|
Phil Zimmerman’s Pretty Good Privacy (PGP) and its offspring have been encrypting and decrypting email for almost 25 years—but require enough knowledge and determination to use them that adoption has never taken off outside the technoscenti. Now initiatives from several quarters aim to fix that—but will it all “just work,” and will end users adopt it even if it does?
According to a new Pew Research Center study of Americans’ attitudes after two years of disclosures about widespread government surveillance, 61% of respondents are less confident that these efforts are serving the public interest, and 57% said it is “unacceptable” to monitor the communications of US citizens. Despite this strong sentiment, only 18% of those surveyed indicated that they had changed the way they used email—even “somewhat”—as a result. Add this gap to the high bar end users have had to overcome in order to adopt email encryption, and how likely is it that these new tools and services will trigger a change in behavior?
Not Widespread After Two And A Half Decades
People who regularly say things that can put them in danger—activists, dissidents, journalists—may come to depend heavily on encrypted email. But it never really caught on with average email users, probably because in the past it never occurred to them to worry about who might see their messages other than a nosy spouse or partner. Even if they felt the need, the steps involved were fairly arcane for the average consumer. And if they overcame that hurdle, and if somebody they wanted to swap encrypted messages with did too, then exchanging and loading the necessary keys was often a bridge too far.
The community around PGP tried to make key exchange easier by creating public keyservers and programming plugins for just about every email client written. But the adoption curve was never driven far enough to trigger the network effect—likely because of the number of unusual (to Joe Sixpack) steps involved in generating a key and getting it onto that keyserver in the first place. Similar issues affected alternatives like S/MIME, outside of certain business environments and platforms, where the equivalent hurdle was obtaining or exchanging valid certificates. Each system worked well where it was used, but none of them really impacted the use of email on Main Street.
Instead the form of encrypted email most often encountered by consumers has typically been a small, self-contained system—often deployed by banks or healthcare providers—that only allowed them to exchange messages with people at those organizations. In many cases this was just a captive webmail service accessed from a web browser over a TLS-encrypted session, with content-free “you have a message” notes going to a customer’s regular email address to prompt them to visit the portal. In these highly regulated industries the expense of deploying these systems is often easy to justify, especially when the alternative is an envelope sent via courier or next-day service.
Along Comes Citizen Four
Since the Edward Snowden leaks made the depth and breadth of recent government surveillance public, there has been renewed interest in encrypting email—along with just about every other kind of Internet traffic. And after a few years of steady work, a number of initiatives are coming to the fore.
Since 2008 the German government has been working on an email service called DE-Mail. The initial goal was to support the exchange of legally binding electronic communications and documents between citizens, businesses, and government. But according to German officials, beginning in April 2015 the platform will offer end-to-end encryption of messages through browser plugins, which will be based on PGP. While the DE-Mail platform hasn’t been wildly popular with consumers to date, this new service might change that—and the announcement certainly reflects a different attitude on the part of the German government, compared to the official UK or US positions that end-to-end encryption threatens the effectiveness of law enforcement.
In early 2014 a small startup called Keybase.io began getting attention, at least partly because of the founders’ track record with SparkNotes and OkCupid. They set out to update the traditional PGP keyserver and attestation models, incorporating public proofs of identity based on social media and other services. They also offered both command-line and browser-based code that would simplify many of the details of key management and encryption for end users—though perhaps allowing users to upload their private keys for ease of portability is a step too far. Still, the focus on simplifying things for the end user is laudable, and it is a standalone service that you use with your existing email account. Their keyserver is integrated with the existing PGP keyservers, and their simpler user interface can be used on top of publicly reviewed and vetted open source programs.
In June of last year, Google announced it was developing a Chrome extension that would make end-to-end encryption with PGP a lot easier for end-users—which, in a blaze of creativity, they named End-To-End. While this extension still hasn’t reached the Chrome Web Store, the source code has been publicly available for over six months and other messaging players such as Yahoo have been participating in the project. Like Keybase.io, Google is revisiting the keyserver—but this time taking a new look at the entire key distribution model, with an approach that draws on concepts from Certificate Transparency.
At Black Hat USA last summer, Yahoo’s Alex Stamos promised that they would deliver an end-to-end encryption capability for Yahoo Mail users. On Sunday at the South by Southwest Festival, he announced that Yahoo was on track to deliver the functionality by the end of this year, showed a video of a beta version that was much easier to setup and use than traditional PGP clients, and announced that their version of the code from their collaboration with Google was available on Github for public review.
But If You Build It, Will They Come?
Whether or not DE-Mail sets a standard for Europe, whether or not Keybase.io can revitalize PGP for users at large, anything that is interoperable and adopted by both Google and Yahoo is going to be hard to ignore. And having that many potential correspondents in their key distribution system is going to be a powerful motivator to maintain interoperability with existing keyservers, even if the latter have to make some changes to do so.
Will end-users adopt it? I think so, though even if it only takes a few extra clicks, it probably won’t be used for most messages—you just aren’t going to encrypt a quick note about when to take the kids bowling this weekend. But if it is just a few clicks, and if most of your contacts are in the same position, then I think you’ll see modest growth from the consumer side. However the ability to reach so many consumers without deploying expensive, specialized infrastructure might represent an opportunity for lots of businesses to communicate more securely with their customers, and vice versa. And I think that could be the one-two punch that finally changes expectations, and gets adoption moving on a broad front.
Note: I have referred almost exclusively to Pretty Good Privacy, or PGP, which is the progenitor and probably most-recognized of a family of compatible implementations. But the GNU Privacy Guard, also known as GnuPG or GPG, may be the most widely deployed example as it is found in most Linux distributions. OpenPGP refers to an IETF Proposed Standard or protocol that these programs implement, RFC 4880, and which is free for use without licensing fees.
Sponsored byDNIB.com
Sponsored byWhoisXML API
Sponsored byCSC
Sponsored byVerisign
Sponsored byVerisign
Sponsored byIPv4.Global
Sponsored byRadix
The big problem with end-to-end e-mail encryption has always been key generation. That’s always been the unsolvable part. PGP had automatic seamless encryption, key exchange, key lookup and all that solved. But none of that helps if users don’t generate keys, and that has to be tied into the e-mail program itself so that if it’s got encryption available and the user doesn’t have a key it’ll start stepping the user through the process of generating one. Without that most recipients won’t have keys and you can’t encrypt e-mail to someone who doesn’t have a key. The move to webmail’s made it worse, since webmail and secure keys are almost mutually exclusive.
It’d be useful to look at the problem as an offshoot of SSL certificate generation, not in isolation, because most of the issues of generating and distributing e-mail keys are much the same.