The Inability to Opt Out of DPI (or Why the Marketplace Cannot Cure Paul’s Worries)
posted by Danielle Citron
Some might respond to Paul’s Ohm’s terrific article, The Rise and Fall of Invasive ISP Surveillance, by suggesting, as network providers do, that the marketplace will sort out our privacy concerns about Deep Packet Inspection practices because consumers can opt out of DPI tracking of their online life with a single click. Optimism about a proper functioning marketplace, however, is misplaced for several reasons. First, as Arstechnica reports, network providers bury notice of their inspection practices in densely worded privacy policies and do not email users to note the change in policy. Thus, a basic information asymmetry problem arises—consumers cannot reasonably be expected to know about, and protect themselves from, opaque practices. Second, even if consumers opt out of the creation of behavioral profiles for use in delivering ads, they may not be opting out of the copying of their traffic. And, third, as Dr. David Reed testified before the Subcommittee on Telecommunications and the Internet, even if some network providers switch to an opt-in approach or reject DPI entirely, consumers cannot totally control the use of DPI technologies by those with whom they communicate, thus rendering consumer choice illusory. Thus, the privacy concerns that Paul raises likely are not self-correcting.
September 12, 2008 at 1:54 pm
Posted in: Privacy (Consumer Privacy), Privacy (Electronic Surveillance)
Print This Post
Responses (3)
Paul Ohm - September 12, 2008 at 2:17 pm
I agree with all three points, and I would add a fourth: it is technologically difficult for an ISP to distinguish between users who have and have not opted out. For Google, this is easy to do–they merely need to look at the cookies sent to it. But you usually don’t send cookies to your provider.
So the provider is left remembering the IP addresses of those who have opted out, which they probably don’t want to do because it is clumsy and imprecise. Or, they can “trick” users into sending them cookies by redirecting traffic. This is much more precise, but it is even more clumsy and the user suffers a performance hit.
According to Richard Clayton’s report on Phorm, Phorm redirects user web requests as many as four times before allowing the user to connect. As far as I can tell, this is in part to allow Phorm to distinguish between those who have opted-out and those who have not.
Logical Extremes - September 12, 2008 at 5:55 pm
Another reason why ISPs doing it the IP address level is not a good idea… many devices can share an IP address by using a NAT router (most homes and businesses do this). IF I wanted, for some bizarre reason, to opt-in, it might be impossible if my IP address was opted-out. But the reverse is worse… my IP address got opted-in, but I want to opt-out for my device.
Logical Extremes - September 12, 2008 at 5:59 pm
Ah, I see now this is probably what you meant by imprecise”.
The other alternative is no good from the consumer perspective either… redirection reduces performance as you mention, adds a finite probability of introducing errors, and also makes the operations done within those redirections opaque.
Leave a Reply