In the modern web, Google and Facebook are the feudal lords and people are the peasants — at least when it comes to control of the photos, comments, 'likes' and other data that each person posts online.
To use Google+ and Facebook, people yoke themselves to the providers by handing over their data in exchange for use of the services. It's like a feudal system: the social-networking companies are sustained by the data flooding into them, and gain in power from the exchange. People upload their photos, their messages and other data from their personal life, but the service providers control how that information is presented to the world.
"The users contribute their own content to you for free. You sell it
back to them with banner ads put on there. And on top of that, you spy
on them to gather profiling data," says Michiel de Jong, of the Unhosted project to decentralise user data.
Compare this with feudal lords in the Middle Ages — 'the castles' — who took in taxes in the form of wheat, cattle and other resources, consumed them and then demanded more. The castles held all the political power and could talk to other castles, while the peasants who lived on their land had little influence, even though the resources they produced kept the castles going.
The online form of feudalism is more insidious. With Google and Facebook, the resources these castles take in — images and search terms, for example — are not used up, as they were in the original system. Instead, the data is analysed again and again, and the castle grows in power with each bite of information.
Network of data
If you have 50 friends on Facebook, you get 50 friends who can all contact you and vice versa. However, Facebook gets a network of some 1,225 possible combinations of one-to-one discussions between you and your peers, which it then makes money from through advertising.
As your friends talk to each other, they feed Facebook data about how information flows between its users. It's likely that your friends will have their own friends and will talk to them as well. Every time these first- and second-level contacts interact, it gives Facebook more pointers to where you fit within your network. To you, it's a bunch of your mates; to Facebook, it's an expanding cloud of data to be harvested.
This means that for operators of social networks, the value of each piece of data uploaded is multiplied by every other piece of data in the system. The co-inventor of the Ethernet transfer protocol, Robert Metcalfe, came up with a rule to demonstrate the information disparity between operators and users. According to Metcalfe's Law, the value of a communications network is proportional to the number of users connected to the system (illustrated below).
According to Metcalfe's Law, the value of a communications network is proportional to the number of users connected to the system. Image credit: Wikipedia
The Locker Project leader Jeremie Miller believes that once user data goes into a system, it can become difficult to get it back.
"Things like email addresses and the stuff we've uploaded become entangled in the larger systems," says Miller, whose project aims to give people ownership of their personal data. "It takes real time and effort to move/migrate/announce [data] relocations."
People mistrust this handing over of their data, in much the same way IT managers have concerns about uploading their enterprise's data into the cloud, or web users have misgivings about Gmail and Yahoo automatically scanning their emails.
"The thing that disturbs me is [more information] being centralised to Google," Alec Muffett, a security engineer with two decades of security industry experience, says. "In itself, it's not an evil thing — it's a power exchange. You have my data, I get services from you for free, but I don't believe it's the only model that need exist."
Even top executives in these companies recognise the problem. Google's former chief executive, Eric Schmidt, has admitted if you want a new identity online, you need to change your name. In addition, Facebook's former head of marketing Randi Zuckerberg has said that "anonymity on the internet has to go away".
"If a building company put up a tollway and made drivers cede ownership of their cars whenever on that tollway, the traditional justice system would prohibit that," he argued. "Yet this is exactly what is happening on the 'information highway', and the situation is largely overlooked by justice departments, who still live largely in a brick-and-mortar world."
Google and Facebook have the same business model: pull people in, make sure they stay within the technological landscape and make as much money as possible out of their data. Both services need people to be accessing pages where they can serve ads or encourage them to buy something, so they add a social element to keep them there.
For example, Facebook has brought more features within its domain — such as Skype video calling and online payment service Facebook Credits — so that people do not stray away.
For its part, Google is spread across a variety of properties. Because of this, it is salting the web with social elements in a bid to ensure that people are always logged into a Google profile account, which allows it to serve better-targeted ads. For example, it has created the +1 button to help make its search results social and has introduced badges to its news aggregation service, Google News.
Methods of control
What makes this modern feudalism powerful is that the key parties are keeping their methods of control from the users.
Neither company openly gives details to users about how their data is being used. We never see inside Google's algorithms, or gain a view of how our connections interweave with every other person on Facebook, but their services see all.
What makes this modern feudalism powerful is that the key parties are keeping their methods of control from the users.
There are exceptions to this. For example, Google exposes its inner intelligence to developers via Google App Engine, its rentable cloud platform. However, the exact goings on of the algorithms and data frameworks in the underlay is kept from developers, as they must mount their applications in a high-level language, such as Python, Java or C, to run them on the cloud service.
Facebook does this too; it rolls out features made possible by its mammoth network of data, but never gives the user access to the data itself. It also does this with hardware — its Open Compute Project purports to let outsiders build a datacentre to rival its own, but it leaves out
any mention of Fusion-io, the PCIe-linked flash hardware that gives Facebook's services their speed, in the documents it released to describe the structure of its datacentre.
Beyond this, both companies are consolidating their systems...
...around similar data analysis engines. Facebook relies heavily on Hadoop, a data analytics framework that mimics Google's internal MapReduce and Google File System tools.
Ultimately, whether a person's data is being uploaded into Facebook or Google, it is being processed in a similar way. Though the person may not be aware of it, the castles have created infrastructures amenable to the processing of each other's data. Their data analytics systems can exchange data, as the structures of Google File System and Hadoop are logically similar.
Escaping the castle grounds
Feudalism isn't necessarily a bad thing for those it exploits. It provides a measure of stability, assures a degree of security and helps to keep communities centred. But movements are gathering online to try and give people data sovereignty.
Feudalism isn't necessarily a bad thing for those it exploits. It provides a measure of stability, assures a degree of security and helps to keep communities centred.
Google's Data Liberation scheme aims to give people control over their information. However, while it allows people to load data into and extract data from a variety of user services, it doesn't shift the storage method any closer to the user: as open as the system may be, people's information will still reside in Google's infrastructure.
A wider decentralisation project is de Jong's Unhosted, which works by breaking the typical client-server relationship down into separate fragments: the server does the application's source code, the client does the presentation and processing, and a separate storage entity holds the data.
Unhosted exists, de Jong says, because "if everybody hosts their own data, then there is no central repository that needs to be financed. If storage becomes part of commodity infrastructure, then this big data [or social] service becomes part of a commodity infrastructure".
To be truly effective at liberating their users, the castles must "let every individual become their own feudal system, or at least have their own property and home", Miller says.
Leaving can be difficult, as demonstrated by one ZDNet UK reporter's attempts to transition to a purely open-source social layer via the Twitter-replacement Identi.ca, Facebook-replacement Diaspora and the Skype stand-in Jitsi. As with feudalism, if something displaces your castle, all the resources you have given it may go as well. However, if you can do it, you gain a level of control over your own data that the castles cannot give you.
Alec Muffett is working on a scheme called the Mine Project. This aims to give consumers a local place to store their credentials and sensitive data, so they can choose which services they want to expose the data to.
Movements are gathering to try and flee the feudal systems of Google and Facebook. Photo credit: One Lucky Day/Flickr
"I believe the structure of the internet encourages individuals to host their own data. In some ways, it's a little unfortunate that everyone thinks it's easier to have a big company do it on their behalf, but it's entirely understandable," he says.
Muffett believes the current social web is part of a cycle that has seen computer power oscillate between centralised and decentralised setups for decades.
In the '60s and '70s, mainframes ruled as central repositories of data, then this shifted to client PCs. With the rise of the cloud, the cycle is swinging back to centralisation.
"It's a karmic cycle," he says. But sooner or later it is going to swing the other way, and people should prepare for an age when their data is once again their own, he argues.
The success of these companies means that when new social-networking technologies are developed they tend to have a lot in common with their predecessors, according to Muffett.
"Everyone wants the Facebook experience, everyone wants the Google+ experience, which looks awfully like Facebook," he says. "There's a degree of convergent evolution and giving people what they're familiar with."
The remedy for the concentration of data into a few hands could be technologies such as Hadoop, which is used by Facebook, Twitter and others to manipulate and store the data they hold. Because Hadoop and similar tools are open source, people can easily find out what the companies are doing with the data. Also, because the tools are developed in the open, people can anticipate where the services using Hadoop are heading and make sure their data is kept easily transportable.
Even here there are caveats. For example, EMC's distribution of Hadoop uses proprietary technology from MapR. This means that though the data is mobile, some redundancy within the file system and certain high-availability features fade away when users decide to move their data to the non-commercial version. It's another form of lock-in.
There are some steps people can take to counter the castles and keep their data mobile. If you use Twitter, then cross-post your tweets to Identi.ca. With Facebook, do the same with Diaspora. With Google, keep a log of all your search recommendations.
If people keep control of the data they put into the world, they will be able to search it themselves as the social networking providers do. An open-source revolution could decentralise the data and bring the castles down.
Get the latest technology news and analysis, blogs and reviews delivered directly to your inbox with ZDNet UK's newsletters.