce uncertainty, especially because high political stakes can prevent that reduction (see Sarewitz 2004). Therefore, assessment needs to be complemented by strategies to enhance learning. No doubt technology assessment can help. But rather than focus on analyzing and forecasting risk, benefit, and harm, the aim of ITE is to uncover sociotechnical system characteristics that may hinder learning, preventing quicker feedback on harms, reduced damage wrought by errors that do occur, and better responses to unexpected consequences.

Expositions of ITE vary (cf. Collingridge 1992; Woodhouse 2013; Dotson 2017). I will part ways by describing the framework in terms of meeting challenges though precautionary strategies, most importantly the epistemological challenge of technological change: Can learning happen quickly and without high costs? Neglecting this challenge not only risks collective mistakes, if not negligence, with emerging technologies but can also stymie harm reducing innovations. The ills of suburban sprawl persist, for instance, because feedback comes far too late; developers stay the course because innovation ‘errors’ are clear only after building large swaths of housing (Dotson 2016). Regardless, meeting this epistemological challenge requires three interrelated kinds of precautionary strategies.

The first set of precautions are cultural. Are those involved prepared to learn? Is feedback produced early enough and development appropriately paced so that participants can feasibly change course? Does multi-partisan monitoring by appropriate experts occur? Is that feedback effectively communicated to the affected and deciders? Biotechnologists unknowingly applied ITE strategies at the 1975 Asilomar Conference: They put a moratorium on the most uncertain genetic engineering experiments until further testing, proceeded gradually as hazards and precautionary measures became better understood, and communicated the results broadly (Morone and Woodhouse 1986). In contrast, large-scale technological mistakes – from nuclear energy to irrigation dams in developing nations – tend to occur because they are led by a group of ‘true believers’ who fail to fathom that they could be wrong (Collingridge 1992).

Other strategies entail technical precautions. Even if participants are disposed to toward learning, does the technology’s design enable prudence? Sociotechnical systems can be made forgiving of unanticipated events by ensuring wide margins for error, including built-in redundancies and backup systems, and giving preference to designs that are flexible. The designers of the twentieth century nuclear industry pursued the first two strategies but not the third. Their pursuit of economies of scale combined with the technology’s capital intensiveness all but locked-in the light water reactor design prior to a full appreciation of its inherent safety limitations (Morone and Woodhouse 1989). No doubt the technical facet of ITE intersects with its cultural dimensions: A culture bias toward rapid innovation can create technological inflexibility just as well as overly capital-intensive or imprudently scaled technical designs (cf. Collingridge 1992; Woodhouse 2016).

Finally, there are political precautions. Do existing regulations, incentives, deliberative forums, and other political creations push participants toward more precautionary dispositions and technologies? Innovators may not be aware of all the hazards or uncertainties if deliberation is insufficiently diverse. AIDS sufferers, for instance, understood their own communities’ needs and health practices better than medical researchers (Epstein 1996). Their exclusion slowed the development of responsible research and treatment. Moreover, technological designs are less likely to be flexible if risk deliberation occurs too late. Finally, do regulations protect against widely shared conflicts of interest, encourage changing course or modulation, and distribute the burden of proof fairly? ‘Sound Science’ regulatory approaches force the least empowered participants (i.e. victims) to prove harm prior to passing regulation, failing to incentivize prudent innovation. In contrast, if innovators paid into victim’s funds until harm was disproven, the financial incentive would encourage precaution (Woodhouse 2013, 79). Indeed, mining companies already post remediation bonds to ensure clean up after valuable materials have been unearthed.

To these precautions, I would add the need for deliberative activities to reduce polarization (see Kim 2008, 474; Fleck 2016). Indeed, those studying environmental mismanagement have outlined how establishing trust and some vision of a common future – often through informal communication modes – are essential for effective collective governance (Ostrom 1990; Tembly et al. 2017). Strong mediation practices aided the participatory design of the ‘green’ neighborhood of Quartier Vauban in Germany (Dotson 2016). Deliberations are unlikely to lead to productive disagreement and action when overly antagonistic.

The ITE framework overlaps with RRI in several respects but also differs from it in a number of ways (Table 1). Expositions of RRI vary as much as they do for ITE. Some describe RRI in terms of the four pillars of anticipation, reflexivity, inclusion, and responsiveness (Stilgoe, Owen, and Macnaghten 2013). These pillars are reflected in ITE’s focus on learning. Trial and error is intelligent insofar as decision makers are pushed to anticipate possible failures and deleterious consequences and are made to include the interests of and be accountable to potential victims, so as to incentivize ameliorate action (cf. De Campos et al. 2017). ITE also shares RRI’s connection to deliberative democratic traditions and the precautionary principle (see Wong 2016; Reber 2018). The first key difference is that inclusion and deliberation is sought not only to broaden participation but also to enhance the policy process’ ‘intelligence’: All participants have expertise and expand the number of policy considerations. Secondly, reflexivity is associated not only with deliberative settings but is distributed across interactions between plural partisans (Genus and Stirling 2018, 63). Thirdly, although ITE aims to spur adequate precautionary steps, it recognizes that such steps may not be anticipatable in advance of actual trials with the technology. In other words, it adds the pillar of ‘incrementalism’ to that of anticipation.³

Table 1. ITE Strategies and their relation to RRI. Bolded items are RRI pillars introduced by ITE.

Intelligent Trial and Error
Epistemological challenge: How to learn about, avoid, and inexpensively mitigate errors in light of incomplete understanding of a complex reality?
Cultural Precautions	Technical Precautions	Political Precautions
Need for learning recognized?	Technology designed to be forgiving of error?	Diverse and early deliberation and decisions required?
Focus on detecting potential errors?	Safety and backup systems present and maintained?	Burden of proof fairly distributed?
Effective communication of errors to those who need to know?	Designed flexibly enough to change course?	Innovators incentivized to avoid and correct errors?
Efforts to lessen enthusiasm and groupthink?	Gradually scaled up in size/developed at appropriate pace?	Efforts to reduce polarization among stakeholders?
Overlap with RRI: Anticipation, Reflexivity	Overlap with RRI: Anticipation, Responsiveness, Incrementalism	Overlap with RRI: Responsiveness, Reflexivity, Accountability, Pluralism

ITE has much to offer RRI. Others have argued that RRI could better account for the material barriers, costs, and prevailing power structures that prevent well-meaning innovators from innovating responsibly (De Hoop, Pols, and Romijn 2016). ITE addresses those limitations by highlighting the need to ensure technological flexibility, which zeros in on a structural barrier to another RRI pillar: responsiveness. ITE also sees inclusiveness as better supported by countering conflicts of interest, and fostering diversity in decision-making and fairness in the burden of proof. As Genus and Stirling (2018, 67) contend, such ITE strategies provide ‘qualities’ for constituting RRI in the face of entrenched power and interests; most important is ITE’s focus on the irreducibility of ignorance: although anticipatory scenario-building helps, it cannot substitute for ensuring incrementalism, flexibility, and reversibility. Finally, in contrast to RRI’s foregrounding of deliberative reflexivity, ITE emphasizes the pillar of political pluralism: Innovators need not be circumspect about their innovations’ social and ethical consequences but merely incentivized or otherwise encouraged by political opponents to act as if they were. Again, ITE posits democratic intelligence – and thus also reflexivity – to be a kind of distributed cognition (Lindblom and Woodhouse 1993, 23–32).

ITE is usually applied to already entrenched technologies. But it has potential as an analytical lens for evaluating emerging technology. In the next sections, I describe nascent efforts to recycle oil and gas wastewaters for agricultural uses – namely by the Cawelo Water District and Chevron in Kern County, California. Viewing wastewater recycling as an emerging technology, via ITE, uncovers a range of oversights, uncertainties, and political discordances that must be addressed if participants are to avoid behaving irresponsibly.

Extracting fossil fuels, producing water

Produced water has myriad sources: it is ‘produced’ when pumping crude oil, separating bitumen from tar sands, fracking a natural gas well, or extracting methane from a coalbed (Clark and Veil 2009). Oil and gas deposits often overlay brackish underground hydrological structures. Extracting coalbed methane, for instance, requires pumping out water to allow natural gas to flow to the surface. For other oil resources, water is intentionally injected (e.g. ‘fracked’ natural gas wells or steam extraction of heavy oil deposits). Injecting water also maintains production rates in declining reservoirs. Hence, much more water than oil is produced by wells nearing the end of their lifespan. Indeed, increasing injection rates signal California’s oil production decline (Tiedeman et al. 2016). In contrast, produced water rates peak during the early months of fracturing a natural gas well, declining thereafter. The volume produced is often immense. In California a hydraulically fractured well might produce 39 million gallons of water, while a conventional oil well averages nearly 16 gallons of water for every gallon of crude.

Although the composition of produced waters varies from well to well, there are a number of typical contaminants (see Shaffer et al. 2013). The amount of total dissolved solvents (TDS) may run from 1000 to 400,000 milligrams per liter (saltwater runs 35,000 mg/L), potentially including salts, heavy metals, and radioactive isotopes. Also routinely present are organic molecules – oil and grease – as well as production chemicals. Such contamination often makes the treatment of produced waters technically challenging and expensive. And most of the promising technical solutions that remain under investigation are highly energy intensive.

What produced waters mean for oil and gas firms and other social groups depends on a range of factors. Significant for oil and gas firms are the economics of various treatment and disposal options, which is influenced by the regulatory environment (Hagström et al. 2016). Produced waters are recycled for further well development in the Marcellus Shale region because the region lacks disposal injection wells and municipal treatment facilities will not accept untreated wastewater (Shaffer et al. 2013). The sociopolitical context can drive firms to view produced water as a resource rather than a waste stream. In any case, most oil and gas wastewater in the United States is currently injected underground, either to dispose of it or to ‘enhance’ recovery of fossil fuels.

Geography further combines with disposal costs and regulations to shape the perception of produced waters. In the drought-prone Southwest some look to treated produced waters – as well as other brackish, low-quality aquifers – as a potential ‘diversifying’ resource; because climatic changes may decrease the magnitude and reliability of flows in the Colorado, Rio Grande, and other rivers, it is hoped that technologies used to tap and treat unconventional water resources could help stabilize supply and recharge aquifers (New Mexico First 2014, 57).

For oil and gas firms, produced water has potential as a political resource: Recycling and hence ‘saving’ water allows them to demonstrate their green credentials. In the same way that BP covered their gas stations with solar panels – albeit pointed downwards – in the 2000s to become ‘green’ (Zehner 2012), the recycling of produced waters could be used to ‘greenwash’ fossil fuel extraction. Consider a National Geographic article that framed recycled produced waters as ‘green fracking’ (Kiger 2014).

Such meanings are not guaranteed. Produced waters and their concomitant technologies exist in a state of considerable regulatory uncertainty. To begin, produced waters are not considered to be hazardous waste under the Resource Conservation and Recovery Act, and the EPA exempts any waste (e.g. highly saline condensate) resulting from water treatment (Romo and Janoe 2012). State regulation of produced waters can fall to very different institutions – from the Railroad Commission in Texas to the Oil Conservation Division in New Mexico. Moreover, there is ‘no clear law regarding who owns produced water treated for beneficial uses’ (8). Such uncertainty is especially problematic in arid regions, where a clear adjudication of access to water becomes important during droughts. In New Mexico, produced waters’ categorization as ‘non-potable’ and ‘deep’ exempts them from usual water regulations (Bossert, Olson, and Bushnell 2015), a categorization that seems arbitrary as technical advances and growing water stress transform such waters into potential resources (see Hagström et al. 2016). From an ITE standpoint, there is considerable reason for caution. Indeed, nascent efforts in Kern County, California foreshadow future controversies.

Produced water irrigation in Kern County

Kern County has historically been defined by agriculture and oil and gas extraction (see Reisner 1993; Miller 2011). It ranks second in California in agricultural output and produces 71% of the state’s oil – largely from heavy oil deposits (CDFA 2015; Cooper and Sedgwick 2015). Kern County agriculture, like in much of the West, relies upon intensive irrigation, in turn enabled by massive engineering works and generous public subsidies. Kern County ranks only behind the Los Angeles Metropolitan Water District in water received from the State Water Project: the multibillion dollar collection of dams, reservoirs, and canals leveraging billions of kilowatt-hours of energy to move Northern California river water to southern metropolises and the irrigation-dependent farms of the Central Valley. The oil fields of Kern County are just as dependent on scarce water resources as its farms. Extracting heavy oil itself requires injecting heated water, and reservoirs both demand and produce more water as they are depleted. A typical Kern River oil barrel requires 3.5 barrels of injected steam and ‘produces’ another 10 barrels of water (Harris 2015).

Although produced water has reportedly flowed through irrigation ditches since 1994 in Kern County’s Cawelo Water District (Waldron 2005), its use has expanded only recently. The drought that gripped California between 2011 and 2017 played no small role. Indeed, by 2014 recycled water from Chevron’s Kern River Oilfield accounted for half of Cawelo’s supply, spurred by Chevron selling recycled wastewater for a fraction of market price (Onishi 2014). An unusually low level of contamination (i.e. low TDS) of the water produced in this oilfield allowed Chevron to use inexpensive treatment technologies (Waldron 2005). As a result, oil firms and local water districts have become increasingly economically interdependent.

This symbiotic relationship between energy and agricultural industries is nothing new in the American West. Reisner (1993) describes how large dams built in the twentieth century West were mechanisms for subsidizing irrigated agriculture – often in places ill-suited for it. For instance, much of the electricity produced via the California State Water Project’s hydroelectric dams is consumed by the project itself to pump water over mountains to farms and urban agglomerations. The relationship between the Cawelo Water District and Chevron simply signals the involvement of the fossil fuel industry – not just hydroelectricity – in providing discounted water to desert farmers.

Yet there are some salient difference. While twentieth century dams directly subsidized irrigation projects through generation revenues, oil companies supply recycled produced water in order to reduce disposal costs and transform a waste material into a source of revenue and positive public relations. Chevron (2013) touts its relationship with the Cawelo Water District to prove it is ‘being environmentally conscious’ and helping farmers. The recent establishment of a Bakersfield company promising to recycle 42 billion gallons of oil field wastewater for Kern County farmers suggests that it is also a lucrative arrangement (Hardcastle 2016).

Despite the economic benefits to Kern County farmers and oil firms, broader public awareness of wastewater recycling has provoked considerable controversy. Worries about potential environmental and human harms have been stoked by admissions that routine testing only looked for a handful of contaminants (Harkinson 2015; Heberger and Donnelly 2015). Although consumer concerns that chemicals could accumulate in produce predominate, farm workers are at greater risk via chronic low-level exposure. Indeed, many accounts describe the treated waters smelling of hydrocarbons (Xue 2016). Others worry that contaminants could accumulate in soils or infiltrate freshwater resources – including the canals where some residents fish.

The dispute over the risks of produced water in Kern County mirrors other environmental controversies, like climate change or genetically modified food: it is ‘scientized’ (see Kinchy 2012). That is, certain participants aim to reduce the conflict to a debate between differing scientific risk assessments. Political opponents, in turn, marshal their own test results and expert facts to their advantage. In response to outcry over the previous paucity of testing, the State Regional Water Quality Board assembled a Food Safety Panel and the Cawelo Water District contracted an independent toxicologist to test its irrigation waters and citrus crops, promising to monitor for 160 potential contaminants. The toxicologist reported finding little to no hydrocarbon contamination in Cawelo citrus fruit and traces of organic compounds in irrigation water below drinking water standards (Robles 2016). The Environmental Working Group disputed the findings, referring to the toxicology report as ‘poorly designed’ and ‘extremely short-term’ (Stoiber 2016, 3). Indeed, its conclusions seem suspect given that others have uncovered elevated levels of benzene, arsenic, mercury and other heavy metals in irrigation ponds (Shariqu 2013; Schlanger 2015; Srebotnjak 2016).

The State Water Resources Control Board has also attempted to transform the controversy into one of consumer risks, obscuring the possible harm to farm workers, a potentially damaging buildup of contaminants in the soil, or the economic risks of becoming dependent on an oil industry waste product. This move ‘scientizes’ the debate, narrowing it to its most easily risk assessable and quantifiable dimensions and excluding non-compatible values and rights concerns.

Despite the existence of considerable uncertainties and underlying value commitments, advocates of the arrangement nonetheless appeal to ‘science’ to bolster their position. One recent headline reads ‘science confirms oilfield produced water safe for irrigation’ (Hislop 2016), while another journalist declared that environmentalists had now ‘run smack into a wall of facts’ (Henry 2016). The State Water Resources Control Board (California Water Boards 2016), in any case, insists that ‘no studies have shown that irrigating food crops with produced water poses any threat to public health.’

Scientizing fact-rhetoric obscures what scientific studies actually do. Increased scientific scrutiny of what previously seemed like unambiguous knowledge only seems to uncover new uncertainties and complexities. For instance, greater controversy and, in turn, more scientific attention paid to the geology beneath the proposed nuclear waste disposal site at Yucca mountain only raised further questions and overturned previous assumptions (Sarewitz 2004). One expects the same for assessing the safety of produced waters. Consider how the toxicologist testing Cawelo’s irrigation waters cited federal drinking water standards to demonstrate their safety, presuming the adequacy of those standards. As Szasz (2007, 116–122) describes, United States drinking water standards suffer from three deficiencies: First, some standards are too lax; second, novel contaminants that lack a standard are not tested for; third, laboratory studies do not test long-term exposure to combinations of different chemicals. These deficiencies undermine any claim of scientific certainty regarding the risks of produced water irrigation.

In other words, much of the science necessary to evaluate the risk of recycled oil wastewater in Kern County remains ‘undone’ (see Woodhouse et al. 2002; Hess 2009). Indeed, for many oil wastewater contaminants, safe residue levels for crops are unknown, much less for combinations of hundreds of chemicals that might be interacting synergistically (Heberger and Donnelly 2015, 38). Thirty eight percent of the constituents of the produced waters used to irrigate in Kern County have not been publicly identified, due to their status as ‘trade secrets’ (Shonkoff, Stringfellow, and Domen 2016).

Even worse, some of the undone environmental science may now be undoable (see Frickel et al. 2010). Despite recent monitoring improvements, regulatory protections against environmental contamination in Kern County has been historically insufficient. Untreated wastewater was originally dumped into drainage canals and, more recently, into unlined ‘evaporation-percolation pits,’ where they drain into the ground or diffuse into the air (Miller 2011; Heberger and Donnelly 2015). As of 2015, hundreds of these pits had been operating without permits (Cart 2015). It was also revealed that a bureaucratic error allowed several oil firms to dispose of wastewater in aquifers designated for irrigation or human consumption (Baker 2015). A legacy of chronic monitoring failure makes it more difficult to discern the source of contamination in future studies and, hence, responsibility. Would the presence of hazardous environmental chemicals be due to Cawelo’s irrigation practices or simply the product of decades of ‘organized irresponsibility’ (Beck 2013; cf. Owen et al. 2013)?

Contra scientized framings, the dispute over produced water irrigation is as much a matter of values as facts. Environmental controversies are driven by differing visions of humanity’s place in the natural world (Hulme 2009) and political questions like ‘Who benefits?’ and ‘Who decides?’ (Macnaghten, Kearnes, and Wynne 2005). As Rayner and Cantor (1987) put it, the question ‘How fair is safe enough?’ motivates citizens as much as the perception of danger. Some stakeholders see current governance systems for produced water recycling in Kern County as illegitimate. A member of the local Center on Race, Poverty, & the Environment justifies their skepticism of monitoring efforts by citing the lack of state enforcement and the tendency for ‘blind faith in the industries’ (DuPont 2015). Produced waters’ automatic classification as non-hazardous waste and the fact that a mere three to four people are responsible for monitoring a wastewater stream twice as large as San Francisco’s do little to inspire public confidence (Miller 2011).

The State Water Resources Control Board has attempted to legitimize the practice of irrigating with recycled wastewater, assuring citizens that no water has ever come from hydraulic fracturing wells. Yet similar chemicals are used across oil and gas extraction methods (Stoiber 2016, 40), and there is little reason to suspect that their produced waters differ significantly. The difference is that hydraulic fracturing has lost legitimacy to much of the public through a series of controversies, becoming more ‘politically unsafe’ (see Morone and Woodhouse 1989) than other methods. The relative success of opponents of hydraulic fracking in informing the public of its potential hazards has led to older fossil fuel extraction processes appearing less risky by comparison.

The controversy over the agricultural use of produced water in Kern County is indicative of future disputes, if the practice were to spread. What lessons should one draw regarding how other regions could more intelligently and democratically face up to the uncertainties and complexities posed by the recycling of oil and gas wastewater?

The challenge of intelligently steering produced water

The case of Kern County represents an incipient instance of an emerging innovation that could present serious unintended consequences. Produced water production will increase as traditional reservoirs decline and unconventional oil and gas extraction (e.g. shale oil) rises. Much like Chevron in Kern River, extraction firms will be evermore awash in produced waters, waters expensive to treat and dispose of. Moreover, concerns about induced seismicity (Weingarten et al. 2015) and groundwater contamination may make disposal via reinjection politically infeasible, increasing the attractiveness of recycling. At least some rural western districts facing declining oil and gas production and intensifying drought and population-induced resource pressures will join Cawelo in seeing cheap, recycled wastewater as a lifeline. Indeed, a manager of a Central Valley agricultural firm insists that ‘it wouldn’t be responsible not to use [produced water]’ in a drought situation (as cited in Grossman 2016). Even absent such pressures, the historical preoccupation with taming the desert by putting land under the plow – regardless of the economic or environmental costs (see Reisner 1993) – may spur further developments.

No doubt more technological assessment of the relevant consequences posed by produced waters could help spur more responsible innovation. Yet analysts are not be capable of detecting and fairly evaluating all the risks to every social group or the environment, lacking the time, resources, and knowledge to do so. Neither would assessment resolve political contention or reliably reduce uncertainties, given the high stakes (see Sarewitz 2004). Moreover, the technology’s seeming simplicity obscures several challenging complexities: most produced water contaminants are proprietary and their interactions with agricultural chemicals and broader effects are uncertain; scaling up its use may have unpredictable effects on the subsurface hydrogeology; and it is difficult to know in advance the broader secondary and tertiary sociopolitical consequences. Hence, ensuring responsible innovation in the recycling and reuse of oil and gas wastewater requires asking, ‘How amenable are its technical, cultural, and political facets to learning and adjustment?’ How might they be reconstructed to be more so?

There are significant barriers to learning the risks of produced waters. Decades of limited regulation of oil and gas extraction have already increased environmental contamination levels, shifting the baseline of ‘normal’ with which one would assess new efforts. For instance, unlined evaporation and percolation pits were only banned in New Mexico in 2008 and remain legal in California (Heberger and Donnelly 2015). Furthermore, some experts have noted that knowledge regarding the effects of unconventional water resource use is in short supply; not only is information lacking regarding water usage and quality throughout oil and gas production processes, but the hydrogeology of deep water deposits is poorly understood (Bossert, Olson, and Bushnell 2015; Graham, Jakle, and Martin 2015). Tapping such deposits may cause the land above to subside and could impact other water sources in ways as of yet unrecognized. Indeed, it took decades for it to be realized that groundwater pumping in Albuquerque prevented New Mexico from delivering promised Rio Grande flows to Texas and even longer to end the practice (Phillips, Hall, and Black 2011). Advocates of produced water recycling will need to be incentivized or forced to proceed in ways that reflect the historical lesson of ITE scholarship: new, surprising complexities will almost invariably emerge.

Hence if decision-making regarding recycled produced water innovation is to be both responsible and intelligent, participants must recognize the need not only for building capacities of learning and adjustment but also for additional research, especially to the advantage of less-empowered groups. When advocates overestimate the certainty of current regulatory science and policy (e.g. Henry 2016; Hislop 2016) they stymie efforts toward more stringent initial testing. Rather than wait until evidence of potential harm – as has been the case with ‘fracking’ (see Ottinger 2017) – detailed studies into the environmental and social risks should precede widespread deployment. But would more assessment – like self-serving calls for more research – only make things worse (see Sarewitz 2004)? The fact that controversies remain unresolved and decision-making can become more gridlocked with additional scientific research is not necessarily a problem within the ITE framework. Because many innovations proceed without sufficient recognition of uncertainty or of disadvantaged social groups, fostering greater disagreement would help slow down the pace of innovation to a level that would enable more trial-and-error learning and encourage advocates to pare down grand proposals to more incremental changes (see Woodhouse 2016). If farm worker organizations and other groups representing less privileged stakeholders are provided access to experts who could help them dispute water districts’ and fossil fuel firms’ risk assessments, they can delay a too rapid roll out of recycled waste water innovation. Of course, critics of scientific assessment (e.g. Sarewitz 2004) are correct that debates cannot be ‘scientized’ if incremental political change is to occur; early deliberation has to forefront value differences.

If the innovation proceeds, monitoring would also need to be well funded and multi-partisan. having a mere handful of people monitor wastewater streams – as has occurred in Kern County – would be unlikely to ensure a desirable outcome simply due to a lack of person-power and insufficient diversity. Accomplishing RRI’s pillar of inclusion via multi-partisan monitoring combats the biases or conflicts of interest that exist within any one decision maker, institution, regulatory body, or firm. Oil and gas firms seek to maximize profits, and water districts are likely to be biased toward maintaining the status quo – much like bureaucracies more generally (Rayner and Cantor 1987). Such institutional motivations stymie efforts to more responsibly manage risk or seek out a more complete understanding of how failures have arisen (see Sagan 1993). Indeed, recall how many viewed the Cawelo Water District’s efforts to monitor recycled wastewater to be untrustworthy and the oil industry’s influence to be disproportionate. Without multi-partisan monitoring groups, through the increased involvement of citizen groups, university scientists, and others, recycled produced waters may also end up becoming ‘politically unsafe’ regardless of the results of official risk assessments (see Morone and Woodhouse 1989).

Furthermore, in line with technical precaution of ensuring flexibility, a moratorium might restrict recycling produced waters from beneficial use until adequate feedback is obtained from efforts already underway (e.g. Kern County). More responsible produced water innovation could be made undoable if advocates scale up or lock-in the technology before there is a chance to learn from prior experience.

A broader cultural barrier is the ‘productivist’ bias that often skews decisions over sustainable resource use; enthusiasm regarding new technologies of energy production routinely overshadows the often more economical and practical methods of energy reduction (see Zehner 2012). This same barrier may present itself in the case of water. Although there are many successful efforts toward water use reduction in the Southwest (see Fleck 2016), increasing supply of any resource tends to reduce pressures to alter unsustainable behaviors (Alcott 2005). Some of the relevant experts do recognize that recycled produced waters are not sustainable, essentially being a ‘mined’ water resource – though it is unclear if a majority does (New Mexico First 2014). Such recognition, however, may not alter the behavior of politicians, businesses, or the lay public. The reflexivity pillar of RRI, from the view of ITE, would need to be enhanced by strong structural incentives to conserve, such as more stringent water pricing combined with tax rebates for efficiency gains.

The technical features of recycling produced waters will further challenge efforts to be responsive and to proceed intelligently, namely its inflexibility (see Collingridge 1992). Indeed, the network of dams and irrigation canals on which the West relies is a large, capital-intensive, and highly inflexible sociotechnical system. In one sense, utilizing produced waters increases flexibility by diversifying water resources. However, it may further entrench unsustainable modes of desert living, as difficult choices may be delayed by additional decades. Produced water recycling itself is also highly capital-intensive. Even though some treatment technologies, such as membrane distillation, are highly modular and can be incrementally expanded, the delivery of recycled water and disposal of treatment wastes require expensive dedicated infrastructure (see Shaffer et al. 2013; New Mexico First 2014). Oil and gas producing regions are far from urban centers where recycled water could garner a high price, necessitating long and costly pipelines. Even in rural areas, where water is often priced an order of magnitude lower, transporting water still demands significant capital outlays. Produced water is delivered to Cawelo, for instance, via an eight mile pipeline. Moreover, desalination processes output highly concentrated brine, which is disposed of via expensive evaporation ponds or deep geologic injection. Given such costs, firms face a strong incentive to maximize production, treating and selling as much recycled water as they can manage to offset initial capital investments. Seeking economies of scale, however, inevitably increases the costs of unexpected errors. Moreover, firms are more likely to fight more stringent regulation if it renders infrastructural investments unusable. As such, RRI’s pillar of responsiveness would be advanced via greater opportunity for learning and adjustment, namely by restricting recycled produced waters to local uses and less capital-intensive forms of delivery (i.e. tanker trucks).

Most concerning is the possibility of regions becoming dependent on recycled produced water, especially as flows in major western rivers decline. Unless it proves possible to take greater advantage of more sustainable power sources, the energy intensity of treatment alongside rising fossil fuel prices will make recycled water exceedingly costly economically and environmentally; reliance on energy-intensive water resources would further lock-in fossil fuels and nuclear energy. It is unclear what, if any, precautions could prevent inflexibility via resource-dependence. In that regard, such endeavors resemble the light-water nuclear reactor, whose safety systems could only delay or mitigate disaster – not prevent it (Morone and Woodhouse 1989). What technical arrangement for extracting and treating brackish water would not exacerbate the underlying causes of the Southwest’s growing water pressures, namely climate change and burgeoning populations?

Prudent political precautions toward lessening the barriers to intelligently deploying recycled wastewater technology would include ensuring the maximum feasible diversification of decision making – in line with RRI’s call for inclusiveness. Unfortunately discussion of the risks of produced waters are happening considerably late and behind closed doors. Recall that supervision of Kern County’s produced water falls mainly to regional water authorities, while other states – such as New Mexico – leave the regulation of such waters to oil and gas bureaucracies. Moreover, it is a reasonable worry that greater dependence on produced water may only amplify the privileged political position of fossil fuel firms. Such arrangements can breed distrust and political polarization, strongly contrasting with the decentralized networks of diverse partisans that have achieved many of the recent successes in Southwest water management (Fleck 2016), what Ostrom (2010) terms ‘polycentric governance’ (see Tembly et al. 2017). Deliberative networks, including managers, scientists, environmentalists, and politicians, have managed to accomplish what previously seemed politically impossible: for instance, the renegotiation of the distribution of Colorado River flows during droughts and allowances for ‘wasted water’ to support ecosystem maintenance (Fleck 2016). Moving past a scientized and polarized stalemate regarding produced waters seems more probable with the support of similar stakeholder networks.

Bringing produced waters under the partial jurisdiction of other institutions (e.g. offices of the state water engineer, worker and environmental protection, etc.) and inviting the participation of environmentalists, farm worker activists, and others would also be necessary in order to avoid the further scientization of the debate. Indeed, some experts are already conceptualizing public outreach regarding produced water in terms of ‘allaying fear’ and decreasing the ‘yuck factor’ (Hagström et al. 2016). This framing presumes that public opposition is rooted in an irrational ignorance of the purported facts – disposing the technology’s advocates to overestimate their own understanding of the hazards. Hence, countering such frames would be necessary to advance the RRI pillar of anticipation.

Making contamination costly for oil and gas firms, as well as for water districts, would lessen the cognitive pull of financial conflicts of interest. Given that some of the effects may not be detected for years, if not decades, and responsible parties typically fight financial penalties in court even longer, adequate compensation to potential victims could be assured via environmental bonds (Kim 2008). Groups that stand to benefit from recycled produced waters would be required to pay into human and environmental victims’ funds, setting aside money to rectify errors as they arise, only being returned if resulting consequences are less significant than anticipated. This would, in addition to advancing RRI’s pillar of accountability (see Foley, Bernstein, and Wiek 2016; Hartley et al. 2016) and incentivizing anticipation, enforce a higher degree of cooperation among those producing water and environmentalists, by realigning their incentives in a similar direction (see Zehner 2012, 180).

Conclusion

This paper has attempted a prospective strategic analysis of the recycling of oil and gas wastewaters for beneficial use, imagining how produced water innovation could proceed more responsibly. The sociotechnical landscape has been analyzed and found wanting: Significant cultural, technical, and political barriers stand in the way of RRI in the case of produced waters. Anticipation, reflexivity, and inclusiveness are limited without multi-partisan assessment and monitoring to foster disagreement and increase the number of considerations; governance cannot be adaptive or responsive absent measures to slow deployment or limit it to more flexible technologies (i.e. trucks rather than pipelines); and accountability and anticipation are weakened without victims’ funds or environmental bonds. Lacking a concerted effort toward a more learning-oriented process – one countering the various structural barriers – is it reasonable to expect that such innovation will be responsible?

Further research could uncover other ITE strategies to make the sociotechnical landscape of produced waters more amenable to RRI. For instance, how to reduce productivist biases (see Zehner 2012)? Could lessons be drawn from psychological research on combating climate denial and other system justifying ideologies (see Dotson 2015)? The psychocultural barriers to more responsible innovation might loom as large as the material and the political ones. Moreover, how might successes of diverse governance networks in other resource areas be best replicated? How to recreate the incremental successes of Colorado River management (Fleck 2016) or the collective stewarding of irrigation systems and forests described by Ostrom (2010)? Examining how to reduce polarization among otherwise fanatically divided opponents may be an essential research question for achieving RRI in the twenty-first century.

Moreover, from an ITE perspective, not all innovations are equally compatible with RRI, varying in terms of how flexibly they can be designed. The technical barriers to RRI may simply be too large for some technologies, which cannot be built except at large scales and with significantly delayed feedback or that have potentially catastrophic unintended consequences that cannot be adequately anticipated. For other innovations, significant power asymmetries may stymie efforts to develop a more inclusive, plural political process. Technologies that provide similar benefits as recycled produced waters: desalinization, the exploitation of non-fossil-fuel-related brackish reservoirs, and other forms of water reclamation, might compare more favorably in terms of the feasibility of cultural, technical, and political precautions. Might efforts to spur responsible innovation be more effective by supporting these alternatives?

My broader aim has been to demonstrate how the ITE framework can provide a proactive – rather than retrospective – strategic analysis of emerging technologies. Researchers need not wait to ask critical questions until after long-lasting harms become obvious and alternative pathways mostly forgone (see Woodhouse 2005). No doubt early inquiry entails its own risks: The researcher must reason hypothetically in the face of considerable uncertainty. Yet the professional costs of one’s conclusions being found later to have been incomplete – if not incorrect – seems minor next to the harms potentially resulting from decision makers proceeding without critical social scientific guidance. Indeed, more intelligently meeting any of the major challenges facing twenty-first century technological civilization seems improbable without more social scientists pursuing such risky scholarship.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Taylor C. Dotson is Associate Professor of Social Science at New Mexico Tech. His research focuses on the possibilities for more intelligently steering – and potentially even dismantling – technoscience in order to reconstruct more democratic, communitarian and sustainable technological societies. He is the author of Technically Together: Reconstructing Community in a Networked World and is currently writing a book about facts, fanaticism, and the future of democracy.

ORCID

Taylor C. Dotson http://orcid.org/0000-0002-1777-3769

Notes

1 ‘Risk’ is sometimes seen as privileging expert views. ITE uses it in a non-technocratic way. Hence, I have retained the word despite its baggage.

2 ITE’s skepticism of analytical forecasting is rooted in a public administration literature that predates other critiques of prediction, including within RRI, by several decades.

3 In some accounts, RRI’s pillar of ‘responsiveness’ resembles incrementalism in certain respects.