This is the third and final section of "The liberty of the Networked".
In part 1 of this essay I used Benjamin Constant's characterisation of the modern, individualised liberties as being dependent on the republican liberty of collective self-determination to characterise the ways in which technology can be seen to be simultaneously freedom enhancing while also dauntingly threatening. Part 2 of the essay considers the specific ways in which freedom-enhancing characteristics can become pathological -- anomie and alienation are pathologies of private liberty; social tyranny is a pathology of collective self-determination. Technology offers them all new and frightening scope.
In this part, I look at the very Californian view of the web as a new medium for emergent, quasi-market phenomena. The analogy has its attractions--especially for analyses of the web--but the centralising dataphagous mechanisms driven by Web2.0 advertising business models should cause alarm in a way that markets themselves need not.
- How are Web2.0 and market mechanisms different? Google or Wikipedia
- Picking the battles
There is a temptation to equate Web2.0 mechanisms with disaggregated markets. This is partly useful. For example, Google can be seen as a great auction for attention, with payment made in links (Curzon Price, 2008). Or Wikipedia can be seen as eliciting evolutionarily stable memes. (A mutation is a change in a Wikipedia article ...does it survive the environment of selection?).
But in other ways, the analogy misses important centralising features of Web2.0 services like Google. In the realm of attention, Google might also be thought of as the ultimate central planner. Google established its dominant position in search through its patent on the PageRank algorithm. In essence, PageRank counts a link to a page as a ``vote" for that page. These votes are weighted by the number of votes that a page itself receives. If a page that many people link to links to yours, that is considered an important vote of confidence. In signalling-theory terms, linking to another page is considered to be a costly and therefore meaningful signal of value. PageRank worked very well in the early days of the web, but its own dominance has reduced the meaning contained in each link as content providers self-consciously try to increase their page ranks. The number of links to pages are like bids into an auction, and the auction has become increasingly gamed.2
Just as prices work as allocative signals as long as costs are real, so links work as allocative signals in search rankings as long as they represent relevance.3 But the very existence of the mechanism undermines it, and does so much more than analogous ``market manipulation" does. Models of market manipulation show the cost inherent to flexing market power: buyers cut back on quantity consumed, entry is encouraged, etc. In the classic account of a English auction, it is even a dominant strategy to tell the truth about value (or cost, in a reverse auction) and be rewarded transparently for market power. The dominant strategy argument relies on precisely the self-limitation of the exercise of market power: push your luck too far, and you will end up buying something for more than you value it or not owning something you wished you had had. Scarcity continues to do its limiting work, even under market manipulation, and prices continue, therefore, to carry meaning. But the manipulation of ranking suffers no real scarcity constraint. Links can be invented at very low marginal cost, and at a cost which has no real relation to the relevance, to the genuine meaning, of the link. Search Engine Optimisation experts build ``link-farms"--web sites rented out for their potential to link to others. PageRank, predictably according to price theory, was very good at indexing the web before it became generally understood, and is now doing poorly. Google, instead, plays a complicated and obscure cat and mouse game hoping that its historical lead has enough natural monopoly in it to keep search working.
Google hoovers up the work of Stakanovite linkers and aggregates it by simple formula into tables of relevance whose quality is certainly questionable compared to the organisation of knowledge that preceded it (think of the search results from a good library compared to Google's). But, as in so many industrial processes, Google makes up in quantity what it sacrifices in quality against the old artisianal methods. More disturbingly, however, Google's business model encourages it to operate a Stasi-worthy accumulation of personal data. Google does not simply license search to people who demand search. It gives away search in exchange for the right to a small piece of users' screen real estate. It optimises the value of that real estate by knowing as much as it can about the property--what is the behaviour of its eyeballs.
Most markets are extremely decentralised.4 You can enter a market without registering with some central authority. This is a significant factor in the association of free markets with the freedom of the moderns. No centralising authority or registry is required; entry and exit are easy, etc.
But Google is different. It is much more like a formal exchange, and even then, it is not a simple bilateral match of trades. Google (in its search service) processes an input--"raw" web pages--and offers sorted web pages to the searcher. You cannot compete in the information economy without doing the equivalent of "registering" with Google: you make the web pages visible to its spiders, and usually further than that, you "sex them up" with (search engine optimisation) SEO to increase their salience to Google's algorithm. Google, in the sense of the behaviour that it elicits, has a huge degree of control over content on the Web. Entry into the information economy has a gatekeeper, registrar and rule-maker. No wonder Google needed to persuade us that it would "not do evil".
So although PageRank looks as if it is doing something market-like in its processing of information, it should probably be seen instead as the ultimate tool in the centralised processing of information. Google should be considered to be a "mechanism", in the sense of "mechanism design", and one that is at a particularly un-spontaneous, centralised end of mechanisms.5
Google's business model is to sell your screen-space to advertisers and swap you free-search in exchange. This strange barter economy is very common to Anderson's world of ``Free" (Anderson, 2007). But note that there is no fundamental reason not to split out those transactions: I could rent out my screen space through a third party, on privacy terms I could specifiy, while I could buy search services from a search provider. The hidden cost of the ``Free" lunch is, ultimately, Freedom.
Wikipedia, on the other hand, does not centralise personal information. We can think of Wikipedia in terms of costly signals: a change to a wikipedia entry is an effortful move ...it is like a mutation in genetics. Will the mutation survive? If it elicits others to overcome the cost barrier of making a counter-modification, it does not survive. Wikipedia is therefore mostly composed of ``evolutionarily stable statements".6
The evolutionary method is distributed in a way PageRank is not: on most articles, anyone can make a change, even anonymous users. Moreover, Wikipedia has no advertising revenue to optimise, and therefore has no particular interest in the systematic collection of personal information. Wikipedia is much more like the libero-genic market than Google--a background institutional framework that allows and encourages competing entry without any systematic monopolising of manipulating tendency of its own.
Although this characterisation is true of much of Wikipedia, it is not true everywhere. Some articles are locked. Here is a film about the details by my colleague Felix Cohen, showing the locking of the Israel entry.
Wikipedia: Stability, manipulation and locking.
These hard cases, of course, are an important part of the collective action of the Wikipedians. They are never more responsibly solicited than in cases of lock-down. The deliberation amongst the self-appointed elite is a great example of the freedom of the ancients that the web, at best, re-creates.
The similarities between Web2.0 services and markets: both filter information and create ordered rankings that help with choice; both take decentralised behaviour as inputs ...should not lead us to think that the most libero-genic characteristics of markets automatically flow from this. Centralisation of information is a business necessity for Google as it optimises click-through. Wikipedia is not data-phagous only because it does not have much of a business model beyond the highly Nozickian one of voluntary taxation.
Follow the money \ldots to the database
Whatever services can capture large portions of the data relating to who, what when, where and how about every habitual human activity will be sitting on an advertiser's dream. The commercial incentive to build large information collections is huge. As argued in Part 2 of the essay, these corporate databases pave the road to Kafkaesque and Orwellian tyrannies as surely as government data-phagy does.
The forces of tyranny--whether social or technocratic--seem aligned to make the most of the power of technology. Should we disconnect from the cloud? come off the grid? return ourselves, if not society, to primitive innocence?7No. The hopeful potential for the network technologies exists, even when we recognise their power to help tyranny. Once the forces that tend technology towards unfreedom are identified, we can fight them in the right ways.
- Kafka, Orwell, McNealy
- The fight here has to be almost entirely political--our civil liberties need to be fought for against naturally power-loving technocracies. When a state can be trusted, we would want it to deploy surveillance responsibly. When it cannot be trusted--or is not trusted by crucial parts of society--then it must be stopped from using the power of surveillance. Of course, the task of creating trust-worthy states is huge and never-ending. a good place to start is to take a lesson from the freedom of the ancients: collective self-determination was a reality in Athens largely because of its scale. The gigantic states of today, made possible by their efficient use of technology, are not possibly the sorts of bodies over which anyone can feel they exercise determination.
De-gigantifying the state is a huge political task, and one that can certainly be helped by technology. Here are just a few links to organisations that are using information to ``shame'' states into better behaviour. FarmSubsidy.org collects all the information from different EU member countries about who receives what payments from the Common Agricultural Policy. Making the information easily available--for example showing how much the CAP benefits Nestle, large landowners etc--should make it harder to keep acting like this. Mwali Matu, at Mars Kenya catalogues the private interests of members of the Kenyan parliament to keep a check on corrupt legislation.
These positive uses of information technology against the might of gigantic states are small. Traditional political action--the support of campaigns to protect civil liberties, for example; or movements for the reform of political institutions, like the effort to re-localise much of our politics--has to be a large part of a sensible course of action. This is what the Convention on Modern Liberty is looking to achieve in the UK . Once we have trustworthy government, we can empower it with technology. Until that point, we should resist moves to the creation of the database state.
As entrepreneurs and consumers, there is much we should and could do. This is repeated from the section above:
Google's business model is to sell your screen-space to advertisers and swap you free-search in exchange. This strange barter economy is very common to Anderson's world of ``Free" (Anderson, 2007). But note that there is no fundamental reason not to split out those transactions: I could rent out my screen space through a third party, on privacy terms I could specifiy, while I could buy search services from a search provider. The hidden cost of the ``Free" lunch is, ultimately, Freedom.In an environment where responsible citizens understand the dangers of dataphagy, there will be markets for liberty-friendly technologies.
- In the fight to preserve the libero-genic nature of the Internet, the solutions proposed by Zittrain seem right. Prefer those solutions that rely for their incentives and organisation on the freedom of the ancients. Prefer Wikia to Google. Be a good digital citizen and do the equivalent of picking up the litter in public spaces--contribute to the properly decentralised web services. Use what means there are (for example anti-trust) to level the playing field against those whose business is, at heart, the centralisation of information. Putting this into practice has implications for us as users, as entrepreneurs and for public policy.
- The Sunstein effect seems like a minor worry compared to the previous 2. Although communities may splinter online, the digital world offers opportunities for multiple overlapping identities. I can be ``admin" in the OpenDemocracy forums and ``bigtone" on kiteforum.com; TCP on twitter, and I can choose what people see of my activity on FriendFeed. My colleague David Hayes and his co-author Keith Kahn-Harris argue the opposite point in their very interesting "The politics of ME, ME, ME". They deplore the micro-fragmentation of politics and communities that the web has enabled, and they argue that this undermines genuine efforts of collective self-determination. The commentary on their article, somewhat paradoxically, advances the argument.
Walzer (1983) argued that the USA more than anywhere else practised an equality between overlapping spheres of justice. A fireman might not be a billionaire, but was certainly a valued firefighter ...For Walzer, the multiplicity of spheres gave rise to a ``complex equality" wherein lay the strength of American society. Online, the idea can be extended. Sunstein's narrowing of visions should be countered by the proliferation of identities. The Health Ranger is not only that ...somewhere else, he will be asking for technical help on how to get rid of a (computer) virus ...the beliefs, behaviours and people whom he finds in that role will have an effect on all his other beliefs. as digital citizens, we should always think of who we might be interacting with, of how this particular interaction might break through a Sunstein-barrier. The web is likely to re-inforce the trend that other forces of globalisation contribute to of multiplying our identities (Sassen, 2008).8
A call to action
Here is a final summary in terms of the freedom of the ancients and the moderns . The unfreedom of the moderns stems from the dangers of agency, dangers exacerbated by gigantism, itself permitted by technology. The unfreedom of the ancients comes from the tyranny of society, the absence of any notion of the rights of the individual against the group. If techno-society can combine that unfreedom with gigantism--maybe the rise of online nationalism in China is just that sort of a move--then freedom lovers must work to stop it. Technology empowers, but power is the raw material of both freedom and tyranny.
Anderson, C.: 2007, The emerging world of ``free'', Video.
Curzon Price, T.: 2008, Google's attention deficit disorder, openDemocracy .
Sassen, S.: 2008, Fear and strange arithmetics: when powerful states confront
powerless immigrants, openDemocracy .
Walzer, M.: 1983, Spheres of Justice, Basic Books.
- ... 2)1
- Many thanks to all the people who have commented on early drafts of this paper--Selina O'Grady, Graeme Mitchison, Victoria Curzon Price, Anthony Barnett, Jonathan Zittrain, David Hayes, Jeremy O'Grady, Stefaan Verhultz. This paper owes a great deal to a seminar funded by the MacArthur foundation in March 2008, "Credibility in the New News" in London. Many thanks to Kathy Im and Elspeth Revere for making that gathering and space for thinking possible. I presented a version of this paper to the Annual Meeting of the Mont Pellerin Society in Tokyo in September 2009. It was a personally emotionally charged occasion, being the child of two members of the society while feeling uncomfortable with most of the positions taken by its members. On the question of the authoritarian state, however, we were on common ground - at least at some level of abstraction.
- There is a second aspect to Google's operation which is less innovative but logistically very important. It needs not only to judge the ``worthiness" of a page through PageRank, but also needs to come up with a list of relevant pages to return for each query. This is done through statistical text-processing heuristics.
- One of the most troubling features of the subprime crisis has been that one of the most important prices in the economy--the price of risk--was found to have been set through processes that would almost surely vitiate against it being a reflection of real costs. This was the lesson of the layer upon layer of agency and regulatory failures. Getting the price of risk wrong has huge repercussions in the world of goods; getting the price of attention wrong--as is happening with Google today--has a similarly leveraged effect in the world of bits.
- Not exchange traded markets, but all others.
- Just like the mechanisms that were meant to produce "good allocations" in utility regulation, PageRank cannot of itself undo any deep forces towards uncompetitive and manipulative behaviour.
- The evolutionary process can, fascinatingly, be observed on Wikiscanner, the very brilliant piece of software built on top of Wikipedia by Virgil Griffith, a student at Caltech, lets you find out what anonymous Wikipedia edits have been made by which organisations. The subtle change in the Walmart entry, from a Walmart computer, describing its average wage rate not as ``20% less than other retail stores", but rather ``double the federal minimum wage," shows the kind of micro-mutation that lies behind the Wikipedia process.(Although Wikipedia does not centralise information, a trace of the identity of who has made the change is kept, allowing for a degree of control, for example to exclude consistently unhelpful contributions. Wikiscanner has taken advantage of that information to produce its information sleuthing service.)
- There is certainly a conservative current that has come to this sort of conclusion. Here we have the French conservative, Alain Finkielkraut and a commentary on his "worrying ecstasy".
Mark Pesce has recently written
about a "hyper-Sunstein" effect.
Fasten your seatbelts and prepare for a rapid descent into the Bellum omnia contra omnes, Thomas Hobbes' "war of all against all." A hyperconnected polityÑwhether composed of a hundred individuals or a hundred thousandÑhas resources at its disposal which exponentially amplify its capabilities. Hyperconnectivity begets hypermimesis begets hyperempowerment. After the arms race comes the war ...
Naturally, governments will seek to control and mediate these emerging conflicts. This will only result in the guns being trained upon them. The power redistributions of the 21st century have dealt representative democracies out. Representative democracies are a poor fit to the challenges ahead, and 'rebooting' them is not enough. The future looks nothing like democracy, because democracy, which sought to empower the individual, is being obsolesced by a social order which hyperempowers him.Apocalyptic techno-visionaries are also always with us. The Sunstein effect is a slow and weak force, too little to be hanging this sort of apocalypse on.