Google's decision to combine the behavioural data of users of its different products has given rise to concerns about the encroachment of big business into the private lives of ordinary citizens. There are two complementary perspectives normally deployed when discussing these concerns.
The concern about privacy
Some of this is just cynical pandering to fashionable paranoia by lazy journalists, but there are sincere civil libertarians who regard Google's action as a worrying step towards an infrastructure of mass surveillance, if not The Matrix.
The first point to note is that many of the commentators who have articulated this view don't understand how the technology works. This does not invalidate their concerns, but it does explain why they tend to see the situation through the prism of traditional surveillance tropes.
The old East Germany became the leading example of surveillance following the opening of the Stasi's files in the 1990s. This reached its apotheosis in the film The Lives of Others in 2006, in which the agent of the state is inexorably drawn into the lives of his targets. This is a high quality (i.e. middle class liberal) model of surveillance, giving substance to the lives of both the snooper and the snoopee. It allows for small triumphs of humanity and free will, and acts of theatrical self-sacrifice.
The reality of the Stasi was institutional incompetence. In part, this was down to its genuinely totalitarian ambitions. It is estimated that files were kept on a third of the population and that as many as 10% were occasional informers. This meant that the Stasi was inundated with huge amounts of worthless noise. The act of gathering so much, and the co-option of so many, gave the impression of omniscience, which in turn produced self-repression. If you were an informer, even one who produced nothing but banal chaff, you would assume that you too were under surveillance and would keep your nose clean accordingly.
Google are running a business in order to make money, not trying to create a totalitarian state. It suits their purpose to imply omniscience, as that is what attracts the fee-paying advertisers, but the reality is not unlike the Stasi: humongous amounts of worthless data.
Making money is a numbers game, which means a business is generally discriminating about its customers. In particular it makes an effort to exclude those who are not profitable. This is the opposite of the totalitarian state, which tends to spend a disproportionate amount of its effort on the recalcitrant minority. It's a good example of the Pareto principle: business focuses on the 20% that provide 80% of the profit, and avoids the 20% who generate 80% of the costs; the totalitarian state focuses 80% of its public order effort on identifying and controlling the 20% who constitute a perceived threat. A totalitarian state does not offer an opt-out, even one buried in the small print.
The libertarian critique of Google and others usually fails to take into account that the business is simply not interested in those who are most concerned about privacy, as these self-selecting individuals are the ones that provide the least value to the business. The savvy user who limits their online trace is a much less substantial, and thus less monetisable, user than the average. They are probably also more discriminating in their consumption, and thus less open to the persuasion of advertisers, with the exception perhaps of those that appeal to their savviness or paranoia. The intelligentsia is a niche market, and a not particularly valuable one at that.
The concern about surplus labour
The economic perspective is often expressed in the maxim: "the product is you". Your online activity creates a commodity, your aggregate preferences, through explicit "likes" and behavioural inferences. This surplus value of your labour is taken by the likes of Google and Facebook and sold to advertisers. The capital of these companies has been built by millions of users creating an enormous, multi-dimensional dataset as a by-product of their online lives.
Most users understand the quid pro quo nature of free online services, namely the price you pay is advertising. This is familiar from the world of broadcast TV. The more tech-savvy also understand that this advertising aspires to narrowcasting, i.e. it is targeted to you and your assumed interests. This is sometimes justified as a better service for the user, in the sense that it presents ads that are more relevant, however the only real justification is that it is more valuable to advertisers, as relevance is assumed to lead to a better conversion rate (views to purchases).
At a macroeconomic level this is a wash. Increased relevance does not lead to increased aggregate expenditure. At a microeconomic level it is crucial, as it gives one medium (online) a potential USP over other media (TV, paper, billboards etc). It should be noted that this USP remains good even if only a minority of users provide sufficient preference data to support targeted adverts. Accurate and detailed data for 20% of users would be enough to allow online to dominate the ad budgets of most advertisers.
This last point highlights the problem of equating the traditional notion of surplus labour value with the expropriation of user value online. No one is obliged to go online so the real coercion that underpins economic life is not present. You will not be impoverished if you decline. Though it will become more and more difficult to avoid online altogether, as the medium becomes pervasive across many areas of social exchange, the extent to which you are exposed can be controlled by you. Most importantly, the amount of surplus labour value you produce, as a percentage of your online activity, can be minimised. In the traditional factory setting, this can usually only be done through sabotage.
Google's move provides an interesting counterpoint to Marx's theory of surplus labour value and the related tendency for the rate of profit to fall over time. Their investment in constant capital (their software products and server infrastructure) can only lead to a productivity gain if it increases the surplus value of the users' labour, as that is the product they sell to advertisers. However, they don't need to increase the surplus value of all labour, only some of it, i.e. those who don't object to their revised policy.
Meanwhile, the "saboteurs" who minimise their online data, and thus their surplus labour value, are largely ignored by Google. They are a vocal minority, but they constitute neither a significant threat (so long as Google remains law-abiding) nor an opportunity. They are actually more like factory inspectors than browbeaten proletarians.