A couple of years ago, I was asked to write about co-operative visions of technology. This never led to a publication due to a variety of different reasons. However, I did write a fairly decent draft of a short, ‘popular’ contribution on cloud platforms. I still think this is valid, but I’m not going to be pushing this anywhere I don’t think. So, I thought I would publish it on my blog so that if anyone finds it useful, they can use it, and build upon it…
Amazon, Google, Uber, Facebook, and almost all other major technology companies run on cloud technology. Clouds are a staple of today’s economy, efficiently distributing computing tasks on a large scale – but they are controlled by a few powerful companies. The centralization of cloud infrastructures that host sharing platforms has led to a corporate monopoly of service. If this centralization of corporate wealth and power in cloud infrastructures is to be challenged, platform co-ops must have access to bold alternatives. In this piece, I argue that developing alternative (co-operative) business models, that can provide better protection for privacy and security, need platform co-ops to become part of an ecosystem of co-operative cloud ownership of infrastructure and technology.
Challenging the corporate monopoly on platforms must be in step with developing co-operative clouds – as one reinforces the other. Co-operative clouds not only allow for citizens to take back control of common economic, social, and cultural needs from corporations, but allow for an ecosystem of platform co-ops to develop. This is essential because as cloud infrastructure is owned by so few, corporations can extract large rents – sapping money out of platform co-ops and often allowing corporates to invest in their own platforms to the detriment of a co-operative ecosystem. Co-operative clouds instead should offer a broadly-owned and democratically-controlled alternative, sharing their proceeds in the co-operative and/or the community. In this piece I:
- Provide an overview of what clouds are and how corporations operate them;
- What co-operatives are doing now, and;
- What secure decentralized co-operative clouds can offer for the future.
Clouds could be described as a large collection of servers that distribute computing resources. But they are also political in who controls them and what they are used for. Corporations provide cloud services for lots of internet applications with some of the largest being Amazon Web Services (AWS), Microsoft, and Rackspace, extracting huge operating profits (AWS generated $1.35bn alone in the last quarter of 2017, roughly 64% of all operating income for Amazon). These corporate cloud providers frequently operate gigantic data centers, such as in the deserts of Arizona or the forests of Oregon, with questionable environmental impact.
Corporations have dominated the sharing economy providing free or rented services and products. This is structured under three cloud layers:
- Infrastructure as a Service
- Platform as a Service
- Software as a Service
These three services are layered with infrastructure at the bottom and software on top. Cloud Infrastructures includes the hardware, storage, networking, and physical data centers. Cloud platforms include spaces for app development, database management, analytics, and operating systems. Cloud software includes the way most internet services are provided to customers such as Google Docs, Dropbox, and Slack.
As current cloud infrastructure is centralized in data centers, large corporations are able to monopolize their position through scale. The up-front costs of buying servers, deploying networks, and maintaining data centers stifles cloud providers that cannot compete on similar levels. This has implications for who controls the cloud, limits innovation, and diminishes chances for large-scale platform co-ops supported by co-operative clouds to exist. Clouds do not need to be based in massive data centers. All cloud infrastructures require are computers tied together and an enforcement of rules; so that everything works efficiently, securely, and doesn’t go down.
Rethinking the cloud means we rethink the whole Internet economy. Corporate clouds directly support most internet services, extracting big data often undermining privacy, and money through rents, that sit at the core of its monopolistic working model. New ways of engaging online through reinvestment, support, and generating a platform co-op ecosystem can go some way to remedy concerns of dominant corporations. As New School professor David Carroll, who has written about co-operative alternatives has said, “the corporate cloud, really, is just someone else’s computer; it is at odds with platform coop ethics, especially when we realize we’re just renting access and computation.” So how do we become co-operative owners of cloud technology and challenge the platforms of today?
There are several co-operative clouds either already out there or in the prototype stage that we can look to. However, a lot of these efforts are currently disjointed – and are heavily reliant on some parts of co-operative clouds being run by corporations. Below I outline the important contribution each layer must make to a full realization of co-operative clouds, with some examples of alternatives.
Infrastructure as a Co-operative
Owning the physical infrastructure that underpins cloud computing is essential to break the corporate monopoly on which most platforms rely, including potential platform co-ops. Tech worker co-ops like Sheffield-based Webarchitects.coop offer a different model, collectively owning their infrastructure, running on green energy. Another cloud co-op is May First/People Link in the USA and Mexico, which own servers, deploying the open-source cloud collaboration software ownCloud, as part of a broader social action movement. Crucially, cooperative ownership of infrastructure allows for the fairer distribution of resources, community development, and sharing that is not possible under corporate clouds and their rents. Yet these have not scaled to a point to deem them as a viable alternative.
Platform as a Co-operative
To generate a non-extractive platform co-op economy, these platforms should be hosted by co-operative cloud infrastructures to develop a co-operative business ecosystem. Platform clouds provide managed services, such as web hosting. Webhosting.coop is an example, but its servers are rented from the corporate cloud provider Rackspace. This cloud layer is crucial if a platform co-op wants to build an app. In particular if it wishes to run services on behalf of customers without having to worry about buying servers or the underlying cloud infrastructure. This means that platform co-ops must be hosted on co-operative clouds as part of a broader co-operative ecosystem to restrict money flowing out to corporate providers.
Software as a Co-operative
This front-facing cloud for customers means that platform co-ops can deliver services to customers by providing interfaces and tools. This can include replacements for many major corporate equivalents and new platform co-op services. Examples include the tech co-op CommonsCloud, or open-app ecosystems such as Sandstorm.io, that could be hosted by co-operatives to provide alternatives to the dominance of Microsoft Office, email services and other apps.
All three cloud layers must come together in a full stack to launch a true alternative to the sharing economy. Platform co-ops will require cloud infrastructure, platform and software layers to work together to escape the stranglehold of corporations. At present, we have some options, but it is not enough.
Co-operative P2P Clouds
A radical alternative to relying on centralized cloud infrastructures is decentralised P2P (peer-to-peer) clouds. P2P is an automated protocol-based distribution mechanism that can challenge the power of corporate clouds. Providing P2P alternatives dismantles the likelihood of large, extractive cloud infrastructures that fund and make profitable corporate platforms (think Amazon’s AWS supporting its loss-making international operations). The difference between a centralized cloud (1) and a P2P cloud (2) is shown in the diagram. In the figure, (1) computers and devices connect with centralized (often corporate) clouds. In (2) the central actor is removed, with computers and devices sharing resources between them. Clearly, no system loses its powerful actors, with some P2P co-operative clouds requiring servers to maintain availability (preferably provided by worker co-ops) and brokers to manage tools for distributing and securing data. Yet, the difference is that no one person or organization owns the majority of the underlying infrastructure.
There has been little advanced technical development of P2P co-operative clouds, due to the difficulty in implementation, but there are options out there. One way forward would be a co-operative version of Storj.io for cloud storage. For distribution, they are developing ‘smart’ contracting between the person who owns the data and those who store it. Smart contracting could straightforwardly be adopted in a co-operative model, with individuals or other co-ops offering spare computing capacity to a co-operative cloud infrastructure. Availability and reliability is maintained by data shards that are distributed among different computers with multiple copies. Most importantly, co-operative clouds can be as secure if not more than corporate clouds where data may not be encrypted before it is sent to them. Co-operative clouds would secure data through encryption (with the individual or platform co-op holding the keys to their data). As the shards are distributed across many computers in the cloud co-operative, it maintains both security (through encryption) but also enhances privacy as no one organization can access or hold all data. This is similar to work by MIDATA.coop, which provides enhanced privacy and security of health data for medical research. However, this goes a step further so that the infrastructure is decentralized, where platform co-ops could develop similar models to privacy and security.
Storage should not be the only aim of cloud co-operatives, but about allowing platform co-ops to interactively use and scale their operations. Promising avenues emerge in P2P social networking. One example is Diaspora, a decentralized social network, with ‘pods’ powered by servers connecting to one another, avoiding centralized control. Social.coop is another example of this form of infrastructure, where individuals can work with instances of the social network Mastadon, rather than centralized clouds that extract your data for free services. If you do not trust a pod, you can set up your own, which improves privacy as no one outside the pod owns or can collect your data. This provides a true alternative to the likes of Facebook, which analyze and monetize their users. Importantly, it shows how interactive work can be done on P2P networks – and could be scaled to provide not only storage but hosting of platform co-ops.
Though we’re not yet there with P2P co-operative clouds, there are both storage and distributed social P2P networks, which are not far from clouds that can store, host, and process data. The challenge for cloud co-operatives is to develop an ecosystem of layers with a fundamental reconfiguring of cloud infrastructure. This does not mean we abandon current cloud infrastructure co-operatives but build them in broader networks. Co-operative P2P clouds can be secure, privacy-enhancing, and scalable while also sharing proceeds in a co-operative ethic. Without this reconfiguration, rents to host platform co-ops will flow to a few corporations, allowing them to reinvest in their own platforms, killing off any emergent platform co-op ecosystem. This is an opportunity to move from the large data centers of the corporate sharing economy to a decentralized ecosystem of co-operative clouds which can help platform co-ops to thrive.
 Carroll, D. (2017). “A Different Kind of Startup Is Possible.” In Scholz, T. and Schneider, N. eds. Ours to hack and to own: The rise of platform cooperativism, a new vision for the future of work and a fairer internet. OR books, pp. 113 – 118.