In mid-March, the group dubbed by Wired Magazine 20 years ago as Crypto-Rebels and Anarchists—the IETF—is meeting in London. With what is likely some loud humming, the activists will likely seek to rain mayhem upon the world of network and societal security using extreme end-to-end encryption, and collaterally diminish some remaining vestiges of an “open internet.” Ironically, the IETF uses what has become known as the “NRA defence”: extreme encryption doesn’t cause harm, criminals and terrorists do. The details and perhaps saving alternatives are described in this article.

Formally known as the Internet Engineering Task Force (IETF), the group began its life as a clever DARPA skunkworks project to get funded academics engaged in collective brainstorming of radical new ideas for DOD. It never created an actual organization—which helped avoid responsibility for its actions. During the 1990s, the IETF became embraced as a strategic home for a number of companies growing the new, lucrative market for disruptive DARPA internet products and services—coupled with continued copious funding from the Clinton Administration which also treated it as a means for promoting an array of perceived U.S. political-economic interests.

Over subsequent years, as other industry technical bodies grew and prospered, the IETF managed to find a niche value proposition in maintaining and promoting its legacy protocols. During the past few years, however, the IETF’s anarchist roots and non-organization existence have emerged as a significant security liability. The zenith was reached with the “Pervasive Encryption” initiative, bringing Edward Snowden virtually to the IETF meetings, and humming to decide on radical actions that met the fancy of his acolytes.

The Pervasive Encryption initiative

The IETF began doing Snowden’s bidding with the “Pervasive Encryption” initiative as their common crusade against what Snowden deemed “Pervasive Monitoring.” The IETF activists even rushed to bless his mantra in the form of its own Best Current Practice turned into a mitigation commandment called RFC 7258.

The initiative will come to fruition at a humming session in London at the IETF 101st gathering in a few weeks. The particular object of humming is an IETF specification designated TLS 1.3 and designed to provide extremely strong, autonomous encryption for traffic between any end-points (known as “end-to-end” or “e2e”). TLS = Transport Layer Security. The specification has been the subject of no less than 24 versions and more than 25 thousand messages to reach a final stage of alleged un-breakability. In the IETF vernacular, the primary design goal of TLS 1.3 is to “develop a mode that encrypts as much of the handshake as is possible to reduce the amount of observable data to both passive and active attackers.” How this occurs leverages an array of cryptologic techniques to achieve perfect “forward secrecy.”

There are perceived short-term benefits for some parties from the essentially invisible traffic capabilities between two end-points on devices anywhere in the world that are described below. However, the impacts are overwhelmingly, profoundly adverse. Innumerable parties over the past two years have raised alarms, and include multiple organizations and venues: workshops and lists within the IETF itself, vendor concerns, effects concerns, major enterprise users such as Financial Data Center Operators, major malware software vendors, the IEEE, the 3GPP mobile services community, the ITU-T Security Group and TSB Secretariat, a plethora of company R&D activities in the form of remedial product patents, trade press articles, and literally hundreds of research studies published in professional journals. The bottom-line view among the IETF activists, however, is “not our problem.”

The use of TLS by the IETF is somewhat ironic. Transport Layer Security (TLS) actually had its origins in early OSI industry efforts in the 1980s to provide for responsible security for the OSI internet. Indeed, an initial acceptable industry specification was formally published in the early 90s as a joint ITU-T/ISO (International Telecommunication Union Telecommunications Standardization Sector and International Organization for Standardization) joint standard that remains in effect today.

IETF crypto-activists a few years later took over the ITU-T/ISO internet TLS to roll out their own versions to compensate for DARPA internet cyber security deficiencies. However, it was the Snowden affection that primarily drove zealots to embark on TLS 1.3 as the crown jewel of the Pervasive Encryption initiative. A secondary but significant factor was the interest of Over-the-Top providers in free, unfettered bandwidth to customers leveraging the NetNeutrality political mandate, and added substantial fuel to the TLS 1.3 fire. Indeed, OTT providers have pursued a TLS variant known as QUIC—which allows for multiple simultaneous encrypted streams to end-user customers. QUIC creates major operational and compliance challenges similar to TLS 1.3 and is already being blocked. So as those in London hum for TLS 1.3 anarchy, what is gained and what is lost?

What is gained with TLS 1.3?

There are several “winners.” TLS1.3 makes eavesdropping significantly more difficult. There are fewer “handshakes,” so it should be faster than previous TLS versions. The platform enhances a sense of confidentiality for some individual users—especially the paranoid and those seeking increased protection for activities they want unknown. Those who profess extreme privacy zeal will likely be pleased.

For those engaged in any kind of unlawful activities, TLS 1.3 is a kind of nirvana. It includes those who seek to distribute and manage malware on remote machines—for either programmed attacks or for clandestine campaigns such as those manifested by Russian agents in the U.S. elections. Symantec has already presented statistics on how a considerable amount of malware is distributed via end-to-end encryption tunnels.

The platform also potentially enhances business opportunities and revenue for Over the Top (OTT) providers, and for vendors that leverage it for PR purposes. The latter includes some browser vendors and a few cloud data centre operators who cater to hosting customers for whom opaque end-to-end encryption for unaccountable activities is a value proposition.

TLS 1.3 also provides a perceived sense of satisfaction for those eternal “crypto anarchists” who have been labouring for so many years to best the government agency cryptologists and law enforcement authorities.

In a somewhat amusing, unintended way, the biggest winners may be the vendors of devices and software that detect and block TLS 1.3 traffic. They will benefit from the enormously increased market for their products.

What is lost with TLS 1.3?

TLS 1.3 (and QUIC) are already known to be highly disruptive to network operators’ ability to manage or audit networks. This occurs through a number of factors, but one of the most prevalent is that it breaks the functionality of the enormous number of network “middleboxes” that are essential for network operation. The problem is exacerbated in commercial mobile networks where the operator is also attempting to manage radio access network (RAN) bandwidth.

Because encrypted e2e transport paths in potentially very large numbers are being created and managed autonomously by some unknown third parties, a network provider faces devastating consequences with respect to providing sufficient bandwidth and meeting network performance expectations. It is in effect an unauthorized taking of the provider’s transport network resources.

As noted above, TLS 1.3 significantly facilitates widespread malware distribution, including agents that can be remotely managed for all kinds of tailored attacks. In the vernacular of cybersecurity, it exponentially increases the threat surface of the network infrastructure. The proliferation of Internet of Things (IoT) devices exacerbates the remotely controlled agent attack potential. Although, the counter-argument is to somehow magically improved the security at all the network end-points, the ability to really accomplish this fanciful objective is ephemeral and not real. It seems likely that most end users will view their loss of security and control of their terminal devices as much more important than any perceived loss of privacy from potential transport layer monitoring in transit networks.

A particularly pernicious result for enterprise network and data centre operators, including government agencies, is the potential for massive sensitive data exfiltration. A peripheral intruder through a TLS 1.3 encrypted tunnel into a data centre or company network could leverage their access to command substantial resources to gather and export intelligence or account information of interest. This potential result is one of the principal reasons for a continuing awareness campaign of the Enterprise Data Center Operators organization—coupled with proffering alternative options.

Most providers of network services are required to meet compliance obligations imposed by government regulation, industry Service Level Agreements, or insurance providers. The insurance impact may arise from an assessment that the potential liabilities of allowing TLS 1.3 traffic exposes providers to substantial tort litigation as an accessory to criminal or civil harm. The long list of compliance “by design” obligations are all likely to be significantly impeded or completely prevented by TLS 1.3 implementations:

• Availability (including public services, specific resilience and survivability requirements, outage reporting)
• Emergency and public safety communication (including authority to many, one to authority, access/prioritization during emergency, device discovery/disablement)
• Lawful interception (including signaling, metadata analysis, content)
• Retained data (including criminal investigative, civil investigative/eDiscovery, sector compliance, contractual requirements and business auditing)
• Identity management (including access identity, communicating party identity. communicating party blocking)
• Cyber Security (including defensive measures, structured threat information exchange)
• Personally Identifiable Information protection
• Content control (including intellectual property right protection, societal or organization norms)
• Support for persons with disabilities

Lastly, the implementation of TLS 1.3 is likely to be found unlawful in most countries and backed up by longstanding treaty provisions that recognize the sovereign right of each nation to control its telecommunications and provide for national security. Furthermore, nearly every nation in the world requires that with proper authorization, encrypted traffic must be either made available in decrypted form, or the encryption keys provided to law enforcement authorities—which TLS 1.3 prevents. Few if any rational nations or enterprises are going to allow end-to-end encrypted traffic transiting their networks or communicating with end-point hosts at data centres or users without the ability to have some visibility to assess the risk.

Myth of “the Open Internet”

The reality is that there have always been many internets running on many technologies and protocols and loosely gatewayed under diverse operational, commercial, and political control. In fact, the largest and most successful of them is the global commercial mobile network infrastructure which manages its own tightly controlled technical specifications and practices. With the rapid emergence of NFV-SDNs and 5G, internets on demand are beginning to appear.

The myth of a singular “Open Internet” has always been a chimera among Cyber Utopians and clueless politicians riding the Washington Internet lobbyhorse. The myth was begun by the Clinton Administration twenty years ago as an ill-considered global strategy to advance its perceived beneficial objectives and Washington politics. It came to backfire on the U.S. and the world in multiple dangerous ways. In reality, the humming approval of TLS 1.3 in London will likely diminish the “openness” within and among internets, but it will also properly cordon off the dangerous ones.

Thus, the perhaps unintended result of the IETF crypto zealots moving forward with TLS 1.3 will be for most operators to watch for TLS 1.3 traffic signatures at the network boundaries or end-points and either kill the traffic or force its degradation.

Innovation and a major industry standards organization to the rescue

Fortunately, there are responsible alternatives to TLS 1.3 and QUIC. For the past two years, some of the best research centres around the world have been developing the means for “fine-grained” visibility of encrypted traffic that balances both the security interests and privacy concerns. Several dozen platforms have been published as major papers, created innovative university programs, led to a major standards Technical Report, and generated even a seminal PhD thesis. A few have been patented. A number of companies have pursued proprietary solutions.

The question remained, however, what major global industry standards body would step up to the challenge of taking the best-of-breed approaches and rapidly produce new technical specifications for use. It occurred last year when the ETSI Cyber Security Technical Committee agreed to move forward with several Fine Grained Transport Layer Middlebox Security Protocols. ETSI as both a worldwide and European body has previously led major successful global standards efforts such as the GSM mobile standards now spun out as 3GPP, and the NFV Industry Standards Group, so it had the available resources and industry credentials.

Considerable outreach is being undertaken to many other interested technical organizations, and a related Hot Middlebox Workshop and Hackathon are scheduled for June. The result allows the IETF to hum as it wishes, and the rest of the world can move on with responsible alternatives that harmonize all the essential requirements of network operators, data centres, end users, and government authorities.