Platform Capitalism

The emergence of the platform model of business in which a corporation provides the tooling used or required by others to operate their own services. The platform model enables efficient and near-total harvesting of data concerning user and provider activities and can self-strengthen using this data and machine learning techniques to consolidate and increase its position, and to improve its offering; this tendency is known as a network effect and quickly leads to monopolisation.

= History = Platforms emerged from the cash glut of the dot-com boom, during which time vast sums were invested in technology companies, even when these were not profitable. How did this lead to monopolising platforms?

There is a newer trend towards growth-first startups in the technology sector: profit is not considered and it is the acquisition of data and of users that is important. The belief is that, by creating a monopoly, profit will eventually follow.

= Future = One belief regarding the eventual tendency of platforms is the segregated of each platform from its competitor - in the context of the Internet, separate networks from the physical infrastructure upwards. Such segregation would allow the platforms to gather information across the entire vertical stack and such lock-in would ensure customers become embedded within and dependant upon the infrastructure (Apple has already begun to do this with its cloud-mobile-laptop synchronicity).

= Revenue = Platforms tend to be either lean or product platforms. The former is able to survive by removing all unnecessary costs, including any assets traditionally owned by the industry (such as vehicles owned by Uber or properties owned by Airbnb); the liability and the cost of owning such assets is, instead, passed onto the ‘employees’ which are, themselves, self-employed contractors and thus not entitled to traditional benefits such as sick leave (saving additional money). The latter operates more traditionally, owning assets that are provided for public user in exchange for a fee. Amazon is an example of a product platform: the AWS service provides computing resources in exchange for a competitive fee. The advantage for Amazon, as a platform, is that it can gather data from across the entire lifecycle of its products - rather than up until it is sold, as with traditional capitalism. The additional data helps product platforms tweak their services and monopolise.

It is interesting to note that economies of scale are not available in lean model as contractors must buy them individually. This reduces the resource available to the corporation by increasing the barrier to entry for contractors. This may contribute towards an apparent trend for lean platforms to become product platforms, such as Uber procuring a fleet of self-driving vehicles. There is a prediction that all lean platforms must eventually become product platforms or cease to exist due to failure to turn profit.

In Marxist capitalist models labour can be defined as ‘work resulting in an increase in realisable value’. It is unclear, in this novel form of capitalism, where the labour originates: is it the data-producers, such as you and I on Facebook, that are labouring for free? It seems counter-intuitive to describe freely and willingly creating digital content as labour as so one argument is that there is no labour in many digital platforms - a possible explanation for the slump in global markets as the digital sector grows.

An alternative explanation is that platforms are mining data as a primary resource, extracting value from raw data as a refinery may metals from ore. In this approach, the platforms are taking unfiltered, unprocessed data (or partially-filtered and processed) and finding useful correlations, patterns or otherwise important features for use or sale. One problem with this approach, although it seems more justified that the ‘free labour’ argument, is that the value is only ever the manipulation of information - nothing tangible is ever produced or modified, simply services or understanding improved.

The argument that claims that all lean platforms must eventually transition to product platforms is based around the assumption that neither of these approaches will ever produce meaningful value and will, thus, never be profitable in the long term.

= Real World Examples = Facebook is a digital platform, freely offering the public the ability to store data, message friends and stay connected with one another. Almost all (96%) of Facebook’s revenue came from advertising data, tuned by its immense dataset on the interests and interactions of its users. Advertisements on the Facebook platform can be minutely targeted as a result of its data mining.

There has been an effort made by Facebook to install ‘free’ Internet across much of the third world. This program, known as Free Basics, allows ‘free’ access to Facebook and certain hand-picked services, whilst charging a fee for access to the rest of the web. The benefit to Facebook and any other involved parties is enormous: all user data can be stored and analysed, drawing from a much greater pool than simply those users accessing Facebook.com.

Uber is an example of a lean platform (albeit one transitioning to a product platform). The taxi provider contracts employees in thousands of cities across the globe, using a rapidly-expanding dataset to predict service peaks and to direct its fleet pre-emptively to areas of high demand. Interestingly, Uber is built on top of a host of other platforms: it uses Google Maps, a third-party payment processor and requires users to install the app on their iPhone, Android device or Windows Mobile.

Spotify and Airbnb are other examples of lean platforms.

Netflix is a hybrid platform, moving further towards becoming a product platform as it produces ever more of its own content.