The internet’s engineers see the network under attack, but do not quite know what to do about it
By M.E. | VANCOUVER
“THE NSA has turned the internet into a giant surveillance platform.” Security guru Bruce Schneier (pictured) did not pull his punches when he addressed the 1,200 engineers gathered for the meeting of Internet Engineering Task Force (IETF) in Vancouver last week. But when it came to the question of what should be done about it, he and the other participants in a panel discussion had less to offer.
Mr Schneier, a fellow at Harvard's Berkman Centre on Internet and Society, is one of the few people who had seen most if not all the NSA documents downloaded by Edward Snowden. Only a few have been made public so far, with the most recent revelation being the stealth tapping of Google's internal networks.
“There is a lot more to come,” Mr Schneier warned. “But there is a lot we are never going to know.” Neither details of the encryption standards said to have been manipulated, nor the names of the vendors were in the documents, he admitted.
In some respects, Mr Schneier argued, it is all the internet’s fault—and that of the engineers who built it. The possibility of surveillance is baked into the network. Data, he contended, are a by-product of the information society. All computer processes produce data, and these are being "increasingly stored and increasingly searched". The result is "wholesale surveillance, surveillance backwards in time, the loss of ephemeral conversation, systems that never forget."
And, he added, it is not that the NSA woke up one morning and said: “Let’s spy on everybody.” Instead, they looked around and said: “Wow, corporations are spying on everybody, let's get ourselves a copy.“
Unsurpringly, not everybody on the panel agreed. "Engineers were not complicit,” insisted Stephan Farrell, a researcher at Trinity College in Dublin who heads the IETF's security arm. True, the internet’s engineers knew a lot about the technologies used by the NSA—after all, they had developed many of them. Yet the boffins' "threat model" did not entertain the possibility that so many could be deployed on such a scale and simultaneously.
Encryption is at the core of the defense proposed by Mr Farrell and many other IETF engineers. Even if not unbreakable, encryption ensures that text at least does not lie bare before prying eyes. After an emotional debate in Vancouver, consensus was reached that as much encryption as possible should be injected into the internet’s protocols.
The IETF engineers also agreed that henceforth a new threat model of pervasive surveillance had to be taken into account. Ted Hardie proposed that developers, when writing code, ask themselves: "Can a gay kid in Uganda use this safely?” If the answer was "no" it would be back to the drawing board.
This sounds promising, but there are trade-offs and the internet is (mostly) a business. When asked why encryption had not been put in place before, Jari Arkko, the IETF’s chair, confessed that this was because it is hard to do. One problem has been the ambition to make encryption close to perfect, Mr Farrell argued. The deployment of the secure versions of protocols—https instead of http, IPSEC instead of IP and so on—have been such a huge investment of time and money that the market decided against it.
Many new ideas have been presented in the wake of Mr Snowden's revelations, for instance to "encrypt all lit fiber“. Mr Arkko prefers "lower hanging fruit", such as better security for the web protocol http which is now under review. But developers, he also said, should use the post-Snowden momentum to push forward with the harder and more expensive task of actually implementing more secure systems.
