Lessons from a libertarian

Earlier today, Aaron Ross Powell of Libertarianism.org wrote a post reflecting on the lesson that techies should have taken away from SOPA. No one individual has enough expertise, knowledge, or time to fully comprehend the breadth and complexity of all legislation. This ignorance, Powell argues,

forces us to either accept at face value what our legislators tell us or to adopt a general attitude of skepticism. The latter leads in the direction of libertarianism. The former leads to bigger, more damaging government—and government more and more in the pockets of those very interest groups we hope it would instead work against.

Of course skepticism is good, while complacency is bad. But skepticism about the impact or integrity of a legislation does not equate to Powell’s skepticism that “the state, if only given more power, can fix whatever ails us.” After all, a supporter of universal health care can (and should) err on the side of caution when evaluating a universal health care bill. The problem with Powell’s breed of skepticism is that it dismisses the possibility of thoughtful evaluation by non-experts:

Even if we turn to the experts to explain to us why we should or shouldn’t support a given policy, those experts rarely agree and often have competing interests. After all, SOPA’s proponents were able to present “expert” arguments for their positions, too.

Powell argues that our default attitude should be one of mistrust, since the proponents of a bill can always confound the general public with misleading or downright false “expert” testimony. Mistrust of large, sprawling bills is a good thing– but the logic behind Powell’s advice is problematic, for it excuses the individual from thoroughly investigating a given bill: after all, whatever opinions she finds will inevitably be biased, in ways that she cannot comprehend. Since she doesn’t understand it, she had better vote it down.

Rather than adhere to the mantra that inaction is safer than action, we should form opinions by striving to understand legislation to the best of our abilities. This means that just as we refuse to blithely trust politicians, we should refuse to blithely trust “expert” opinions. Just as we inspect the motivations and connections of politicians– for example, which corporations provide financial backing for their campaigns– we should weigh testimony given by an expert against his incentives. The SOPA battle isn’t necessarily an argument for smaller government; it’s an argument for a government that is responsible to its people and vice versa.

 

 

 

The Net Delusion

“The premise of this book is thus very simple: To salvage the Internet’s promise to aid the fight against authoritarianism.” –Evgeny Morozov, The Net Delusion

In The Net Delusion, Evgeny Morozov argues that the fetishization of information technology undermines the promotion of democracy. Morozov’s central contention is that American policymakers have proved dangerously susceptible to the “Net delusion”— the naïve assumption that the internet is bound to be an agent for positive change in authoritarian regimes (“cyber-utopianism”), coupled with the tendency to craft policies around the logic of technology without regard to political or social context (“internet-centrism”).

Morozov’s most effective main points are as follows:

  1. The intellectual legacy of the Cold War has rendered policymakers overly-exuberant when it comes to the democratizing potential of the internet; furthermore, viewing the “war for internet freedom” through the lens of the Cold War obscures and misleads rather than informs.
  2. Authoritarian regimes have adapted to the cyber-landscape and are now poised to strengthen their apparatuses of censorship, surveillance, and propaganda to unprecedented degrees. The internet’s empowerment of oppressive regimes may, in the long run, outweigh what good it brings the oppressed.
  3. Unfettered access to information may stifle dissidence or activism rather than foster it; glorification of the internet facilitates turning a blind eye to the very real harms it may engender.

The fallacy of the Cold War metaphor. In America, the language of internet freedom as a foreign policy goal is often heavily couched in Cold War metaphors. To this, Morozov raises several objections (quibbles), ranging from the general (the pitfalls of metaphorical reasoning in general) to the specific (drawing parallels between blogging and samizdat is a dangerous oversimplification).

The heart of his critique, though, is essentially that information-centric interpretations of Soviet collapse have got it wrong. The idea that information and communication technologies are toxic to authoritarian governments derives from triumphalist readings of the Cold War, argues Morozov. In the wake of the dissolution of the Soviet Union, the notion that information had been central to the state’s collapse quickly gained traction in the political sphere. Statesmen and journalists were happy to wax poetic about the role of samizdat, Radio Free Europe, and smuggled photocopiers—all of which had been subsidized by the United States government—while paying scant attention to the structural causes of the Soviet Union’s breakup. Information technology had certainly served as a medium for the virus of dissent, had certainly helped unify and organize protestors. But, Morozov argues, widespread dissidence as enabled by the flow of information was not what brought the Soviet Union to its knees; nourishment of resistance via communication was possible only because the state had already been crippled in a fundamental way.

“These may all be worthwhile policy recommendations, but they rest on a highly original—some historians might say suspicious—interpretation of the Cold War.”

The takeaway: the internet may aid democratization, but policymakers cannot hope to make gains for freedom with technology alone. Hillary Clinton can feel free to trumpet the importance of internet freedom, but the cause of the internet must not displace other policies for human rights protection.

Technology is Big Brother’s friend. 

“A new mode of obtaining power of mind over mind, in a quantity hitherto without example.” Jeremy Bentham on the Panopticon

Propaganda, surveillance, and censorship on steroids: the internet amplifies both the dominion of the state and the ability of the determined individual to evade it. What is the net effect? Morozov seems to believe that the internet favors government more than it helps dissident movements (of course, he is a bit of a pessimist, having once written that “David Sasaki […] recently challenged me to write at least one positive post or essay for every ten that are negative. I can’t yet live up to that challenge but I am eager to try bit by bit.”) Some interesting examples from Morozov below:

On the propaganda front, governments are realizing that the internet is well-suited for ideological indoctrination, indoctrination that may be all the more effective for its subtlety. Take the persistence of the death panel myth in 2009 and the controversy over Obama’s birth certificate in 2008. Even in the absence of an established propaganda office, these easily debunked claims were often believed to be true. The idea that access to information precludes propaganda, then, is but a fantasy. Access to a resource does not imply usage, and the internet is no exception. Moreover, these examples corroborate Morozov’s argument that

“small doses of propaganda are still bad for you… All the government has to do is to ‘seed’ a pro-government movement at some early stage—inject it with the right ideology and talking points and perhaps some money—and quietly withdraw into the background.”

Surveillance is also made infinitely cheaper and easier by technology. (For some reason, Morozov spends some time arguing that surveillance in the digital age is a sharpened threat because it loses the dimension of human sympathy present in analog surveillance. Luckily, this is not his sole point.) Digital surveillance is dangerous because it can be easily exported to swathes of people: for example, as Morozov documents, through poisoned download links on trustworthy sites. Morozov also makes the chilling observation that

Search engines have inadvertently become extremely powerful players in the business of gathering intelligence and predicting the future[...] Google does know how often Russian internet users search for the words “bribes,” “opposition,” and “corruption”; it even knows how such queries are distributed geographically and what else such potential troublemakers are searching for.

If internet users continue to shift from Western corporations to local corporations, as they have done in China and in Russia, this information may one day be readily available to the state.

And finally, censorship. The algorithms used in targeted advertising and searching, in which ads and search results are tailored to a user’s profile, can also remove results based on a user’s profile—targeted censorship. China’s predictive censorship engine GreenDam was to be pre-installed on every computer sold in the country. Though China has put GreenDam on the back burner for now, GreenDam 2.0 will almost certainly come to market someday. When it does, companies such as Acer, Sony, and Lenovo will likely have no qualms about shipping their PC’s to China with censorship software preinstalled.

The takeaway: It is possible to circumvent government control to some extent, but even then, users of technologies such as Tor do not fully understand the limits or security vulnerabilities involved. The internet provides the state with flexible, sophisticated, and more subtle tools of manipulation and scrutiny. Moreover, as long as the internet and mobile networks are tied to physical infrastructure, states have the option to off all communication capabilities in a certain area if a strong enough threat is perceived.

The internet is not always good for you. Morozov has two main points here: the internet is the opiate of the masses, and the internet may displace more effective forms of activism.

Mass media is soma, the drug of the hedonist. It caters to people’s desire for cheap, agreeable entertainment at the expense of intelligent discourse. Postman’s 1980’s critique of television extends easily to much of the internet: though the form of the internet does not inherently exclude rational inquiry, most people, most of the time, choose to experience the internet as “entertainment programming.” The freer flow of information also means the freer flow of amusement and distraction: this is why Russia censors very little but provides high-quality, state-run entertainment sites; and it is why after the 1960’s, East German officials chose not to block Western broadcasting in the GDR. This is not to say that we should condemn the internet—simply that we should neither glorify those whose internet freedoms we are fighting for, nor should we expect that the provision of internet freedom will automatically politicize a people.

Morozov also argues that the low cost of Internet actions, such as joining Facebook groups or signing online petitions, encourages slacktivism. This is a perfectly valid critique, especially, I think, in countries with high standards of living, such as the United States. But I suspect that slacktivism does not significantly displace real activism, as those who do naught beyond joining Facebook groups are unlikely to have agitated for change offline in the first place. On the other hand, Morozov claims to detect a “certain triumphalism about online politics,” one that leads Belarusians to the internet in search of “solace from the dreadful political realities of the real world.” I’m unable to speak for the state of Belarusian politics, but it seems that in general, next to the benefits represented by greater connectivity and faster communication (think Arab Spring), slacktivism casts but a short shadow.

Slacktivism aside, a focus on internet activism is still problematic if it comes at the cost of “more effective” organization. Morozov likes to draw parallels, for example, between Ringelmann’s experiments and Twitter activism, but in some ways he seems as Manichean as his unfortunate cyber-utopians. Morozov is dismissive of internet activism, but the efficacy of an internet campaign needs to be evaluated on an individual basis. I wonder what Morozov would have made of the Occupy movement in its inception—a protest that swelled thanks to the ease of internet mobilization, a protest that even prided itself on its lack of hierarchy and control. The story of Occupy shows that internet-organized, decentralized dissidence does not quite fit into Morozov’s mold: it is a new beast, a form of protest that achieves a significant and unique change despite its break with traditional models of “successful” or “effective” dissidence.

Takeaway: We need to internalize that the Internet is not a cure-all. Morozov’s is a sobering message, a call for policymakers to proceed with more discretion, sophistication, and realism in technological matters.

My main dissatisfaction with The Net Delusion is that Morozov barely questions US domestic policy at all. He admits that the US has ceded its moral high ground over Iran and China when it comes to internet regulation, and he calls out the US for the inconsistencies between foreign policy (internet freedom) and domestic policy (internet regulation). Yet Morozov seems to accept that internet regulation is inevitable—in some cases acceptable or even welcome. He spends an entire chapter documenting “internet freedoms and their [Big, Bad] consequences.” The internet can be used by nationalists, religious fanatics, gangs, and white supremacists to further hate, violence, divisions, indoctrination. So regulation, Morozov argues, is an attractive option. As long as hatred and ill-will exists (so for the foreseeable future), the internet needs to be tamed.

This is where Morozov falls into a trap of his own devising. Throughout The Net Delusion, Morozov lambastes policymakers for attempting to solve social or political problems with purely technological fixes:

“We avoid the search for a more effective nontechnological solution that, while being more expensive (politically or financially) in the short-term, could end the problem once and for all. We should resist this temptation to fix technology’s excesses by applying even more technology to them. […] As it is irresponsible to prescribe cough medicine for someone who has cancer, so it is to apply more technology to social and political problems that are not technological in nature.”

Yet Morozov’s examples of justifiable regulation are exactly technological fixes of problems that are not technological in nature. Concerned parents wanting to shelter their children from inappropriate? Even the Tunisian Internet Agency (which functioned in part as a censorship agency during Ben Ali’s reign) has acknowledged that endpoint-based filters (for example, a filter that may be turned off with a password), rather than regulation of the network, is the solution. Ethnic minorities seeking protection? This is the jurisdiction of laws regarding hate speech in general. And what about SOPA/PIPA (which, somewhat curiously, Morozov never addresses on his blog)? Again, regulation is not the answer; it’s ineffective and far too susceptible to abuse, while innovations in the entertainment industry itself would address piracy in a much better way.

Clearly, internet freedom translates into the freedom to do ill—but hatred and violence are not fundamentally technological in nature. Morozov’s failure to apply his own critique to domestic internet policy makes him dangerously complacent with regards to internet regulation. Regulation at home makes it all the more difficult to promote freedom abroad, as Morozov recognizes. His argument is that policymakers would “constantly search for highly sensitive points of interaction between the two.” It’s not clear what that means, but Morozov is forcing the situation to be more complicated than it needs to be. How about both freedom abroad and freedom at home?

Last note: An excellent piece on the hypocrisy of the United States when it comes to internet freedom. Perhaps Morozov would excuse these contradictions. In doing so, he would forget that freedom is far from guaranteed in the United States itself.

An introduction to Bell inequalities

In classical information theory, a bit represents a single unit of information: 0 or 1, on or off, true or false. In quantum information theory, the analogue of the bit is the qubit: a quantum state which can live anywhere in the Hilbert space spanned by |0\rangle and |1\rangle. Clearly, a qubit |\psi\rangle encodes at least as much information as does a classical bit, and we might feel intuitively that a qubit can encode quite a bit more than that. (Hoho.) But what exactly could one do with a qubit of the form \alpha |0\rangle + \beta |1\rangle? Measurement will collapse the qubit into a single state, and we cannot even begin to guess at the values of \alpha and \beta without having many copies of the same state. We shall have to be more subtle than that if we want to do better. This is what quantum computation is all about: exploiting the hidden richness of quantum bits for information processing tasks.

Superdense coding illustrates how two bits may be killed with a single qubit. Suppose Alice and Bob agree to communicate via qubits. Alice encodes some classical information in a qubit and sends it to Bob, who recovers the information by measuring the qubit. At first glance, it’s not obvious that Alice and Bob cannot do any better than in the classical case; non-orthogonal quantum states are not perfectly distinguishable, so if Bob wishes to recover two bits of information via measurement, he needs two qubits. However, Alice can send Bob only one qubit!

Of course, nowhere does it say that Bob can’t start with a qubit as well. We could give Alice qubit a and Bob qubit b; Alice could encode information in a and then send it to Bob. Bob’s two qubits would then be living in a four-dimensional Hilbert space, in which a measurement would yield two bits of classical information. Qubit b would only be useful, however, if it were correlated in some way to a. Without some kind of relationship between the behavior of the two qubits, Alice wouldn’t be able to control the result of the measurement made by Bob. Bob would certainly have two bits of classical information at the end of the day, but Alice would only have sent him one of those.

Here’s where quantum mechanics shows up. In classical information theory, bits a and b are independent. But quantum mechanics permits entanglement between qubits a and b. That is, we can have

    \[|ab\rangle = \frac{|00\rangle + |11\rangle}{\sqrt{2}}\]

In such a state, measurements on a and b are perfectly correlated: if Alice gets 0 when she measures her qubit, Bob will also measure 0, and if she gets 1, Bob will get also get 1. Entanglement of qubits a and b will allow Alice and Bob to design a scheme where Alice can communicate two classical bits by performing a local operation on her qubit before sending it to Bob.

In such a scheme, the space of local operations Alice can perform on a is spanned by \sigma_0, \sigma_x, \sigma_y, \sigma_z. (By \sigma_0, I mean the identity operator.) The four dimensions of the operator space correspond exactly to the four possible messages– 00, 01, 10, 11– Alice will communicate to Bob.

  • If Alice wants to send 00, she applies \sigma_0 to a. (In other words, she does nothing.) This corresponds to applying \sigma_0 \otimes \sigma_0 to |ab\rangle.
  • If Alice wants to send 01, she applies \sigma_z to a. We have

        \[\sigma_z\otimes \sigma_0 |ab\rangle = \frac{|00\rangle - |11\rangle}{\sqrt{2}}\]

  • If Alice wants to send 10, she applies \sigma_x to a. We have

        \[\sigma_x\otimes \sigma_0 |ab\rangle = \frac{|10\rangle + |01\rangle}{\sqrt{2}}\]

  • If Alice wants to send 11, she applies i\sigma_y to a. We have

        \[i\sigma_y\otimes\sigma_0 |ab\rangle = \frac{|01\rangle - |10\rangle}{\sqrt{2}}\]

The four resulting states are an orthonormal basis, so Bob can distinguish them by making appropriate measurements. Entanglement allows Alice to transmit two classical bits by sending Bob a single qubit!

As we have seen in the example of superdense coding, entanglement is a precious resource for quantum information; it is, as Nielsen and Chuang aptly remark, “iron to the classical world’s bronze age.” Entanglement looms large in quantum information and computation because it breaks the bounds set by classical information theory. In the larger context of physics, though, entanglement has been much more than a resource; its history is the tale of a revolution, the vanguard of a paradigm shift in the realm of physical reasoning.

Einstein, Podolsky, and Rosen were the first to realize that the phenomenon of entanglement issued a fundamental challenge to locality, the principle that effects of physical processes at one location cannot be transmitted instantaneously to other locations. Locality seemed to follow from relativity, which had convincingly established that information could not propagate faster than c without violating causality. But with entangled states like the one shared by Alice and Bob in the superdense coding problem, Alice and Bob could perform measurements in a causally disconnected way while getting perfectly correlated measurement results. Entanglement, EPR argued in 1935, showed that quantum mechanics was an incomplete description of “physical reality.”

Fundamentally, EPR objected to the quantum mechanical idea that properties of a system were related to the act of measuring that property. EPR favored realism via the hidden variable theory– the idea that there existed a complete theory with variables corresponding to all “elements of reality”– and argued that quantum mechanics was only a shadow, a projection, of this theory. Once supplied with the complete theory’s hidden variables, quantum mechanics would once again recover locality. Though the physics community viewed the EPR paper as an arbitrary imposition of a classical worldview, it was not obvious how to put the critique to rest. The behavior of any apparently random quantum mechanical process could, of course, always be attributed to the deterministic output of a broader theory.

In a seminal paper published in 1964, John Bell proved that it was impossible to construct a locally realistic theory that could reproduce the predictions of quantum mechanics. That is, any theory that reduces to quantum mechanics while preserving realism (the notion that properties are independent of measurement) must be non-local. EPR’s classical worldview could now be empirically tested: experiments could be conducted in the regime where local realism broke with quantum mechanics. One or the other would have to go.

Bell’s theorem was remarkable in that it allowed physicists to outrule entire classes of physical theories, simply by making measurements on particles that interact with one another and are then separated. Bell’s original paper specified an inequality that correlations in locally realistic theories necessarily had to obey, and after 1964, physicists came up with a slew of such inequalities.

The CHSH inequality, derived by Clauser and Horne in 1974, is one example of a Bell inequality. Suppose that locality and realism hold. We ask Carl to prepare two particles in a replicable way, sending one to Alice and one to Bob. Alice’s particle has objective properties M and N which are “elements of reality,” as realism would demand; Bob’s particle has objective properties S and T. The measurement of any of these properties Q, R, S, T yields \pm 1. Alice and Bob randomly measure either Q or R, S or T respectively; their measurements are causally disconnected, which by the assumption of locality, implies that Alice’s measurement will not affect Bob’s. By repeating this experiment many times, we obtain an estimate of the quantity (Q+R)S + (R-Q)T. It is not hard to show that the expected value of this quantity is bounded absolutely by 2. (For a derivation, see page 115 in Nielsen and Chuang.) This is the CHSH inequality:

(1)   \begin{equation*} -2 \leq E(QS) + E(RS) + E(RT) - E(QT) \leq 2 \end{equation*}

So a violation of inequality 1 would contradict our assumptions of locality and realism, but could still be consistent with quantum mechanics, wherein the expression above achieves a value of \pm 2\sqrt{2} when Carl sends Alice and Bob entangled qubits. Experiments have overwhelmingly seen dramatic violations of the CHSH inequality that are consistent with quantum mechanics! And other Bell tests agree: local realism is broken. The validity of these Bell tests, of course, must be weighed against the existence of significant detection and locality loopholes in the experiments themselves (for example, low detection efficiency or incomplete causal separation of measurement sites.) Modern experiments are starting to close these loopholes, but in no way are they infallible; physicists from Norway and Singapore showed in 2011 that they could “falsify” Bell test violations with classical light by exploiting loopholes.

Loopholes notwithstanding, violations of Bell tests have convinced most physicists that local realism is wrong. The majority of the physics community believes in discarding realism, which means that the locality of quantum mechanics can be preserved. (It turns out that quantum entanglement never violates causality because entanglement cannot be used to transmit information faster than the speed of light; and once we accept that no independent physical reality exists, entanglement can be reconciled with locality.) Alternative models also exist, such as non-local hidden variable theories.

We began this discussion of Bell inequalities by recognizing that quantum entanglement plays a key role in computational tasks that are impossible to perform in the classical world. We then saw how the “weirdness” of entanglement inspired a critique of quantum mechanics that culminated in its own annihilation, in the downfall of local realism rather than its confirmation. Historically speaking, entanglement is powerfully linked with violations of Bell inequalities, but the two are certainly not synonymous. There exist non-classical quantum states which do not violate Bell inequalities, and entanglement can be “distilled” in the sense that local operations and classical communication (LOCC) can transform separable states into entangled states with high fidelity. Nor is maximal violation of a Bell inequality preserved under LOCC. Bell violations do capture some information about entanglement, but they are not good quantitative measures of entanglement.

Still, that Bell violations do indicate some degree of entanglement is sufficient for some purposes. We will see that Bell violation can be used to certify the randomness of output generated by untrusted quantum random number generators.