Category: Internet matters

The latest XKCD comic identifies one of the major security failings of the internet today: the tendency of users to use the same password on more than one important site. It’s fine to use the same password for a bunch of news sites that do not store important personal information. What’s foolish is using the same password for a potentially vulnerable site and for something important, like a bank’s website or the password on an encrypted hard drive partition. Doing so risks allowing someone to compromise your information, one step at a time.

Another related risk is password recovery systems. Countless websites allow users to either have their password emailed to them or reset their password via email. That means that anybody who gains access to an email account linked to such features can then gain access to any sites that rely on that sort of password replacement system.

The wisest thing seems to be using strong unique passwords for email and other important sites, then having a couple of lower tier passwords to use for general sites that do not pose security risks. Random.org has a password generator, though the trick of building up a password from a memorable piece of music or poetry is probably less troublesome and still quite secure. An alternative approach is to have unique passwords for everything and rely on a password management program (or a piece of paper kept guarded in your wallet) to keep track of them.

Online security would also be better if all sites allowed the use of passphrases, rather than just passwords (and sometimes ones with an absurdly short maximum length). Two-factor authentication can also help.

In theory, quantum cryptography (mentioned before is as good as a one time pad, without the need for a secure channel through which to exchange keys. Potentially, it could also employ quantum phenomena to verify that nobody is eavesdropping.

In practice – as with all cryptographic systems – there are weaknesses to be exploited. One known attack exploits a weakness in some sorts of photon detector. Another works by manipulating synchronization signals.

Quantum cryptography may well have some useful applications, but people who expect it to be foolproof and completely secure probably aren’t thinking too well.

The above, famously, is Google’s motto. When I first saw it, it seemed like an embodiment of the ways in which Google differs from other large corporations. They are involved in charitable works, in areas including infectious disease and renewable energy. Furthermore, they give away most of their products, getting the financing from those famous automatic ads.

On further reflection, however, “Don’t be evil” isn’t some lofty, laudible goal we should applaud Google for having. Rather, it is the absolute minimum required of them, given just how much of our personal information they have acquired. Think about GMail: many of us have tens of thousands of messages, many of them highly personal, entrusted unencrypted to Google’s servers. If they were evil – or even a few of their employees were – they could embarass or blackmail an enormous number of people. What Google has is, in many cases, far more intimate than what sites like Facebook do. Facebook may have some private messages to your friends, but Google is likely to have financial information, medical test results, photos you would never put on Facebook, etc.

Now, Google has incorporated a very useful phone calling system into GMail. Install a plugin, and you can make free calls to anywhere in Canada and the United States. In my limited experience, it seems to work better than SkypeOut, while being free to boot. Of course, it is another example where we really need to trust Google to behave ethically. For Google Voice, they already developed algorithms to convert spoken words into transcribed text. Users of their phone service need to trust that their conversations are not being archived or – if they are – that the transcripts will not be used in any nefarious ways.

In short, Google must avoid being evil not out of benevolence, but because their whole business model requires people to view them that way. So far, their products have been remarkably empowering for a huge number of people (any other sort of email seems deeply inferior, after using GMail). If they are going to maintian the trust of users, however, they are going to need to avoid privacy disasters, or at least keep them on a pretty minor scale, like when Google Buzz abruptly let all your friends know who else you are in contact with.