Tommi Space

The power of decentralization

This is my final essay for VIU’s Global Internet Governance course.

Preface

The “original sin” of the World Wide Web could be traced back to the anarchism and liberty that are intrinsic in its architecture. From a merely technical point of view, nobody requires a permission from anyone in order to publish a website. The same goes for connecting to one of the infinitely many servers which compose this immense and ubiquitous entity we call the Internet. As already theorized by (Huberman and Adamic 1999), its structure by design enforces a power law describing how the few websites which have a great number of pages grow at an almost exponential rate, while the others are left out in the cold.

Even though the nature of the internet is decentralized—it is not anyone’s property, and any actor is free to openly take part to it (both passively and actively). There are, however, some for-profit entities that are monopolizing its traffic and becoming the main source and destination of virtually all online activities.

This essay focuses on the side effects of this phenomenon with respect to Online Social Networks (OSNs). It then proceeds to analyze how oligopolistic centralization of platforms strongly limits users’ rights. Lastly, it is highlighted how a process of (re)decentralization could help address or even solve some of the most pressing issues.

Note

The present essay will insistently and extensively focus on very few firms in order to provide examples. This is not intended to discriminate those firms specifically, but it is because undeniably their activity is so extensive that it is very rare to observe other less known or smaller actors behaving similarly.

This very point reinforces the thesis sustained by this essay: a tiny number of stakeholders is responsible for most of the discussions concerning the Internet of today and of tomorrow.

Introduction

As analyzed by (Bahri, Carminati, and Ferrari 2018), logical centralization consists in the unprecedented massive and uncontrolled collection and aggregation of different types of data about millions of individuals, from across all areas of the globe, in the hands of a few centralized entities. This results in what (Reviglio and Agosti 2020) call a control asymmetry:

A small group of platforms—above all Google, YouTube, Twitter and Facebook—act as the ultimate gatekeepers for billions of users worldwide. They provide a personalized experience, but little individual or group user control over information filtering processes.

Therefore, the core problem of centralization can be reduced to the fact that not only the ownership, but also the control and the management of OSNs is concentrated in the hands of few corporations. Their good will becomes the conditio sine equa non for smaller presences on the web.

The web is undeniably experiencing a drastic growth, which nevertheless appears to be uneven. Do efforts towards [[Decentralization]] help to achieve more fairness? Why should a more technical matter come before regulation? In the following points I am going to argue that, when an immense amount of power is wielded by very few supranational for-profits, struggle to curb their influence through regulation is in vain if not backed by more technically oriented propositions, specifically pushing towards decentralization.

Algorithms as Black-Boxes

Mark Zuckerberg, the CEO of Meta (formerly known as Facebook), reportedly stated (Edelman 2021) that his company’s algorithms are built to foster meaningful social interactions, while critics rage in claiming that the firm’s main (or only) purpose is to maximize engagement, thus profitability.
Regardless of the opinions and speculations by scholars and by the public, the worrying point does not originate in how the software is used, but—again—in (at least) understanding how it is built. Let alone the thorny issue of algorithms potentially overlooking the user’s wellbeing, we should start by acknowledging that there is no way for anyone outside the companies owning mainstream OSNs to even know how those algorithms work.

Personalization is the primary scope of the algorithms that decide what to show first in a “feed”. A question insistently posed but remaining still unsolved is: what are the criteria according to which such evaluation is carried out? The ultimate purpose is interpreted as pursuing the logic of market segmentation until each individual user is reduced to a unique market (Yeung 2018), yet this process occurs largely beyond the control of users as it is based on implicit personalization—behavioral data collected from subconscious activity—rather than on deliberate and expressed preferences (Reviglio and Agosti 2020). According to the authors, research on personalization is contradictory and ambiguous, hence not completely reliable. Unless and until more details will be provided by companies, it will continue to be tremendously hard to draw conclusions and scientifically assess the situation.

The most straightforward, yet ineffective, measure taken in this direction by “Big Tech” is to enact a superficial transparency, by overwhelming new subscribers with thousands of obscure sentences in strictly technical language. Labeling them as Terms of use, or, alternatively, providing extremely oversimplified and trivial summaries, means everything and nothing at the same time. Instead of clarifying how OSNs work and how their owners operate, Terms of use confuse and disorient the user. It is widespread common knowledge that, in the end, only a tiny fraction of people actually reads them before ticking on the box to accept them. Furthermore, explaining how data is collected, in very broad terms, how platforms work does not mean giving users an actual choice.
The so flaunted and acclaimed transparency eventually turns out to be the perfect cover-up (Hoepman 2018).

What activists, regulators and users should in the first place be asking for is algorithmic sovereignty, and pursue its achievement through user empowerment (Reviglio and Agosti 2020).

First of all, and more obviously, there is a general need for “academic literacy”. We can define it as the basic knowledge on how filtering mechanisms and design choices function and what their impact on one’s own life is and may be.

After a common, basic yet fundamental understanding of how AI works is somehow assured to be provided to users, OSN providers should empower them by allowing each individual to arbitrarily decide what is important for her/him; personalization should be explicit, and (Reviglio and Agosti 2020) argue it should be so for political news in particular.

Ultimately, it appears self-evident that it is very difficult to establish alternatives that question the very principles on which the current business model and its perverse consequences exist, nevertheless it is crucial for personalization algorithms’ scope to mutate from a form of legitimate hedonistic subjugation to an opportunity for new forms of individual liberation and social awareness (Reviglio and Agosti 2020).

As long as OSN platforms will keep retaining to their own evaluation and self-regulation tools and instructions to access the heart of the machine learning driving their infrastructures, nobody will be able to improve the long-term trend.

Proprietary and open source software

Despite (Roio 2018) argues that “openness” and “transparency” are not sufficient conditions to realise the good ethical propositions of the free software movement, it must be noted that they may not be enough, but they are arguably necessary. In the scenario under analysis, we are dealing with software whose functioning, even if thoroughly explained, may still not actually exactly correspond to what its source code instructs it to do.

Every program run and used without allowing the public to access the source code is defined as “proprietary software”. Although intuitively it may not seem so, secrecy concerning the code does not simply correspond to rightful ownership of programs, and it does not automatically guarantee fair competition. As (Stachwell 2005) points out, open source efforts most often held up as examples of how peer production can prosper in the modern information economy have succeeded in spite of, rather than because of, contemporary intellectual property controls..

The true obstacle, rather than limit, represented by proprietary (“close-sourced”) software is strictly intertwined with the topic analyzed in the previous section: it profoundly limits “auditing”, hence thoroughly exploring and understanding the low-level backbone upon which platforms are built. Auditing can happen anyways, but it is bound to remain indirect (Adler et al. 2016).

“GAFAM” (Google, Amazon, Facebook, Apple, Microsoft) profusely proclaim their love for open source, supporting independent projects and even developing widely used tools that are open source themselves (among many others, Google’s Go and Apple’s Swift programming languages, Facebook’s React framework), hence proving commitment is not a purely formal good will parade. Nevertheless, the trend appears to be recurring, and the problem is tackled only superficially: what is open source are the “extras”, distracting more technically-oriented communities, such as independent developers, who are provided for free with appealing feature-packed cutting edge technology for free.

In this case, too, centralization plays a major detrimental role: if there is only one entity possessing the source code of a program, it is the only one that can run it, and—necessarily—the sole provider.

The case of Truth Social

The chief display of the relevance centralized OSNs monopolies exert can be found in their relation with Donald Trump, former president of the United States of America.

After Mr. Trump repeated aggressive behavior and spread of misinformation through social media platforms during his presidency (Nacos, Shapiro, and Bloch-Elkon 2020), he was eventually permanently banned or temporarily suspended from all major social media platforms, by the end of February 20211. The counter-attack by the American billionaire has been to announce to open his own platform, named Truth Social.

With a great surprise of the Free Software community, Truth Social appeared to be based on the most common and known open source alternative social media platform, Mastodon. Glossing over the many speculations concerning Trump’s awareness of the value of the choice to base his platform on an open one, this is a master proof of the merit that a software that can easily be copied and modified has.

Nonetheless, Trump Media & Technology Group (TMTG) underestimated one little but ineludible detail: Mastodon’s GNU Affero General Public License, according to whose deed it is possible to copy and modify the source code of the original program as long as the source code of its by-products stays open, too. As of October 23rd 2021, Mastodon source code was open, while Truth Social one was not. Once Eugen Rochko, Mastodon founder and current maintainer, realized there was something off, he filed a complaint to TMTG, forcing it to make Truth Social’s source public. On November 12th 2021 Trump’s platform code was made open, with a bold Our goal is to support the open source community no matter what your political beliefs are on the page that displays it.

One might consider the resolution of this case a triumph of open source advocates, or read it as a great contradiction of a powerful and capable Company bending to some legal quibble, but without any argument this proves the rising relevance of the open source movement and its ever improving software.

APIs and Interoperability

In Interoperability (Wegner 1996) defines the term as:

…the ability of two or more software components to cooperate despite differences in language, interface, and execution platform. It is a scalable form of reusability, being concerned with the reuse of server resources by clients whose accessing mechanisms may be plug-incompatible with sockets of the server.

In other words, interoperability makes different digital products or tools work together since they are designed to communicate among each other. In computer engineering, interoperability is defined through APIs, which stand for Application Programming Interfaces. APIs are developed and deployed by OSNs owners as conditions to allow or block access to their infrastructure, safeguarding their dominant exclusive position (Bodle 2011). (Milberry and Anderson 2009) support this by sustaining that in their drive to enclose the Internet, online media companies develop synergistic membranes with prescribed circuits that lead less and less to the global Web outside their online properties.

Interoperability is then sacrificed, or better, selectively enacted, in order to maintain control over competitors’ possible interactions with one’s own platforms. With the exception of some tools—such as Facebook’s Open Graph Protocol, which has now become a widespread standard for pre-fetching website metadata— most APIs are intended to keep the user base gravitating around the service, never permanently leaving it.

Always focusing on the world-famous Menlo Park company, (Bodle 2011) underscores:

Facebook’s development and integration of Open APIs demonstrate a strat- egy of controlled openness, innovation, and dominance. Facebook utilizes Open APIs to provide valuable user data to developers and online partners in order to encourage the proliferation of social applications that ultimately harness inbound links from external sites and devices by redirecting traffic to Facebook’s servers.

This observation opens also to the topic of user’s data collection: if the vast majority (if not the totality) of online services offers the possibility to log in via Facebook (or any other authentication provider with Open APIs), it paves the way to an internet where all the activities performed by one user are reported back to a central central profile on one platform, which becomes the only holder of an enormous amount of data for many different activities.

In order to limit this behavior, which contributes to blow up OSNs traffic statistics also when users are visiting other websites, on the celebrated Electronic Frontier Foundation (EFF)’s blog, science fiction writer and open web activist Cory Doctorow proposes the term adversarial interoperability as a value thriving in the original Web, and that should be restored today. The adjective Adversarial is associated to the above explained Interoperability to underscore how the latter should be possible without the company that delivers the APIs.
A master example of Adversarial Interoperability today is email. If the email provider of the sender is different from the one of the receiver, the message goes through anyways. The reason is that the underlying technology is common and it is independent by the company offering email hosting; in order for me to send an email, my provider needs to comply to a technical, non governmental, standard which is independent from it (Simpson 2016).

Social media platforms should work like email. To post something, I should not be tied to the server or the platform I post from, and also who is not a subscriber could see what I published and interact with me. The current brand-focused development of technology make us believe that envisioning such a perspective is utopistic, but we will see it is not.

Surveillance Capitalism

All of the topics under scrutiny up to this point converge to an effect of very broad scope and influence, with deep socio-economical roots. The invention of the term Surveillance Capitalism is attributed to Dr. Shoshana Zuboff, whose writings brought a new, watershed perspective to OSNs capitalization and its effects on society.

Surveillance Capitalism intends to describe a current advanced condition of capitalism, which is able to influence consumers in a so capillary way that it becomes surveillance.

First of all, (Zuboff 2015) starts by observing computer devices as entities whose mediation becomes crucial in order for humans not only to communicate, but also to work and perform most of their daily activities. Nonetheless, computers are not passive, and their operating systems, as well as programs they run, are programmed to accumulate data. Such extensive and comprehensive amount of information does not quietly rest undisturbed in the memory of every device, but it is processed and analyzed. In a vicious circle, the more users provide information, the easier it is for algorithms to group them according to their interests, and shows them the advertisement that suits them the most (Mosco 2018).

This process happens perpetually under the hood, concealed by a fancy design and edulcorated graphics full of pictures and content which aims at stimulating the user and gluing her/him to the screen, no matter if the content is violent or too strong for her/his eyes (Gainsbury et al. 2016). What is worse, is that user profilation achieved a level as advanced as no mass media before, and it follows that advertisements and even political manipulation can happen with unprecedented effectiveness.

At the end of her 2015 work Dr. Zuboff summarizes her conclusions with so wonderfully to appear profoundly worrying indeed:

New monetization opportunities are thus associated with a new global architecture of data capture and analysis that produces rewards and punishments aimed at modifying and commoditizing behavior for profit. […] Structural asymmetries of knowledge and rights made it impossible for people to learn about these practices. […] These developments became the basis for a fully institutionalized new logic of accumulation that I have called surveillance capitalism. In this new regime, a global architecture of computer mediation turns the electronic text of the bounded organization into an intelligent world-spanning organism that I call Big Other. […]

It is impossible to keep ourselves from wondering how the Big Other’s power could be stopped, contained, or—at least—avoided. In my opinion the most effective solution, again, is not in trying to politically reform or regulate internet monopolies as they currently are. The only way to achieve a long-term stability which does not have to be kept from degenerating is to educate users and empower them so that they will be able to see the best in healthier and better alternatives to centralized OSNs.

Such alternatives exists, and arguably the sole element keeping them from taking off is a very low user base.

Decentralized social media: the Fediverse

How are alternative social media different from the ones most of the world use?
The key to the answer to these questions, once again, is found in the platforms’ fundamental architecture. An alternative social media is any media production that challenges at least implicitly, actual concentrations of media power, whatever form those concentrations may take in different locations as (Curran and Couldry 2003) explain. More specifically, Alternative Online Social Media Networks do not have one single provider, one unique version, one gatekeeper deciding everything about it. Alternative microblogging platforms are very different between each other and their unique feature is precisely their diversity, which allows them to be connected nonetheless, thanks to a common technology.

Alternative Online Social Networks (AOSNs) transforms deficits of mainstream platforms in points of strength, since by design they are conceived to solve what ruined their capital-driven predecessors.

  1. No algorithms are involved: the problem of algorithms opacity fades away since they are not exploited to process user-generated data. Posts are displayed chronologically
  2. Open Source Software: no need to speculate on how things actually work. Transparency and replicability are chief.
  3. Interoperability as the motto: AOSNs are based on the principle that there is no main character involved in their deployment. No central processing unit exists, while the value of the social interactions they foster strictly depends by their swarm-like architecture.
  4. No gain, no pain: by design, there is no prospect of profitability for what concerns AOSNs. Although there are exceptions regarding new surging blockchain-based networks (Guidi et al. 2018), AOSNs are community-driven, and their business model is decided by the administrator of every single server, even though it is often donations collection.

By putting all of the above points together, we get what is commonly called the Fediverse, an abbreviation standing for federated universe (Reviglio and Agosti 2020). The most diffused protocol upon which the Fediverse works is ActivityPub, which fully incarnates the fundamental values of adversarial interoperability: it is developed by the World Wide Web Consortium, an entity not linked to any specific company or government. Nevertheless, there are other several relevant technologies, such as MIT’s Solid, led by Tim Berners Lee, the inventor of the World Wide Web.

From those common and open source specifications, a considerable amount of independently or community developed projects was born, the main one being Mastodon, whose history has been already unfolded before, counting more than four million users worldwide and growing (La Cava, Greco, and Tagarelli 2021).

The distributed core-less nature of decentralized social media networks restores what should have always been networking platforms’ main goal: grouping individuals, offering communities a way to stay in touch, enhancing human interactions without any second purpose driven by economical interest (Zulli, Liu, and Gehl 2020). What is argued by (Zuboff 2015) and (Reviglio and Agosti 2020) finds a meeting point in a process of platforms decentralization, in which the gap among those who provide a service and the ones benefit from it is small and it could be easily filled with a message.

A focus on community

In a world where virtual relations wander inside a universe of federated platforms, the ultimate stakeholder is not a board of navigated moneymen, but the ensemble of communities that form it. There is no single point of failure not only from a technical perspective, but there is nobody having the last decision-wise too. New subscribers can freely decide which one of the infinitely many instances run by different software suits her/him the most, and once in it, she/he automatically signs a tacit (yet explicit) contract to meld into the values she/he chose that specific platform for.

Terms of Service, Community Guidelines, Privacy Policy, and all of the elements necessary to keep people together are not handed over from above by some unknown figure, but they are defined commonly and nobody is forced to agree to them, since signing up elsewhere is very easy.
In the case of a report or complaint, content moderation is almost trivial, since all over the world instances hundreds of thousands of users big can be counted on the fingers of one hand, while all the others’ user count does not surpass five figures2. There is no need for under-payed private contractors to stare at a screen for countless hours to take down hate speech3. Users become their own moderators, since when a post from an user subscribed to another instance, appears as offensive to a transgender member of a community whose Code of Conduct explicitly refuses transfobia, the community administrator is permitted to make that post disappear from her/his instance without any further step.

Communities empowerment online brings back the quintessential value of plurality: there may be servers hosting one or little more users, who find their actual interactions only by connecting to “worlds” (figuratively meaning servers and spheres of thinking) that can happen to be profoundly different by hers/his one. Communities as well as individuals unable to live together with others end up isolated, or de-federated. It is the case of Gab social network, originally part of the Fediverse, but gradually disconnected by almost all of the other instances4.

Immortality

The ultimate superpower of the Fediverse is its immortality. Even if the bigger instances will be shut down and software development will stop, the open source of every single aspect of it enables any student of Computer Engineering, as well as any tech enthusiast or company to install the software on a server and keep the Fediverse alive. There is no necessity to “monetize” anything, since there are no capitals required to keep things running. Investments and funding are very welcomed and happily spent on improving the software, but very little is needed to simply make things work.

What is almost romantic about decentralization is its intrinsic value of undefeatability. Even the stoical resistance of one single node can maintain alive the hope that the whole net will be brought back.

Conclusion

The core problem highlighted at the beginning can be translated, under the light of the arguments provided, as a growing trend to virtualize and digitalize everything, keeping the most information possible under control, and transforming Social Networking platforms as the purpose rather than the tool to improve interactions. As pointed out in The Question Concerning Technology (Heidegger 1954) great is the peril stemming from this:

modern technology too is a means to an end. That is why the instrumental conception of technology conditions every attempt to bring man into the right relation to technology. Everything depends on our manipulating technology in the proper manner as a means.

Decentralization spreads and fragments the power of monopolizing intentions and attention, allowing users to restore their humanity and their need for relationship without being driven by foolish tendencies. Only by removing the possibility of living in a lonely swallowed bubble, everyone might find its place.

References

  • Adler, Philip, Casey Falk, Sorelle A. Friedler, Gabriel Rybeck, Carlos Scheidegger, Brandon Smith, and Suresh Venkatasubramanian. 2016. Auditing Black-box Models for Indirect Influence. arXiv:1602.07043 [Cs, Stat], November.
  • Bahri, Leila, Barbara Carminati, and Elena Ferrari. 2018. Decentralized Privacy Preserving Services for Online Social Networks. Online Social Networks and Media 6 (June): 18–25.
  • Bodle, Robert. 2011. Regimes of Sharing. Information, Communication & Society 14 (3): 320–37.
  • Curran, James P., and Nick Couldry. 2003. The Paradox of Media Power. Edited by James P. Curran and Nick Couldry, 3–16. Rowman & Littlefield Publishers.
  • Edelman, Gilad. 2021. Social Media CEOs Can’t Defend Their Business Model. Wired, March.
  • Gainsbury, Sally M., Paul Delfabbro, Daniel L. King, and Nerilee Hing. 2016. An Exploratory Study of Gambling Operators’ Use of Social Media and the Latent Messages Conveyed. Journal of Gambling Studies 32: 125–41.
  • Guidi, Barbara, Marco Conti, Andrea Passarella, and Laura Ricci. 2018. Managing Social Contents in Decentralized Online Social Networks: A Survey. Online Social Networks and Media 7 (September): 12–29.
  • Heidegger, Martin. 1954. The Question Concerning Technology. Garland Publishing.
  • Hoepman, Jaap-Henk. 2018. TRANSPARENCY IS THE PERFECT COVER-UP (IF THE SUN DOES NOT SHINE). In BEING PROFILED, edited by Emre Bayamlioğlu, Irina Baraliuc, Liisa Janssens, and Mirelle Hildebrandt, 46–51. COGITAS ERGO SUM: 10 Years of Profiling the European Citizen. Amsterdam University Press.
  • Huberman, Bernardo A., and Lada A. Adamic. 1999. Growth Dynamics of the World-Wide Web. Nature 401 (6749): 131–31.
  • La Cava, Lucio, Sergio Greco, and Andrea Tagarelli. 2021. Understanding the Growth of the Fediverse Through the Lens of Mastodon. June.
  • Milberry, Kate, and Steve Anderson. 2009. Open Sourcing Our Way to an Online Commons: Contesting Corporate Impermeability in the New Media Ecology. Journal of Communication Inquiry 33 (4): 393–412.
  • Mosco, Vincent. 2018. A Critical Perspective on the Post-Internet World. Javnost - The Public 25 (1-2): 210–17.
  • Nacos, Brigitte L., Robert Y. Shapiro, and Yaeli Bloch-Elkon. 2020. Donald Trump: Aggressive Rhetoric and Political Violence. Perspectives on Terrorism 14 (5): 2–25.
  • Reviglio, Urbano, and Claudio Agosti. 2020. Thinking Outside the Black-Box: The Case for ‘Algorithmic Sovereignty’ in Social Media. Social Media + Society 6: 1-12.
  • Roio, Denis. 2018. Algorithmic Sovereignty. Thesis, University of Plymouth.
  • Simpson, Joel. 2016. Email Archiving Systems Interoperability.
  • Stachwell, Matthew D. 2005. Tao of Open Source: Minimum Action for Maximum Gain. Berkeley Technology Law Journal 20 (4): 1758–79.
  • Wegner, Peter. 1996. Interoperability. ACM Computing Surveys 28 (1): 285–87.
  • Yeung, Karen. 2018. Five Fears about Mass Predictive Personalization in an Age of Surveillance Capitalism. International Data Privacy Law 8 (3): 258–69.
  • Zuboff, Shoshana. 2015. Big Other: Surveillance Capitalism and the Prospects of an Information Civilization. Rochester, NY: Social Science Research Network.
  • Zulli, Diana, Miao Liu, and Robert Gehl. 2020. Rethinking the ‘Social’ in ‘Social Media’: Insights into Topology, Abstraction, and Scale on the Mastodon Social Network. New Media & Society 22 (7): 1188–1205.

    1. What Happened When Trump Was Banned on Social Media (2021), by D.Alba and E.Koeze on The New York Times↩︎

    2. The Federation: Fediverse user-count and statistics↩︎

    3. Bodies in seats (2019), an investigative article by Casey Newton on The Verge↩︎
    4. How the biggest decentralized social network is dealing with its Nazi problem (2019), by Alex Castro on The Verge↩︎
    🔎