The push to the cloud
by Sanne Stevens
While everyone is Zooming like crazy and doom-scrollingly looking forward to an end to this forced on-screen life, tech companies are unabashedly enthusiastic. Their already impressive empire grew explosively during the pandemic. We do not yet know how permanent online working and education will be – but we do know that the mass migration to the cloud is unstoppable. At TU Delft, researcher Seda Gürses dissects the far-reaching consequences and risks, the indifferent attitude of public institutions, the urgency of a public strategy and radical alternatives.
Not any vision
For years, Seda Gürses has been a critical, active voice in the field of digital technology (more about her at the end of this article). In her view, the public institutions have been taken far too much of a wait-and-see attitude in the face of the changes, even though the profit model of the cloud services has far-reaching consequences and the urgency of a strategy in the public interest is greater than ever.
'I don't really see a vision anywhere where people are really considering the full extent of these developments. The way institutions will change, the long-term economic consequences ... I don't see it. That is extremely worrying,’ says Gürses. 'As a result, tech companies determine what the digital transformation looks like and institutions adapt to it.'
In a lecture during the first lockdown last year, she peeled off, layer by layer, what this way of 'going online' – in this case of the university – actually means for an organisation.
First of all, she mentions what is obvious but is often forgotten: all efforts that 'going online' requires. For example, teachers having to learn to deal with the tools and to teach in a completely different way. But also for students or supporting staff. All the work this involves tends to be overlooked, as if the digital transformation comes about all by itself, taking much of our work off our hands and making things easier for us. The retraining, adjustments and frustrations involved in a new system or tool are rarely taken into account.
In addition, Gürses shows that the logic with which digital tools are designed also shapes our way of working. Zoom, for example, was intended for a business market, with all kinds of options for managers to monitor the productivity of call participants. Some of that very American managerial logic was now seeping through into online classes, and some teachers used the feature to see if students were paying attention. Any feature out there is quickly used without considering whether it is desirable or necessary. For example, the specific logic with which a digital tool is designed influences educational practice. A logic that, for many tech tools, focuses on profit and data collection. 'All too often I see that our institutions are adapting to the way tech companies have decided to design things, rather than the other way around,' Gürses explains. There's no strategy.
'The vision should be: we want to have public institutions that serve the public in this-and-this way. And the technology we use must be implemented in line with that vision. If not, it should be a no-go.'
Current policy frameworks, such as the Digital Agenda or the Digital Roadmap, are in themselves a step in the right direction, but do not constitute the long-term vision that Gürses refers to. "An 'honest' framework, that is not a vision," Gürses says resolutely.
'Honesty... That's only about what you encounter in an algorithm, but it doesn't tell you anything about what has happened in the institution that is going to use this algorithm, in terms of its economic and administrative structure. I would like to see something where an institution says: if we reduce costs in this way, where will we be in 10 years' time, and what kind of economy have we supported with that? There is not enough insight into structural economic consequences.'
'I don't think these policy frameworks are economically thought through. I don't think ... people really understand. The idea is that we saves costs by going to the cloud, by using data tools and machine learning – fair and accountable or not. But I think most people don't realise what it means to become dependent on these services and what the political and economic consequences are. What it really means to cut a workforce and replace some of it with machine-learning algorithms. I don't think they have thought about that the staff they shed will be partly replaced with massive underpaid click-work in the production chain of these models, that these algorithms depend on all kinds of exploitation. I also don't think they understand the long-term cost to the institution itself, of the way your board functions. About what we cut away from our institutions in terms of knowledge and expertise, about what kind of human relationships and insights we delete.'
Staggering capital injections
Under the heading Programmable Infrastructures, together with Martha Poon, Gürses investigates the political economy of the mass migration to digital infrastructure, which is welcomed under the guise of innovation. 'In a way, this huge push into digital data services is due to the immense amount of investment that has been put into these types of companies since the last financial crisis.' The Sillicon Valley tech industry became a major investment vehicle and receives dizzying capital injections. In addition to numerous start-ups and platform companies that are not yet making money but are monopolising markets, cloud services and software-as-service models are promising for investors: they generate money, data and dependence. 'What we call Smart Cities, or Mobility as a Service of Smart Energy, actually expresses the ambition of cloud infrastructure to grow by integrating with existing public infrastructure. The promise that is made is that this infrastructure will be improved – more efficient, more flexible and above all programmable according to specific wishes of policy-makers.'
It remains to be seen whether such promises will be fulfilled. Critics regularly point to the high degree of tech solutionism of these applications, which reduces everything to a data problem. But the lobby is strong. All major industry players, from Microsoft to Cisco, have dedicated departments with roaring names like Country Digital Acceleration Program that "help" governments in the digital transformation. Recently, one of the top executives of the Smart City players stated that thanks to them, municipalities know that artificial intelligence (AI) and machine learning are crucial. ‘For example,’ he added, ‘for the rapid change of traffic lights.’
Marketing hype or not, the result is that each time public institutions engage in this form of 'innovation', they in a way go along with a new economic model. ‘The tech industry benefits from the failure of public institutions as a result of neoliberal policies,’ says Gürses. 'Privatisation and continued austerity are making all these institutions struggle, including health care. If the proposal is to "optimise" what's left – then that sounds alluring.'
Once again, Gürses stresses that the consequences will be greater than a classic privatisation. Just as Zoom's functionalities brought a manager's logic into the classroom, the programmable infrastructure and cloud services also come with a certain logic. Gürses predicts that this will shape our public institutions in far-reaching ways, that the push to the cloud will help determine which priorities are set and how they will be optimised. She cites the example of the mantra of 'individual optimisation' that is central to a lot of AI. 'What many people don’t realise is that this isn't about what's best for the individual at all, but ultimately how institutions can be adjusted in such a way that the distribution is most efficient for the system and people are pushed in a certain direction that is more optimal.'
A well-known example are social media algorithms that seem to tailor the content to personal preferences. But they are mainly programmed in the interest of the company, namely profitable data collection. Another well-known example is the 'individual' credit model: groups receive a 'personalised' offer with an interest rate based on a risk calculation. Proponents present this as progress, a "democratisation" of loans, because people who couldn't get a loan before now have access to credit – albeit at sky-high interest rates. Gürses calls it personalised exploitation. For her, the logic of hyper individual alignment – or rather, manipulation - has little to do with democratic access or other public values. The result is above all an even further fragmentation.
Recently, David A. Banks outlined what that personalised exploitation could mean for the city. A Subscriber City, where access to all kinds of services and places will differ depending on the subscription you have. He emphasises the disastrous consequences for the little stability and collectivity that remains in a city: the space that we share and experience together. And like Gürses, he sees how the interests of the financial markets, tech and real estate intertwine. On the other hand, speculative art projects present hyperflexible individualised experiences by using AI as the way to a truly inclusive city, which adapts to each person's individual wishes.
Either way, there's a lot at stake with the migration to cloud services and the different software-as-service models, a lot more than what current strategies regarding digitisation focus on. Understanding the potential social disruption caused by the economic model of tech companies and a life that revolves around licenses appears to be lacking. Just like an analysis of how the financial sector is a driving force behind the 'marketing' of all kinds of infrastructures by tech companies.
Feminist and queer
Whether and how a meaningful counterpower might grow from the bottom up is not an easy question. But Seda Gürses also shows that things could be done in a fundamentally different way by developing alternative models. Take, for example, POTs, i.e. Protective Optimisation Technologies, a radically different model of optimising algorithms, which protects the public interest. She and others recently founded the Institute for Technology in the Public Interest to further explore technologies and experiment with them in the public interest. Gürses emphasises that 'the public interest' is in itself evolving and can never be unambiguous. In doing so, the institute recognises the limited vision of what’s 'public' and the excluded or ignored interests of certain groups. The institute's fertile sources of inspiration for this are found in feminist and queer theory, where principles of intersectionality and decolonisation are at the forefront. Combined with computer science and design practices, interventions are emerging with which they aim to initiate a greater involvement in the design logic of technology. A first success was the DP3T project, a protocol for privacy by design of Covid tracking apps. A concrete, feasible alternative that shows that we have a choice, that the corporate models are neither necessary nor indispensable and that technology serves and must serve a public interest.
Seda Gürses is currently an associate professor at TU Delft with the Department of Multi-Actor Systems (Faculty of Technology Policy and Management, and a subsidiary of the COSIC Group of the Department of Electrical Engineering (ESAT), KU Leuven). Previously, she was a postdoctoral researcher at COSIC/ESAT, research fellow at Princeton University's Center for Information Technology and Policy, and fellow at NYU Steinhardt's Department of Media, Culture and Communications and the Information Law Institute at NYU Law School.
Her work focuses on privacy-enhancing and protective optimisation technologies (PETs and POTs), privacy engineering, as well as questions relating to software infrastructures, social justice and political economy that intersect with computer science.