As governments around the world digitise their services, citizens are being subjected to conflicting forces of identity consolidation and fragmentation. In the name of ‘customer focus’, governments and their agencies are constructing systems that aggregate personal data from multiple sources. The ostensible reason is to provide convenient, personalised and accessible service on par with that of Facebook or Amazon. Through using corporations as models, governments are developing systems of surveillance and control based on commercial marketing technologies. As a result, citizens are left to reconcile the fragmented personas constructed from data created for incommensurate program goals.
While the recent Australian government robo-debt offensive provides a few insights into this new online retail mode of governance, it is just a sneak preview of the directions in which digital government is heading. This episode has taught us more broadly about the perils of data sharing, in particular the careless use of incommensurate data. It has also provided insights into the willingness of bureaucrats to prioritise compliance over ‘customer service’ and, despite internal and external warnings, to (illegally) force citizens to explain discrepancies in the data.
As demonstrated by James Scott, when a government department seeks to administer an aspect of the natural or social environment, it works from a stripped-down map of the entities involved. The field of attention is abstracted and simplified within the goals of a program. The real world, whether a forest or a city, must be reduced and ordered to a point that makes it manageable from above. Moreover, any given community is subject to disparate programs that have been developed and legislated in a piecemeal way, each addressing the perceived circumstances in the time and place of their conception. Traditionally, program data has been service-focused and transactional. As a result, government information about a single person has been scattered between programs and the individual has been represented in the context of the program, for instance as a student, a passenger, a patient or a taxpayer. This means that each of us has to deal with government such that both parties assume a separate identity defined by each program. This sometimes leads to frustration in dealing with ‘the government’ when people assume it to be a single entity, but it has the advantage of limiting the ability of a government to have a singular view of a citizen.
Digital-government advocates propose to overcome these problems by delivering ‘seamless’ or ‘joined-up’ government, in which a government acts as a single entity in a relationship with a joined-up ‘customer’. Digital government combines web and app-based interfaces with large-scale back-room social and technical infrastructures. Digitising and automating large programs such as those operated by Services Australia can involve a comprehensive organisational ‘transformation’. The online retail mode of governance that pulls this picture together is based on computing infrastructure developed for corporate marketing systems that is designed to address analogous bureaucratic trends in the private sector. The origin of these customer-relationship-management (CRM) systems was a general move by many corporations from transactional or product-based marketing to relationship marketing. This connects with the widespread availability of modern computing and communications technologies that can collect, exchange and analyse large quantities of information at a micro level. CRM systems consolidate all information about a corporation’s customer across all its products and services, enabling precise tracking, prediction and influencing of customer intentions, preferences and behaviour over time. Analysing data over time opens up the possibility of predicting customer behaviours and identifying the opportunities and risks they present to the corporation. Consultants, systems integrators and software vendors have promoted this capability to government bureaucrats, such as those in Services Australia, who are bent on removing the uncertainties created by their old transactional systems and the unreliable behaviour of their ‘customers’. With such weapons in its arsenal, Services Australia has foreshadowed its ability to profile citizens who present a risk, thus enabling it to take pre-emptive action to prevent fraud and error and tighten compliance.
Services Australia delivers programs on behalf of thirty-four federal agencies. It is currently undertaking a $1.5-billion, seven-year program to fundamentally transform the delivery of social-services payments and services. It aims to use CRM and other large modular corporate systems to consolidate the data it has on each citizen across those programs. In doing so it wants to create ‘a single view of the Customer’, built up from thirty years of existing data and drawn from real-time links to other government and non-government sources. Services Australia wants to minimise the amount of information provided by customers in favour of data obtained through linkages to the systems of external organisations. It also wants to become a (digital) platform for the services of other federal, state and local governments.
However, this process of consolidation collides with the service-specific renditions of customers and their circumstances. This was amply illustrated in the case of robo-debt, which foundered in part on the differing definitions of income between Services Australia and the Australian Taxation Office. It was up to customers to reconcile the differences. This is just one simple example of a problem identified ten years ago by Paul Henman, who pointed out that it is customers who must navigate the incommensurate requirements of multiple policy areas when they encounter joined-up interactions with government. The continuation of this trend clearly highlights the entrenched power imbalance between the public and private system designers and a disparate and individualised citizenry.
There is another dimension to this accumulation of customer-centric data. Michel Foucault described a ‘disciplinary government of each and all’, where, on the one hand, individuals are rendered visible to the state in fine detail and, on the other hand, populations can be sliced and diced according to life circumstances or current and future risk to the state. In the documentation for its transformation program, Services Australia envisages ‘a circumstance based approach’ to managing customers. It explains that:
Customer Circumstance data [is] the data that relates to the events that have or will occur in a Customer’s life, as disclosed to the department by the Customer or authorised third party… As a Customer moves through their life, their Circumstances (such as marriage and other relationships, residence, employment, birth, death and disability status) change and in turn these changes affect the Customer’s Eligibility and Entitlement.
Cutting across this consolidated view, according to Services Australia, customers can also be:
identified as part of a micro-segment according to their level of complexity, access needs and preferences, and level of risk. This is measured through indicators and Circumstances (both reactive and proactive) that can be observed by the department. [This] will enable the department to target service delivery based on an assessment of the individual risk, access needs and complexity of Customers, tailoring service offers to match customer circumstances.
This may be interpreted to have a benign meaning: that Services Australia could be more inclined to ensure that every eligible person receives their due benefit. But the robo-debt experience demonstrated that it might be more prudent to interpret ‘target’ in a negative sense: to mean that only eligible people receive their benefits. Such an interpretation is supported by the predictive-risk frame engaged by the department. Robert Castel observed that such preventive techniques ‘promote a new mode of surveillance: that of systematic predetection’, which can ‘dissolve the notion of a subject or a concrete individual, and put in its place a combinatory of factors, the factors of risk’. In either interpretation, predictive risk assessment is a licence for an entity to surveil all its clientele all the time in an effort to assert control.
What does a citizen look like in the eyes of the state when she is constructed from multiple databases and how does she respond to the resultant kaleidoscopic rendition of her? Let us suppose that she has had multiple contacts with government agencies and programs over time. She could ask to see all the information that a government has on her. (Of course, despite current freedom-of-information laws, it would be almost impossible to get all her data. The citizen-customer would be obliged to seek her data separately from each program and agency, not always with success.) She would get something like a bunch of spreadsheets with data points (and, hopefully, field names) that have been stripped of the context of the agency applications that extracted the data from and about her. Unlike an old-fashioned paper dossier, with structured documents in chronological order, she would get the raw data but not the ‘business rules’ that give it meaning and that link the data elements together into a narrative. In a scenario of networked databases, her identities are constructed on the fly within the policy logic of each circumstance and the generic global logic of customer management embedded in the CRM system.
Like other major centres of surveillance in the Australian government, such as the Department of Home Affairs, Services Australia seems to assume that multiple incommensurate views of the subject can be reconciled without difficulty. But this approach assumes that the subject could be rendered in a coherent way that would serve all the programs drawing on that single rendition—a variant of the ‘view from nowhere’. This demands that all participating programs agree on the same definition of shared entities and their characteristics. Even if such a cultural mind-melding were possible, it would bring about a power struggle to agree on a dominant world view and the meanings of its language. What tends to happen in this situation is that interested parties, especially in bureaucracies, will only push so far, leaving inconsistencies to be resolved through informal arrangements or—such as occurred with robo-debt—other less powerful parties. Thus, inconsistencies must necessarily persist and the task of resolving them will generally fall to the isolated neoliberal subject; it will be their task to prove that their lived reality is more complex than the bureaucrat’s model.
Technocratic Urban Governance and the Need to Localise Computing Infrastructure
Andrés R. Yaksic, 21.7.2020