By SERGIO AMADEU DA SILVEIRA*
The algorithmic catastrophe and the “blackout” cloud
1.
The sociology of modernity has produced a set of reflections that need to be deepened, especially in these times of spreading reactionary waves that coexist and feed on the rise of technologies that propose to mediate all human activities. Sociologist Ulrich Beck in Risk society, published in Germany in 1986, warned that risks and uncertainties had become central in modern societies driven by technological and industrial progress.
Ulrich Beck already pointed out that such risks would be increasingly invisible and their perception would be shaped by scientific institutions and the media. The dynamics of risk would be incorporated and the constant search for those responsible and guilty for disasters would lead us to a certain policy supported by risk management.
Ulrich Beck's perception could not be more realistic, since digital technologies have dominated the economy and large companies that control them and command their development have imposed a risk management style. Philosopher Yuk Hui opened his text Algorithmic catastrophe – the revenge of contingency, 2020, technological catastrophes are not simply material failures, but are failures of reason. Taking inspiration from Paul Virilio, Yuk Hui thinks of contemporary technological systems as carriers of catastrophes and techniques for mitigating the very tragedies that their dynamics and purposes generate.
Catastrophes are inevitable by the very nature of automation and automation technologies. Our systems are moving towards the increasing use of machine intelligence solutions based on statistics and probability converted into algorithmic systems that operate from gigantic computational power, generating models that are used to automate activities and their risk.
Norbert Wiener, in the text Some moral and technical consequences of automation, published in May 1960 in the magazine Science, declared that machines could develop unforeseen strategies, since they carried learning algorithms, which could not always be understood and followed by their programmers.
2.
What happened on July 18th and 19th, 2024 is an example of an algorithmic catastrophe. The risk management system, more precisely the mitigation of cyber attacks, failed. An error in the software update from cybersecurity company CrowdStrike that is applied to Microsoft's operating system generated what the world press called a cyber or digital blackout. A message from Microsoft on the former Twitter, now X, said: “We are aware of an issue with Windows 365 Cloud PCs caused by a recent update to the CrowdStrike Falcon Sensor software.”
Every digital system incorporates in some way the attempt to detect and contain errors, failures, attacks, in other words, risks and incidents. Therefore, there are other algorithmic systems that work all the time to analyze failures, errors and attacks. Antiviruses are an example of preventive action to protect a system from sending malicious files that can destroy information and even encrypt a database to obtain ransom from criminals who hold the key to decrypt the information. Interestingly, the problem that occurred and called “blackout” occurred when the protection or attack prevention system ended up promoting an attack on the system that was supposed to defend.
Anthony Giddens and Ulrich Beck wrote that in late modernity, risks are largely produced by society itself, mainly by technology, industrialization and globalization. However, we have been in late modernity for a long time, we are in a rotting capitalist system. The capitalist's dream is dystopian and seeks to replace human work to the extreme with automated systems with the aim of reducing costs and increasing the quality and precision of services and products by increasing productivity.
Thus, in contemporary capitalism, large technology companies advance in the incessant collection of data to improve the extraction of patterns from human, social and machinic processes. However, this dream has socio-technical consequences that are not predictable and cannot be controlled.
It is important to highlight here that the risks are combined with objectives that increase them, among which is the search for market dominance promoted by digital oligopolies, the so-called Big Techs. In the first decade of the 21st century, the business model based on so-called cloud computing spread, accelerating the concentration of computing power and data storage, and consequently, expanding economic concentration.
What is the cloud business like? What does the phrase “my data is in the cloud” mean? Cloud is a metaphor for the business of storing and processing data and systems that are located in data centers which are accessed remotely via the internet. As the joke goes “the cloud is other people’s computers”.
A few companies specialized and ended up dominating the cloud provisioning business. Amazon Web Server and Microsoft Azure, in 2021, held 60% of the global cloud market offering infrastructure as a service. What does that mean. That several companies, institutions, governments replaced their own local data processing and storage infrastructures with contracts for Amazon and Microsoft to “take care of” and “rent” data storage space and computing services.
The cloud contracting costs for companies and governments were attractive. This led to gigantic growth in this market. The consequence was more economic concentration.
According to Gartner Group, the concentration in the Cloud Infrastructure as a Service (IaaS) market was as follows in 2023: Amazon held 39%, Microsoft 23%, Google 8,2%, Alibaba 7,9%, Huawei 4,3%. These five companies dominated 82,4% of the global cloud market. Furthermore, this scenario is getting worse due to the training of large language models, LLMs, which require many available computers with very high processing capacity or computational power. Therefore, Generative Artificial Intelligence based on extracting patterns from large amounts of data is contributing to the concentration of computational power that implies economic power.
3.
On the day of the blackout, many companies went to access their applications and systems in the Microsoft cloud and were faced with the famous blue screen, meaning the operating system was unable to work. Many people who had Microsoft 365 also had access to their files locked. Microsoft 365 is like a subscription service that gives users access to the Office suite and other services over the internet, instead of installing them locally on their own machines.
This means that users' data and files are stored in the Microsoft cloud, allowing them to access their documents and information from anywhere with an internet connection. Except when the company that offers the service has a failure, an attack or blocks, whether intentional or not.
The blackout demonstrated the gigantic power of a mediator of digital relations and a data processing operator like Microsoft. Without a doubt, the unintentional failure caused the blackout. But, it is clear that Microsoft has the power to block companies and institutions from accessing their own data located on their data centers, far from our jurisdiction and our ability to physically access it.
We have a problem of digital sovereignty there. Brazilian State leaders need to assess the risks of continuing to host their strategic data and use everyday software in infrastructures outside our country. Our universities need to debate whether it would not be essential to keep their communication and research data in infrastructures installed in our country, in our jurisdiction and submitted to our ethics committees. The autonomy necessary for development increasingly involves digital sovereignty.
*Sergio Amadeu da Silveira is a professor at the Federal University of ABC. Author, among other books, of Data colonialism: how the algorithmic trench operates in the neoliberal war (Literary Autonomy). [https://amzn.to/3ZZjDfb]
References
BECK, Ulrich; GIDDENS, Anthony; LASH, Scott. The reinvention of politics: Towards a theory of reflexive modernization. Stanford University Press, 1994.
BECK, Ulrich. risk society. São Paulo: Editora 34, 2010.
GARTNER Says Worldwide IaaS Public Cloud Services Revenue Grew 16.2% in 2023
STAMFORD, Conn., July 22, 2024. Link: https://www.gartner.com/en/newsroom/press-releases/2024-07-22-gartner-says-worldwide-iaas-public-cloud-services-revenue-grew-16-point-2-percent-in-2023
HUI, Yuk. Algorithmic catastrophe. The revenge of contingency. Parrhesia: A Journal of Critical Philosophy, no. 34, 2020.
WIENER, Norbert. Some Moral and Technical Consequences of Automation: As machines learn they may develop unforeseen strategies at rates that baffle their programmers. Science, v. 131, no. 3410, p. 1355-1358, 1960.
the earth is round there is thanks to our readers and supporters.
Help us keep this idea going.
CONTRIBUTE
