Countries around the globe have been threatening Research in Motion (RIM) for months now, publicly stating that they would ban BlackBerry services if RIM refuses to provide decryption keys to various governments. The tech press has generally focused on ‘governments just don’t get how encryption works’ rather than ‘this is how BlackBerry security works, and how government demands affect consumers and businesses alike.’ This post is an effort to more completely respond to the second focus in something approximating comprehensive detail.

I begin by writing openly and (hopefully!) clearly about the nature and deficiencies of BlackBerry security and RIM’s rhetoric around consumer security in particular. After sketching how the BlackBerry ecosystem secures communications data, I pivot to identify many of the countries demanding greater access to BlackBerry-linked data communications. Finally, I suggest RIM might overcome these kinds of governmental demands by transitioning from a 20th to 21st century information company. The BlackBerry server infrastructure, combined with the vertical integration of the rest of their product lines, limits RIM to being a ‘places’ company. I suggest that shifting to a 21st century ‘spaces’ company might limit RIM’s exposure to presently ‘enjoyed’ governmental excesses by forcing governments to rearticulate notions of sovereignty in the face of networked governance.

Before I get any further, I need to add a pair of caveats. First: I don’t professionally manage BlackBerry devices or presently administrate an Enterprise Server. I do, however, have a decent high-level understanding of how the BlackBerry ecosystem is set up and how it functions as a cohesive system. Second, and perhaps more importantly, while I have family and friends who work at RIM absolutely nothing written here has been taken from conversations with them nor ‘cleared’ or edited by them. Everything written here has been taken exclusively from the following sources: conference presentations given by RIM Security, my own personal familiarity with the BlackBerry product lines, discussions with information technology staff who deploy BlackBerry products, academics who attend to mobile security issues, and personally performed online research. I have not communicated with anyone inside RIM – that I know personally or otherwise – about the specifics of what I have written here.

The Origins of the Blackberry

The first Blackberry was a glorified Pager that was released in 1999. RIM’s innovation was to combine corporate and wireless mailboxes by setting up a service that relied on their own Network Operations Center  and Blackberry Enterprise Software. Together, this infrastructure collected pager messages into a single mailbox and then pushed them to Blackberry devices. Messages were encrypted using triple DES encryption, with encryption keys supplied by the enterprise instead of RIM. This separation of data transit responsibilities and key provision meant that RIM could not decrypt messages while they rested on RIM’s servers. This basic infrastructure, now ingrained in the BlackBerry Enterprise Server (BES), is a lasting legacy that provides the real security that enterprise customers have come to expect from RIM products. In the absence of this infrastructure (as in the case of consumer BlackBerry use) the security of BlackBerry communications remains largely rhetorical; if governments strongly pressure RIM and wireless companies for consumer data they can usually (eventually) force RIM to turn over demanded information.

BlackBerry Internet Service

RIM’s BlackBerry Internet Service (BIS) is central to the provision of consumer BlackBerry offerings. Assuming that a customer purchases their device from a wireless phone carrier and use that carrier’s services exclusively (i.e. assuming that their phone isn’t hooked up to a BlackBerry Enterprise Server) then the BIS lets customers enjoy many of the corporate features of the BlackBerry without any of the security that is often associated with the BlackBerry. The following image displays the general structure of the BIS ecosystem:

There are several benefits to using a BIS, including access to email, data compression and BlackBerry Messenger. In the case of email, it is as secured as the wireless provider makes it. This means that customers would enjoy levels of security equivalent to or exceeding that of enterprise customers if the provider deployed an asynchronous key infrastructure designed to prevent the provider from accessing their customers’ email in transit and at rest. Unfortunately, I’ve yet to find a single wireless network that provides this level of encryption for the transit of email. Instead, while the communication between the mobile device and the wireless networks’ server is likely encrypted – if using a GSM-based device, the A5 algorithm protects the customer’s data over the air – the rest of the data’s transit is likely unencrypted.[1] Since carriers are often obligated by national law to design networks to facilitate lawful access (e.g. CALEA in the US) government can gain access to carrier-mediated data communications. RIM is (somewhat) explicit about this in their BIS “Security Feature Overview” .pdf document, where they write that “Email messages and instant messages that are sent between the BlackBerry Internet Service and your BlackBerry device use the security features of the wireless network.” Effectively, consumers are prisoners to their wireless providers’ (often quite low) security standards.

Internet access is similarly passed through both the wireless provider’s networks and the BIS. Data is secured for the air using A5 in the case of GSM devices and then passed through the carrier’s servers and RIM’s own network. Where data is encrypted using SSL or some other form of encryption the data experiences two layers of encryption: it is encrypted over the air, and further encrypted using the web-based encryption standards. When the data passes through RIM’s servers it experiences data compression to reduce delays in accessing content. Compression is oftentimes significant; according to Rogers (.pdf) the same email message would be roughly 23KB if read on an iPhone 3G as compared to around 2KB when read on a BlackBerry Bold. Assuming both devices use similar 500MB data buckets this would mean that the iPhone could receive around 22,000 messages before exceeding the bandwidth allotment versus over 250,000 received on the BlackBerry. Significant compression is also noticed when browsing websites and sending/receiving pictures on a BlackBerry versus other mobile devices.

The third ‘key component’ of the consumer BlackBerry experience is the BlackBerry messaging service. Incredibly popular, this service is encrypted using a global key. This means that messages sent from a BlackBerry device are encrypted on the device, transmitted to the other device(s) the message is intended for, and decrypted upon arrival at recipient devices. Specifically, RIM has written that:

The BlackBerry device scrambles PIN messages using the PIN encryption key. By default, each BlackBerry device uses a global PIN encryption key, which allows the BlackBerry device to decrypt every PIN message that the BlackBerry device receives.

It is possible for RIM to decrypt messages that are encrypted with the global key, making them available to third parties if those parties come looking for them. As we will read shortly, RIM has capitulated to various governments by giving up keys enabling decryption of consumer BlackBerry messenger traffic. Importantly, the wireless provider cannot make this information available because they never have access to the global keys – your PIN to PIN messages are secure from your carrier’s surveillance mechanisms but vulnerable to RIM’s own actions.

What is the BlackBerry Enterprise Server

The BlackBerry Enterprise Server (BES) is typically deployed by corporations to secure their communications from public and private scrutiny. Below is a graphic demonstrating the BlackBerry communications architecture that includes a BES.

In this framework, communications are encrypted on the device according to the key management system used by the BES-owning organization. By encrypting communications before to exiting the device, intercepting the data at the wireless network is useless unless engaging in traffic analysis. When the data is passed into the Internet more generally it remains encrypted. The data is only decrypted when it gets behind the corporate firewall. Separate policies will manage encryption at-rest in the internal mail and messaging infrastructure that the organization maintains.

The result of this encryption policy is that email is not subject to access by government at the carrier level; government has to go to the group running the BES and demand the group hand over the data in question. This significantly changes the dynamics of the data request because carriers generally don’t care about the actual privacy of individuals on their network; so long as law enforcement is willing to pay for the effort of collecting and providing customer data the carrier is (generally) happy to help. This attitude changes when authorities come to a particular business or small group of users that are securing their communications using a BES; these groups are motivated to secure their communications (as demonstrated by setting up and running a BES in the first place) and have personal stakes in maintaining communicative security. As a result they are likely less happy than a carrier to cooperate with government agents.

What is key here, is that when running a BES neither RIM nor the wireless carrier can assist law enforcement in accessing email, Internet browsing (which can be encrypted by default) or BlackBerry messenger contents (assuming that the organization isn’t using the same global encryption key consumer messenger traffic relies on). If the BES and surrounding corporate IT infrastructure is outside a country’s legal reach then secured communications can be provided without worrying about government actually going after the mail or messaging servers themselves. Further, if a corporation’s legal assets and identity are also outside the nation, the government may be unable to legally compel the company to turn over the contents of BlackBerry communications. Needless to say, the full encryption of communications prevents the nation’s wireless carriers from effectively tapping BlackBerry device communications. Of course, this degree of security does depend on the device itself being protected from side-attacks, and protecting against these may limit the device’s full functionality.

Retrofitting Communicative Privacy and Security?

Over the past 24 months or so, various governments have decided that accessing secured BlackBerry communications is a national security priority. The actions taken by the Mumbai terrorists, who used BlackBerry devices to securely communicate with one another, have fuelled governmental demands to access privately secured data communications. What exactly has being demanded of RIM, why is it problematic to comply with these demands, and what is the next step from this point forward?

First, let’s showcase some of the countries demanding access to BlackBerry communications. The UAE has argued that BlackBerry devices pose ‘security risks’ on the basis that:

BlackBerry operates beyond the jurisdiction of national legislation, since it is the only device operating in the UAE that immediately exports its data offshore and is managed by a foreign, commercial organisation … As a result of how BlackBerry data is managed and stored, in their current form, certain BlackBerry applications allow people to misuse the service, causing serious social, judicial and national security repercussions.

Similarly, India remains ‘concern’ about their inability to decrypt secured BlackBerry communications. On the basis that encryption prevents rapid content penetration by government code-breakers, the Indian government sees BlackBerry communications as a national security issue. As noted by ZDnet, the general argument is that “India’s intelligence services need to be able to access encrypted data to prevent attacks in a ‘constant setting’: where attacks are likely and have occurred regularly.”

Other countries that have, or are presently, evaluating whether or not to let their businesses and citizens enjoy high levels of communicative security and privacy include:

• Kuwait. RIM has reportedly agreed to block thousands of pornographic websites after the government raise concerns about the cultural impact of these websites.
• Bahrain’s government successfully forced RIM to disable the messaging services for BlackBerry messaging chat groups on the basis that such groups could generate “chaos and confusion” as news was distributed via them. In effect, the BlackBerry limited the government’s ability to censor news that it didn’t want spread amongst its citizenry.
• Indonesia is concerned about BlackBerry encryption on the basis that the government is uncertain whether “data being sent through BlackBerrys can be intercepted or read by third parties outside the country.”
• Algeria, paralleling concerns raised by India, worries that the devices might be a “danger for our economy and our security.
• Lebanon is studying the security concerns around the BlackBerry.
• Tunisia has previously suspended email on the basis of security concerns.

Needless to say, this abbreviated list has a lot of nations citing ‘security concerns’ as driving the impairment of BlackBerry services. Also needless, but important, to note is that many of these same nations are well known for their efforts at censoring communications, oppressing their citizens, and regularly violating human rights.

What has RIM’s response been? In addition to blocking thousands of websites for the Kuwaiti government, RIM has provided decryption keys for the BlackBerry messenger service to India and believed to have provided them to Saudi Arabia as well. In the case of India, this apparently means that RIM is providing some kind of live access to BIS infrastructure that carries Indian messaging data. It’s important to carefully read and parse RIM’s official position regarding Saudi Arabia. Specifically, “RIM cannot accommodate any request for a copy of a customer’s encryption key, since at no time does RIM, or any wireless network operator or any third party, ever possess a copy of the key.” This seems deliberately nebulous, designed to confuse customers of the consumer line of BlackBerry services. While the company cannot crack Enterprise customers’ security on the basis that the BES architecture lacks back doors, the same cannot be said about wireless providers’ customers. Wireless providers’ customers use RIM’s BIS but are (arguably) not RIM customers themselves; RIM lacks a significant business relationship with them, save to potentially assist with hardware problems (and these are often dealt with at the provider level). Wireless customers using BlackBerry devices can almost certainly have their security and private infringed upon – RIM has effectively admitted as much by stating that they use a common global mode of encrypting messenger traffic (that they will disclose if forced) and that email data is subject to wireless companies’ own security policies (meaning it is subject to lawful access requests).

RIM has not yet capitulated to governments by redesigning their BES systems to provide governments access to BES-secured data. RIM repeatedly maintains in public that they cannot provide access to communications that are privately secured using the BES infrastructure, and that the company cannot monitor the content of BES-secured communications despite their flowing through RIM-based infrastructure. This stance may change, with evidence coming from the ongoing negotiations between RIM and the Indian government. The two parties are reportedly working towards some kind of an agreement that will give  the Indian government live access to data flows along BES environments. Presumably, this kind of ‘sneak peek’ would involve letting government officials look at data flows before they were encrypted going out of the enterprise, or after they had entered the corporate network. Alternately, all encryption keys might ‘just’ have to be registered with the national government. Save for in the last case, these ‘solutions’ would likely take the form of some kind of required plugin or module for Indian BES customers. Regardless, any agreement on any three of these lines will present BES customers and IT administrators with a myriad of security and confidentiality issues. It will only be a matter of time until some government official is bought out by a competing organization to perform corporate espionage or the government otherwise inappropriately uses their surveillance powers.

RIM and Single Points of Privacy Failure

Countries advocating for access to encrypted communications are demonstrating the danger of dependence on third-parties to route your communications: if the third-party is compromised then your communication may also be compromised. Many of the countries pressing RIM have already compromised their wireless carriers/ISPs, but RIM poses a somewhat unique danger insofar as it is a trusted and often extra-territorial third-party. RIM’s status alleviates some challenges of implementing and maintaining secure communications for some (typically business) individuals within the nation, but heightens problems for governments seeking access to all facets of their citizens’ communications. Countries are taking advantage of the fact that they can effectively shut down BlackBerry communications within their nations by placing pressure on regional wireless providers; such pressures threaten to deny RIM access to revenues and effectively force the company to the negotiating table. RIM is behaving as any ‘good’ profit-maximizing corporation would in light of threats its profits: it is negotiating deals that maximally enhance its balance sheet, principles and the privacy of wireless carriers’ customers be damned.

The Internet has demonstrated that it is an incredibly robust communications network, but one that does have weak links. Wherever there is a single, necessary, node that traffic must pass through there is a point of attack for government, a point where sovereign powers can be exercised to capture and interrogate citizens’ data traffic. The application of sovereign power demonstrates Goldsmith’s and Wu’s general thesis in Who Controls the Internet?, that the ‘net is becoming bordered as various powers mediate what kinds of data and data repositories are available to citizens. Of course, the general thesis must be nuanced: in each case where the BlackBerry network has come under fire from government we see governments at odds with other governments, governments trying to come to terms with private international data networks, and private corporations struggling to maintain product uniformity while accommodating regional law. In essence, we see governments struggling to adjust to a novel mode of network distributions, see them struggle to realize new approaches to govern communications traffic. Given the transformative nature of Internet governance generally, we would be well advised to take seriously Cowhey’s and Mueller’s (2009) conclusion that network governance has changed how the state system and communications networks interface. Specifically, they write that

…the Internet has not escaped governments, but the governance systems have changed. Changes in the rules of decision making and the forms of stakeholder participation will drive outcomes in novel directions even if the parameters of choice still remain under the control of governments (193).

Should facets of the BlackBerry system become more significantly decentralized we could see additional complications around the governance regime of mobile data communications. Such complications will contribute to additional anxieties around the range of actions available to the state in its self-expression of sovereignty. Whereas states have historically worked on places – a locale whose form, function, and meaning are self-contained within the boundaries of a physical contiguity – they are increasing being forced to work on spaces – instances of crystallized time that operate as a site of flows, and thus lack an international bounding of form, function and meaning associated with places (Castells 2001). The decentralization of BlackBerry services, a shift to a P2P-like infrastructure for their BIS and BES services, would limit states’ abilities to attend to places, undermining law’s ability to address BlackBerry security in a manner paralleling law’s limited capacity to end widespread P2P-enabled copyright infringement. Place has become a space of weakness, a point where time remains closely associated with matter and thus receptive to the “hard geophysical reality of places.” To achieve the advantages of decentralized virtualization places must give way to temporal structures associated with light-time and the manifestation of spaces that challenge geophysical locatedness (Virilio 2005: 117).

Even with the shift from places to spaces, the decentralization of BlackBerry security, and the modulation of state governance models, the switches of information transfers will remain privileged instruments of power. Thus, so long as RIM maintains vertical integration of their product lines the company will acts as a central point of power that is receptive to organized state power. Vertical integration is a problem; the company must shift from a single to pluralistic set of interrelated corporate nodes to transition from a places to spaces company. Disaggregating the vertical integration of the company would see it adopt a layered approach to its business, and manifest by the company spinning itself into a series of unique corporate bodies that maintain integration with other RIM-based corporate bodies without any particular body directly informing or integrating with one another in a centrally planned manner. Open protocols and APIs, instead of centralized corporate design, would be responsible for maintaining BlackBerry device and service integration. Under this framework a hardware, operating system, and network security corporation could emerge from the present whole that is Research in Motion.

Under this model the various corporations would contribute to a cohesive BlackBerry device, though no one party would own the entire stack. Google has demonstrated the viability of this approach, showing what a 21st century information company looks like and how it behaves (for much more on this, see Wu 2010). Exploring how RIM might implement a Google-like approach to corporate design and product architecture could simultaneously help confound sovereign authority and promote modular adjustments to facets of the BlackBerry infrastructure in ways that promote module innovation by giving developers free(r) rein over various ‘hidden components’ (those components that aren’t depended upon by other layers of the BlackBerry device stack). Such innovation and decentralization could continue to fulfill market demands of chasing after profits if open protocols and APIs are appropriately developed and propagated. Adopting this disaggregated approach, where RIM shifts from a places to a spaces company, would have the ultimate effect of challenging and undermining current structures of state sovereignty and accelerate the modulation of state power. By forcing states to engage with a better-entrenched networked governance structure that facilitates secured mobile communications the state might learn new modes of enacting governance requiring cooperation and compliance instead of blunt force. Without the tools of sovereignty the state typically wields, and the requirements to achieve cooperation and consensus, BlackBerry devices would enjoy enhanced security and their users superior communicative privacy. Importantly (for the RIM-spinoffs), the transition from a places to spaces corporation might be implemented whilst improving the conditions for modular innovation and enhancements to existing corporate profit logics.

Notes:

[1]  A5 encryption has serious deficiencies, which have been helpfully summarized by Harald Welte. In effect, A5 has long depended on security by obscurity to an extend and is quickly compromised in the face of a sufficiently motivated attacker.

Book Sources:

Cowhey, Peter and Mueller, Milton. (2009). “Delegation, Networks, and Internet Governance,” in M. Kahler (ed). Networked Politics: Agency, Power, and Governance. Ithaca: Cornell University Press.

Castells, Manuel. (2000). The Rise of the Network Society (Second Edition). Malden, MA: Blackwell Publishing.

Goldsmith, Jack and Wu, Tim. (2006). Who Controls the Internet? Illusions of a Borderless World. Toronto: Oxford University Press.

Virilio, Paul. (2005). The Information Bomb. New York: Verso.

Wu, Tim. (2010). The Master Switch: The Rise and Fall of Information Empires. New York: Alfred A. Knopf.