The Role of HCI in Cross-Sector Research on Grand Challenges

Grand challenges describe problems with socio-political as well as technical complexities. As we approach 2050 and a future with 9.5 billion human inhabitants providing energy, food, water, health-care, and other necessities in a sustainable manner will be grand challenges. The Global Footprint Network [1] estimates that our current demand for renewable ecological resources and services is more than what could be sustainable if provided by 1.5 Earths. An additional concern is that critical infrastructure and ecological services are becoming increasingly vulnerable to both anthropocentric and natural threats. Simultaneously, technological development is occurring at an accelerating pace [2]. One has only to look at the progression of computational power, internet path, and tool use, among other trends for evidence of this acceleration. Grand challenge innovations are often looking for technologically innovative “silver bullets” or what Silicon Valley would term “disruptions.”

Disruptions describe technological innovations capable of providing industry level paradigm shifts [3]. For example, consider the shift from private vehicle ownership to autonomous vehicles summoned on-demand with ridesharing (Google Car + Uber + car2go). Personal vehicles spend on average 95 % of their time parked. Such a paradigm shift could reduce the number of vehicles by a factor of ten, reclaim much of the one-third of city land dedicated to vehicle storage, while reducing per capita transportation expenses and increasing transportation safety [4]. However, solving grand challenges requires more than just technological innovation.

Problems are often multi-faceted with regulatory, cultural, and economic considerations in addition to the ecological and utilitarian ones. For example a shared vehicle model could have unintended negative consequences: economic efficiency gains from autonomous vehicles would likely result in increased consumption and/or a willingness to commute farther distances due to the ability to do productive work while commuting. Cost efficiencies would be nullified if automotive travel per capita saw a net increase. We also observe economic and social disruption for the displaced taxi or limousine drivers that may elevate the economic inequality without effective workforce development and welfare support. The complexities surrounding grand challenges translate to complex solution spaces. The problems cannot be solved by finding a way to optimize a single variable or even a set of variables. Many of the 21st century problems we are facing require understanding that systems are adaptive and exhibit high order interactions with one another and that must be taken into consideration in order to manage tradeoffs. Finding innovative solutions to tradeoffs requires interdisciplinary as well as cross-sector research teams from stakeholders, private sector, government, and academia. Building a shared understanding of the problem from technical, social, economic, and ecological perspectives is an essential first step to working out solutions that stand a chance in the real-world.

Here we begin by examining some of the technological visions of the future that provide the framework for building innovative solutions.

1.1 Cyberinfrastructure Vision

Cross-sector collaboration has many challenges. Here, we begin by examining the technical. Increasing the interoperability of cyberinfrastructure (CI) is a key component to conquering grand challenges. Briefly, cyber infrastructure is the hardware and software that enables the storage and retrieval of data. Current CI is often segmented and specialized to specific domains [5]. For example, a hydrologist might have a set of databases that are compatible with their tools, but those tools and databases may not be compatible with those used by a waste water engineer. In contrast, future CI is “heralded as a transformative force, enabling new forms of investigation and cross-disciplinary collaboration” [6]. The vision would be enabled by CI that is end-to-end from the collection of data to the analysis, storage and dissemination [5, 6]. End-to-end CI would provide guided or automated acquisition and processing of data, run model simulations/forecasts, analyze the results, provide statistical reports, and visualize the results. For this to occur cross-sector/interdisciplinary collaboration is necessary. Academics may be well suited to resolve unknown relationships between variables but are less inclined to know how to deploy those results or models inside of a resilient CI framework and need the assistance of computer information specialists. Industry and public stakeholders need decision support tools to form decisions based on the best available data and models. Ideally, decision support tools can encapsulate expert knowledge for use by industry and public stakeholders.

Future CI will not only streamline human computer interaction (HCI) between domain experts and their tools but increase the accessibility of tools to non-domain experts. In the National Science Foundation [7] CI vision stipulates that open access built upon an open technological framework is essential. This will increase the accessibility of scientific information to the public and enable stakeholders and government agencies to participate in a virtual community alongside researchers. CI needs to be built to simultaneously serve both basic and applied research in academic, governmental, and industrial sectors. Researchers benefit by gaining exposure to real-world needs which can serve as a catalyst to theoretical research. An open-framework also provides educational opportunities for K-12 as well as workforce development for the next generation of technical professionals. Education can be difficult to be privatize for the entire public, but an educated workforce can improve every sector. Additional benefits of the openly available resources are increases in the capacity of the academic and industrial sectors; reducing unnecessary redundancy leads to more sustainable research and stakeholders better served by industry. The community as a whole benefits through better access to data, tools, and computational resources. This is a paradigm shift from the translational research mode, by which basic research may eventually find its way to industry through a unidirectional process. In contrast, the cross-sector/interdisciplinary model features problems posed by government agencies or stakeholders; the community then engages in problem-solving through developing shared understanding of the problem.

Achieving this vision will require robust, complex, multi-layered technologies for monitoring environments in real-time, assessing their state, providing decision support to human supervisors, and implementing control actions. A heterogeneous group of users will need to interact with CI at every layer for a variety of purposes. For this reason human computer interaction becomes a critical consideration in the design and implementation of cyberinfrastructure. The end goal is to produce technology that results in better integration of policy, strategic (management) and operations decision making. However, HCI applies not only to the interfaces used by policy makers, managers, and stakeholders. HCI is important at every layer of the CI stack. CI when combined with highly distributed sensors and data streams could form large scale cyber physical systems (to be discussed). Systems will expose web APIs for other services (machines) to interact with them. However, they will also expose user interfaces for human computer interaction. Some web-services might allow sensor readings to be placed into a database, others provide a means of retrieving data, running a model simulation, or visualizing the model forecasts or predictions.

Trends suggest that implementing solutions to grand challenges will require distributed cyber infrastructure built on the defacto standard TCP/IP protocol. Care needs to be taken in the design of the web APIs as well as the user interfaces for each of these web-services. Web-services will provide the coupling between sectors. In some scenarios it will be technologically feasible for systems to communicate independent of human users. In other scenarios human users will be needed to monitor, operate, maintain, and administer on-line systems. Longer time scale problems with interacting factors might be suited for decision support tools capitalizing on the analytics provided by new cyber infrastructure. Analytics will be able to aggregate the vast amounts of sensor and simulation data as well as social trends from online social networks. While the information processing and storage capabilities are likely to increase exponentially, the cognitive capacities of human users are unlikely to keep pace. HCI will become essential to verification and validation of interfaces to ensure that they meet task requirements. The proliferation of data and processing power provides opportunities and arguably necessity for advanced visualization of data and virtualization of physical systems.

Coupled Natural and Human Systems Research.

Coupled natural and human systems consider how both biophysical and human factors interact with one another. Ecological systems provide valuable ecological services to communities that must be managed to maintain the health of ecosystems and the sustainability of ecological services. Ecological services are becoming increasingly strained by human population growth and climate change. Considering human influences is essential to understand the coupled natural and human systems as a whole. In the United States, federal, academic, and industry, and NGO are actively collaborating to develop ecosystem service concepts to further national environmental and economic objectives [8].

Properly implemented cyberinfrastructure could greatly enhance cross-sector research. Currently academic researchers might use and develop highly sophisticated biophysical models for basic research that go unused or under-utilized by governmental regulatory agencies and stakeholders. These models may still require manually acquiring and processing input datasets despite the enormous growth in data and increased availability of data via web-services. Instead, it remains common place for domain experts to spend their careers learning how to use highly specialized models. The codebases of these models are often the result of decades of research and several “person decades” worth of effort. The problem is that for many of these models, collecting/acquiring and processing the input datasets is still a manual process. The inputs as well as the outputs are in non-standard formats. Domain experts spend a large amount of their time learning technical details of file formats and software systems that are irrelevant to the science they are actually interested in.

Embracing modular, test-driven, scalable, open cyber-infrastructure frameworks could offer productivity gains, leverage citizen science, provide tools and analytics for regulatory agencies, and provide decision support for stakeholders. Web services should be developed that allow for data intensive analysis of geographic regions of interest without needing to know specifics about where to acquire the data or how the data is formatted. Future technology could provide:

• Ability to assess physical systems at multiple time resolutions and temporal scales

• Ability to understand the complex adaptive interactions of human, technological, and ecological systems

• Mitigation of climate change impacts, increased sustainability, managed ecological services

• Regulations that allow trade of “nature as capital” to pursue environmental goals [8].

1.2 Cyber-Physical Systems (CPS) Vision

The proliferation of cyberinfrastructure combined with increased sensor availability and lower costs will enable the monitoring and control of physical systems. Rajkumar et al. [9] describe CPS as “physical and engineered systems whose operations are monitored, coordinated, controlled and integrated by a computing and communication core.” CPS will improve the efficiency of existing systems and provide new capabilities. CPS is in many ways aligned with the cyberinfrastructure vision discussed earlier although it is not synonymous. Cyberinfrastructure is built to address information management needs. CPS arise from engineering domains and operate in real-time employing both sensors monitoring and actuators affecting the physical world. CPS exist at a variety of scales and many technologies can be classified as falling under its umbrella: personal medical devices such as prostheses, adaptive automation in vehicles, drones, smart-buildings, smart-cities, power distribution networks, and even planetary scale risk monitoring of critical infrastructure from empirical data and models of natural resource use and climate change [10]. Traditional embedded systems focus more on the cyber components, whereas CPS takes a holistic approach integrating the cyber and physical components [11].

Cyber physical systems can replace analog and mechanical control systems while offering increased reliability and efficiency. Advanced adaptive control systems are being developed that can change strategy based on the stability of the system. When the system is more stable the controller can optimize efficiency, whereas when the system is close to the operating envelope the controls can act more aggressively to stabilize the system [12].

Strides are being made at the small scale, but at the large scale challenges remain. According to Sztipanovits et al. [13] the largest obstacle to CPS is system integration. Systems can be designed and modelled digitally. The components can be specified and manufactured, but building the physical system is still less than a science. A second challenge to large scale CPS is economic deployment of sensors and actuators at a planetary scale to measure and forecast critical indicators [9].

Critical Infrastructure Research.

Critical infrastructure represents another grand challenge that commonly involves government, industry and academia. For example, phaser measurement units (PMU) enabling smart grids involves substantial collaborative research. However, HCI research targets effective HCI and human factors engineering except for heavily regulated sectors such as nuclear power and commercial aviation. In heavily regulated and safety critical sectors, collaborative HCI research commonly focuses on improving user interfaces and measuring human performance, a conventional application of the discipline for critical infrastructures. However, critical infrastructure can also engage in recent HCI research approaches such as social informatics more common in the private sectors. For instance, UBER is increasing the availability and efficiency of the transportation infrastructure (i.e., roads and cars). HCI research can potentially introduce novel concept of operations for the government to maximize availability, security and efficiency of critical infrastructure but such ‘revolutionary’ changes require additional research opportunities.

Smart-grid communication networks are typically described as connecting power generation plants, power distribution networks, and consumers. However, necessity exists for cross-sector communication. Academics might need access to real-time and historic data for research purposes. National laboratories have high performance computing resources that can aid operations and research. Renewable energy generation and energy consumption is highly dependent on meteorological conditions. Smart-grids will need high-resolution, localized forecasts to schedule power ramp-ups. Regulatory and standards organizations will also need real-time communication to ensure compliance as well as develop and manage existing standards [14]. Lastly, consumers might wish to modify consumption based on dynamic pricing per market demands.

1.3 Participatory Culture Vision

To the dismay of skeptics, the small contributions of millions of users have accumulated in high value resources as well as advancements in science and technology. Wikipedia has over 35 million articles in 288 languages. Foldit is an online protein folding game with a community of 240,000 players. Among several notable contributions to science, these players improved the activity of an enzyme by 18 fold in a matter of weeks [15]. These accomplishments were possible despite the fact that the vast majority of internet users have little or no technological or programming prowess. Good interfaces make it possible for smart individuals with limited technical skills to contribute to crowdsourcing efforts like Wikipedia and Foldit [16]. Crowdsourcing local expertise and cognitive processing power is perhaps an underutilized resource for managing 21^st grand challenges.

Citizen science refers to scientific research conducted by nonprofessional scientists. In the modern context it usually refers to citizens collaborating with professional scientists in the collection and processing of data [17]. Ecological data collection is expensive and it is unlikely that remote sensing will be able to fully capture measurements of biodiversity with the granularity needed for decision making and regulation anytime in the near future [18]. Citizen science can be enhanced through HCI research that can reduce the variability of novices engaging in scientific research tasks. HCI is also needed in validating data collected by citizens.

Providing public stakeholders with tools that can convey scientific knowledge in a manner that is more accessible to non-domain expertise and allowing them to examine how particular actions will affect the future state of their environments and communities could provide positive democratizing effects on public decision making. Without such tools public stakeholders must rely on the authority to gain insight into the inter-workings of their environments. By being able to interactively manipulate policies, environmental factors, or decisions under a variety of scenarios ordinary citizens can gain understanding into the non-linearly dynamics of their system and potential risks that they might want to avoid.

Having already discussed technical challenges of cross-sector/interdisciplinary research we turn our attention to non-technical challenges facing collaborators. The merits and incentives for interdisciplinary research are evident but cross-sector and interdisciplinary research can be challenging. Bryson et al. [19] remark that organizations fail into collaborations. Collaborations take place when organizations cannot get what they want without collaborating. Here we examine the perspectives and goals of government, academic, industry, and stakeholder sectors. Together these sectors comprise the social infrastructure that maintains the availability of critical infrastructure and ecological services through a shared power structure.

2.1 Government (Funding Agencies, National Labs, Regulatory Agencies)

Government is a multidimensional construct, ranging from legislative bodies, to funding agencies, to defense agencies, to research laboratories, and to regulatory agencies. Each one views the role of research and development differently. In the United States, for example, the federal legislature must strike a balance between reducing public spending and ensuring the economic competitiveness and national security through public expenditures. Legislative bodies must therefore navigate short-term electability and long-term vision for government.

A consistent push for national security has ensured some stability of defense research and development, including more recently the advent of cybersecurity. Technology development has helped provide competitive advantage to U.S. defense tools, and HCI has been deployed as a tool to help improve performance of technologies deployed in defense contexts, from better avionics for pilots, to augmented reality headsets for ground soldiers, to better intruder detection systems for cyber applications. Defense agencies like the Defense Advanced Research Projects Agency (DARPA) ensure that emerging HCI technologies are realized and implemented. DARPA regularly hosts grand challenges that feature an HCI component.

Beyond defense applications, there remain many government agencies that champion basic research and development. The National Science Foundation (NSF) and National Institutes of Health (NIH) underwrite foundational research, primarily in academia, to further scientific goals. These goals can include HCI opportunities, such as improved medical device interfaces.

Federal research laboratories, including those operated on behalf of NASA or the U.S. Department of Energy (DOE), serve as a resource to explore research that the private sector cannot or should not perform. As industries become viable, such as commercial space travel, federal funding subsides in these research facilities. Yet, they remain an important capability to meet government research needs and develop technologies that may provide advantage to U.S. interests. Additionally, during periods of economic downturn, private sector funding in research and development may decline; federal research facilities are a key component to maintaining a vibrant national research culture that spans economic ebbs and flows. An important part of the national laboratory framework entails partnering with academia and the private sector. Increasingly, Small Business Innovation Research (SBIR) has become a vessel toward commercialization of technologies developed in federal research facilities. In terms of HCI, these laboratories often house human factors departments and work to apply human factors including HCI to technologies that are under development.

Finally, there are regulatory bodies - agencies serving to protect and promote public safety. Historically, these agencies are at the forefront of HCI because they regulate the safety of technologies used by humans. From consumer safety, to automobile safety, to highway safety, to aviation safety, to nuclear safety, these agencies are large consumers of research to determine safety solutions. While some research is done in-house certain agencies may contract with outside sources to conduct safety research. A grand challenge arises in the face of technological disaster, such as the Deepwater Horizon Incident. The regulatory agencies must quickly mobilize research to address high profile public safety concerns, and HCI is one technology that ensures the operational safety of technologies under scrutiny.

2.2 Academia

Traditional academic positions, namely tenure-track professors, are driven to produce knowledge for publications and train students for their graduate degrees. Though compatible with goals established in government and industry, these two key drivers lead to some characteristics that are shared with other sectors but others that are unique to academia. First, academic research tends to be theory-driven (loosely defined) while laboratory research is tightly-controlled such that findings are both generalizable and repeatable. For repeatability, the rigor on scientific method is also a major focus. The research products are meant to be publishable or public, rather than strictly commercial interests. This contrasts to the applied or problem driven research most typical in the focus of the government and industry, which emphasizes on what works in a practical setting (rather than what is optimal in an ideal setting). For some parts of the government and commonly for industry, the work is often kept proprietary rather than public. On occasions, academic researchers do have “field work” opportunities that facilitate testing of theory-driven findings and formulation of new theories.

Second, academic research tends to produce work in accordance with academic cycles or “academic time scale” as much of the work is carried out by graduate students. The academic time scale means that (i) students can only be recruited at specific times of the year for at least two years, and (ii) some training time is required to prepare the students to carry out the work. The academic time scale is thus different from both government and industry operating per fiscal year cycles and expecting immediate progress on projects. On occasion, academic researchers are able to align their schedule to work with government and industry, typically involving some lead time for synchronizing time cycles across the parties.

Within academic institutions incentives may not exist for cross-sector collaboration. Work that does not result in publications demotivate academics if other incentives are not in place. Interdisciplinary work has more unknowns, more collaborators and consequently more channels of communication. Those factors make collaborative work less productive. If the only metric considered is time per publication it becomes difficult for researchers to justify time spent on collaborative efforts. Collaborative research becomes effective when the key drivers, such as student graduation and time constraints, are aligned across the three sectors. Thus, fostering research across sectors to make progress on grand challenges depends partly on our efforts in creating the setting conducive to the participation. We believe there are three feasible ways to significantly improve the setting for cross-sector collaboration, especially with respect to working with academia.

First, synchronization between “academic” and “fiscal” time is essential for collaborative engagement. One potential method is that the government should strive to provide constant access as a “field” for academic research. In other words, the government may treat itself as a “field” where graduate students can go observe and collect data, regardless of direct project sponsorships. This can lead to formulation basic and applied HCI research to pursue either across or within sectors. Further, graduate students are constantly being trained in the field, and thus can be productive at the beginning of any sponsored project that often has a much tighter fiscal deadline. The government needs to prescribe controls on accessibility and support as appropriate. This method of engaging academics for supporting the government is feasible because academic institutions have some flexibility in conducting unfunded research activities that private sectors generally cannot provide.

Second, long-term or predictable funding could also enhance collaborative research. However, long term planning is often challenging. One potential method is extending the collaboration nature of existing internship opportunities in the government or potentially industry. Internships can be dependent on research participation for the government, improving the expertise and integration of graduate students in the government settings. This method coincides with the proposal to increase government access to academic research.

Third, academic institutions need to develop students not only with competencies within multiple disciplines given the complex nature of grand challenges, but also with the skill of switching between the operating modes of research and practice/industry work. One potential solution is that academic institutions need to provide formal incentives for training students and researchers in translating research into application/technology. Otherwise, collaboration between government and industry with academia can remain “rocky” or only effective for specific research groups.

2.3 Industry

Industry is driven to fill market demands and is generally more application-centric and productivity driven. Industry provides services, engineering and design, construction, equipment to clients in both private and public sectors. Success requires finding economically viable means for delivering goods and services. Technological industry giants such as Google and Apple have sufficient capital to make significant investments in research development and have made significant contributions in advancing and deploying cyber infrastructure [17], although the trend does not hold across industry. Corporations and utilities can be successful without having to develop innovative intellectual property by licensing technology and focusing on business fundamentals.

Open architecture and data access will reduce barriers of entry for startups developing applications that have the ability to connect users to information for both entertainment and decision making. The small scale of startups can actually be advantageous in new markets as it can make them more adaptable and nimble compared to larger counterparts. Funding startups can be problematic and open access to data resources can ease those requirements by not only eliminating direct costs of data but indirect costs for tools to access data.

2.4 Public Stakeholders

Here we refer to public stakeholders as citizens and communities/organizations organized and united around a common cause. They share a concern or an interest in a particular issue or set of issues. Communities can be geographically segregated or segregated in cyberspace. Communities can vary in their level of political organization, resources, authority. Public stakeholders are beholden to only themselves and in this context are problem-centric. They know what they value (e.g. clean water, cheap gas, etc.) and take actions to maximize those values. Freeman [20] describes stakeholders as “any group or individual who is affected by or can affect the achievement of an organization’s objectives.” Their power is derived from influence others in the public sphere to affect the market and politics.

The most challenging aspects of cross-sector collaborations are likely to be non-technical with communication at the forefront. Individuals spend significant portions of their lives in the same discipline and become “united by customs, tradition, and adherence to a largely common worldview(s)” [21]. Cross-sector and interdisciplinary research can become burdened because participants view the world in fundamentally different ways [22]. Collaborators do not need to develop a common philosophical basis for how they view the world, but it can be beneficial for collaborators to gain insight into mindsets of collaborators because they influence how they see the value of their effort and their collaborations [23]. Workforce development of individuals who can span disciplines and sectors are likely going to be highly valuable for cross-sector research endeavors. In particular, there is a need for individuals who are trained to “understand and address the human factors dimensions of working across disciplines, cultures, and institutions using technology-mediated collaborative tools” [5].

Despite the harsh reality of climate change and dwindling natural resources, technology provides reasons to be optimistic about the future. Human and natural systems are becoming increasingly interdependent as new technologies pervade. Technology offers a double-edge sword of enabling us to manage our environments through intervention while producing systems that are technologically dependent. Care must be taken to avoid unintended consequences and ensure future generations have access to currently available critical infrastructure and ecological services. Additionally, care requires both qualitative and quantitative analysis of the systems as well as blunt discussion regarding values, concerns, and tradeoffs when it comes to forming decisions. HCI forms the bridge from our holistic understanding of the world – the nexus of information, models, and analysis represented across the web of cyberinfrastructure – and our collective mindfulness.

This work of authorship was prepared as an account of work sponsored by an agency of the United States Government. Neither the United States Government, nor any agency thereof, nor any of their employees makes any warranty, express or implied, or assumes any legal liability or responsibility for the accuracy, completeness, or usefulness of any information, apparatus, product, or process disclosed, or represents that its use would not infringe privately-owned rights. Idaho National Laboratory is a multi-program laboratory operated by Battelle Energy Alliance LLC, for the United States Department of Energy.