Science and technology

Babbage

Internet security

A sense of false security

Mar 2nd 2011, 15:10 by G.F. | SEATTLE

TO MANY of his constituents, Charles Schumer, a Democratic senator from New York, appeared to spout gibberish on Sunday. "Major web sites [should] switch to secure HTTPS web addresses instead of the less secure HTTP protocol," he told Reuters in a Manhattan coffee shop. Mr Schumer's statement, however, constitutes perfectly sensible advice—he was well briefed by his staff. Such a move would prevent theft of casual digital identities and personal information in public places—and hinder politically motivated interception by repressive (or democratically elected) governments.

HTTPS is the secured or encrypted form of HTTP (Hypertext Transfer Protocol), a communications language that directs the way in which web browsers and web servers interact to request and retrieve pages, images and other files. HTTPS layers encryption on top of plain HTTP using SSL/TLS (Secure Sockets Layer/Transport Layer Security). These are the old and current names for web-page securing technology that dates back to the world wide web's juvenile days, not long after Netscape alerted the masses to its existence.

Websites that offer SSL/TLS security allow connections via a URL that starts with "https" in the location field or link. First, the browser silently requests security credentials that the server provides. Next, it validates this information independently using either its own built-in data or those included in the operating system. If it passes muster, the browser and server exchange an encryption key, unique to each session, which is then used to guard the data that passes between them. Any whiff of interception or rerouting is enough to alert the user. Because of the way browsers and operating systems validate SSL/TLS certificates, an interloping party (the so-called "man in the middle") cannot pretend to be a secured server (to a browser) or a secured browser (to a server) without provoking such warnings. 

Flaws in earlier versions of SSL/TLS were patched up years ago and it is generally regarded as foolproof—and vital. The risk of not using it was readily demonstrated in the early stages of Tunisia's recent upheaval. The government allegedly intercepted connections between citizens and the unencrypted version of Facebook's local site, as Alexis Madrigal explained on January 24th in the Atlantic. The government could then intercept traffic by pretending to be Facebook; users, unaware, would blithely bung in their credentials, handing over access to their account and their entire social network. (To its credit, Facebook decided to flip on SSL/TLS for all of Tunisia and, later, made it available as an account preference worldwide. The internet company has offered HTTPS for some time but users outside Tunisia still have to opt in.) 

Mr Schumer's statement, and a letter he has sent to large web site operators, comes a decade after free software appeared that made it trivial for the mildly knowledgeable to intercept any data over an open Wi-Fi network, like nearly all of those in cafés or at airports. (Office and home networks protected with some form of password are a different matter.) While corporations typically require employees to use encrypted connections known as a VPN (virtual private network), ordinary users have, by and large, remained oblivious. This lack of concern may stem from the near-univeral use of HTTPS by banking, investment and e-commerce sites to protect logins, transactions and credit-card data. The lock icon which pops up in browsers for such sites may have lulled less tech-savvy types into complacency. But the massive growth in the use of web apps for email and social-networking sites exposed information identity thieves and other scammers relish.

In 2007 software was released which could intercept bits of data used by websites to identify a user from anyone on the same public Wi-Fi network. Session tokens, as these bits are called, are generated after a login, in which a secure connection is used just long enough to allow the entry of a username and password before the web browser is redirected back to an unsecured version of the website. By grabbing hold of these, impostors were able to "sidejack" a Gmail account or other services that his victim had accessed. With access to email, an attacker could visit popular sites, reset a user's password and use email to retrieve login information. Following a flurry of sidejacking activity Google began the process, which ended up taking several years, of tweaking most of its services to provide SSL/TLS as an option (though not a requirement).

A smattering of technical know-how was needed to sidejack—and the sidejacker had to be in close proximity of a sufficient number of users to make it worthwhile. Two developments have changed that equation. First, the release of a proof-of-concept plug-in for the Firefox browser, called Firesheep, made worldwide headlines last October. With a couple of clicks, even the most unsophisticated user could take over the identity of anybody else on the same network that happened to be browsing any of a few dozen popular websites. (Mr Schumer fingered Firesheep in his public appearance.) Second, the growth of smartphones and tablets with Wi-Fi connectivity—along with the spread of free networks in America—dramatically increased the number of proximate targets. A few years ago a sidejacker (or "sniffer") might have had access to a handful of laptops from which to siphon data; now hundreds of smartphones and slates can be logged on to such networks at any given time.

More worrisome than sidejacking is the wholesale interception of unsecured web traffic by governments. This allegedly happened in Tunisia and is believed to occur routinely in many countries. The open internet in many countries passes through a series of chokepoints at which interception is trivial and may, in fact, be mandated. HTTPS does not solve the problem entirely—tracking internet addresses accessed by users may provide indirect information about contacts. But encryption hampers governments or other parties that want to view the content of messages. It also makes it more difficult to maintain that interception did not occur.

There are other niggles. A widely known proof of concept from 2009, called sslstrip, intercepts unsecured web traffic on an open network and rewrites HTTPS links into plain HTTP or redirects them to malicious secured sites that use lookalike domain names. Users have to be attentive, or install additional security extensions, to identify attacks using this approach. Should the common form of sidejacking become trickier to execute, the use of this more elaborate ruse would doubtless spread.

Major websites now generally offer secured connections but not as a default option. Mr Schumer recommends they ought to do just that. He is right. Without securing a connection from start to finish, users are vulnerable to identity theft, and much worse, by anyone that happens to be sitting in a convenient spot in the sequence of connections from user to server. While Mr Schumer prefers to stress the public-hotspot end of the chain, HTTPS guards against government meddling, too.

You must be logged in to post a comment.
Please login or sign up for a free account.
1-17 of 17
Mar 2nd 2011 3:19 GMT

When will the economist be switching?

Mar 2nd 2011 4:44 GMT

@Gaeroid
Quite a while I imagine. https:// combined with third party advertising makes for a complicated mess.

Mar 2nd 2011 5:01 GMT

@remythecrasher: Having installed a few web-advertising systems, the SSL limitation is a thing of the best. Typically, JavaScript code handles the advertisement call, and it has a statement to make a request by https if the current page is encrypted. As long as all items on an SSL/TLS-served page are also served from https links, the browser doesn't squawk.

Mar 2nd 2011 5:43 GMT

Good question! We'll be switching on https for user pages (login, registration, etc) in the next few weeks.

Reido wrote:
Mar 2nd 2011 6:16 GMT

Secure email should gather more attention, too. There's no telling for sure how many different "hands" it goes through before it reaches the destination user. But of course it's harder to set up due to the need for individuals to maintain their own public/private keypairs.

Mar 2nd 2011 8:19 GMT

@G.F.
You're correct. I've just been running my paranoia settings too high. (I like to know what comes from where.)

timwills wrote:
Mar 2nd 2011 9:10 GMT

gosh this gets complex fast - I was hoping Rapport did the trick on the secure stuff... Expat Shield to create some confusion and be able to watch the BBC. Darknet is when we move country...

jouris wrote:
Mar 2nd 2011 9:14 GMT

We'll know that things have significantly improved when Yahoo and gmail routinely use https for all of their accounts. Until then, caution is definitely indicated.

D. Sherman wrote:
Mar 2nd 2011 10:58 GMT

This article does a good job of explaining what HTTPS is good for and why more sites should use it. What it doesn't do is explain why they're reluctant to do so. There must be some reason other than mere inertia. It would at least be worth asking the question. Sometimes when a lot of people do something a certain way, they're being foolish, but most of the the time when a lot of people do something a certain way, there's a reason for it.

Mar 2nd 2011 11:08 GMT

@D. Sherman: "This article does a good job of explaining what HTTPS is good for and why more sites should use it. What it doesn't do is explain why they're reluctant to do so."

An excellent question. I have asked this question of many firms for over a decade, and the answer for quite a while was a combination of cost, complexity, and accessibility.

The processing needs for SSL/TLS used to be much higher relative to unencrypted traffic, and thus require much more equipment or much more expensive individual servers. The premium was clearly in the double digits which, for any company with 1,000s of servers, boosts expenses. Newer processors have dedicated encryption components now (even on the desktop: viz., the latest Intel i5 and i7, which Apple is using for full-disk encryption in its next operating system), and processing power has become much less of a cost of running a server relative to other components and the overhead and support. (Symmetrically, this was also an issue for pokier computers and mobile devices, which would bog down in the continuous encryption process.)

Complexity and accessibility: Companies were worried (and rightly so) that introducing a new element in the chain of things that can go wrong would mean fewer customers. This was true in the early oughties when SSL/TLS sometimes went awry and browsers were simply programmed in a much poorer fashion. As for accessibility, some customers could not reliably get SSL connections to work because of government or ISP blocking policies, older browser, and so forth. Those concerns have nearly entirely disappeared.

I expect that the real reason, like Y2K, US stock market decimalisation, and IPv6 address depletion: there is no profit motive in using SSL/TLS rather than unencrypted connections. Companies will have to spend money and time (even if slight amounts) to make the transition without receiving additional revenue as a result.

mekinney04 wrote:
Mar 2nd 2011 11:34 GMT

Although the concerns around SSL protocol or lack of use with popular web services are valid, HTTPS is by no means a panacea for privacy. Such simple mechanisms as browser behavior (e.g, caching session data in CLEARTEXT) or malware sitting on the end user's machine, also create significant privay / data loss issues, even for proper HTTPS connections.

Anjin-San wrote:
Mar 3rd 2011 1:22 GMT

Good to see that the Economist does recognizes it is in Rhodos, and is preparing to jump the jump....

Tim Callan wrote:
Mar 3rd 2011 4:18 GMT

"A widely known proof of concept from 2009, called sslstrip, intercepts unsecured web traffic on an open network and rewrites HTTPS links into plain HTTP or redirects them to malicious secured sites that use lookalike domain names. Users have to be attentive, or install additional security extensions, to identify attacks using this approach. Should the common form of sidejacking become trickier to execute, the use of this more elaborate ruse would doubtless spread."

In fact, there is a great solution to the sslstrip problem which already has been widely implemented by many financial and e-commerce sites among others. It's called Extended Validation (EV) SSL, more commonly referred to as the green address bar. When implemented, EV SSL causes the current browsers to display the authenticated name of the business that's running this site in green immediately adjacent to the web address. You can easily see what it looks like by going to PayPal or Bank of America, among others. SSLstrip and its kin separate the client system from the true server that is being counterfeited, and therefore the green address bar will be absent under this attack.

The reason so many financial and other sites have implemented EV SSL is to counter a family of attacks, including SSLstrip, that depend on fooling a user into giving away login credentials, personally identifiable information, or other confidential information. Usability research indicates that the green address bar is widely noticed among active internet users and, in its absence, an effective signal that a user is on the wrong site.

Disclosure: I am a Symantec employee (which is the parent company operating VeriSign SSL, the most popular brand of SSL), an industry spokesperson, and a founding member of the industry standards body that created EV SSL. I have a more complete response to the SSLstrip attack available on my blog.

tertrtre wrote:
Mar 3rd 2011 4:28 GMT

love friend-welcome come to website :
www. ( vip s hop per.us ) ( Please enter the website
... , you can leave the url written in can see the place or brain
) Have what you want
..Smart oh great

Mar 3rd 2011 4:55 GMT

love friend-welcome come to website :
www. ( loveshopping.us ) ( Please enter the website
... , you can leave the url written in can see the place or brain
) Have what you want

Mar 3rd 2011 4:55 GMT

love friend-welcome come to website :
www. ( loveshopping.us ) ( Please enter the website
... , you can leave the url written in can see the place or brain
) Have what you want

1-17 of 17

About Babbage

In this blog, our correspondents report on the intersections between science, technology, culture and policy.

Follow Babbage on Twitter »

Advertisement

Advertisement

Latest blog posts - All times are GMT

Kabuki comes home
From Asia view - 1 hrs 55 mins ago
Link exchange
From Free exchange - March 2nd, 21:42
An abundance of activity
From Multimedia - March 2nd, 21:14
About that Goldman estimate
From Free exchange - March 2nd, 21:10
More from our blogs »
Products & events
Stay informed today and every day

Subscribe to The Economist's free e-mail newsletters and alerts.


Subscribe to The Economist's latest article postings on Twitter


See a selection of The Economist's articles, events, topical videos and debates on Facebook.

Advertisement