Skip to content

Commit 88fda23

Browse files
authored
Merge pull request #46 from intarchboard/caw/trust-and-centralization
Trust and centralization
2 parents a008893 + 080e998 commit 88fda23

File tree

1 file changed

+24
-4
lines changed

1 file changed

+24
-4
lines changed

draft-iab-privacy-partitioning.md

Lines changed: 24 additions & 4 deletions
Original file line numberDiff line numberDiff line change
@@ -98,7 +98,7 @@ This is a simple application of the principle of least privilege, wherein every
9898
a system only has access to the minimum amount of information needed to fulfill their
9999
function. Privacy partitioning advocates for this minimization by ensuring that protocols,
100100
applications, and systems only reveal user-specific information to parties that need access
101-
to the information for their intended purpose.
101+
to the information for their intended purpose.
102102

103103
Put simply, privacy partitioning aims to separate *who* someone is from *what* they do. In the
104104
rest of this section, we describe how privacy partitioning can be used to achieve this goal.
@@ -528,14 +528,15 @@ Privacy partitions ensure that only the client, i.e., the entity which is respon
528528
can link all user-specific information together up to collusion. No other entity individually
529529
knows how to link all the user-specific information as long as they do not collude with each other
530530
across contexts. This is why non-collusion is a fundamental requirement for privacy partitioning
531-
to offer meaningful privacy for end-users.
531+
to offer meaningful privacy for end-users. In particular, the trust relationships that users have
532+
with different parties affects the resulting impact on the user's privacy.
532533

533534
As an example, consider OHTTP, wherein the Oblivious Relay knows the Client identity but not
534535
the Client data, and the Oblivious Gateway knows the Client data but not the Client identity.
535536
If the Oblivious Relay and Gateway collude, they can link Client identity and data together
536-
for each request and response transaction by simply observing the requests in transit.
537+
for each request and response transaction by simply observing requests in transit.
537538

538-
It is not currently possible to guarantee with technical protocol measure that two
539+
It is not currently possible to guarantee with technical protocol measures that two
539540
entities are not colluding. However, there are some mitigations that can be applied
540541
to reduce the risk of collusion happening in practice:
541542

@@ -603,6 +604,25 @@ and protocol). This has a number of practical implications, described below.
603604
optimized nature of proxy-to-proxy paths leads to improved perforamnce. In general, while performance
604605
and privacy tradeoffs are often cast as a zero sum game, in reality this is often not the case.
605606

607+
1. Increased attack surface. Even in the event that information is adequately partitioning across
608+
non-colluding parties, the resulting effects on the end-user may not always be positive. For example,
609+
using OHTTP as a basis for illustration, consider a hypothetical scenario where the Oblivious
610+
Gateway has an implementation flaw that causes all of its decrypt requests to be
611+
inappropriately logged to a public or otherwise compromised location. Moreover, assume
612+
that the Target Resource for which these requests are destined does not have such an
613+
implementation flaw. Applications which use OHTTP with this flawed Oblivious Gateway to
614+
interact with the Target Resource risk their user request information being made public,
615+
albeit in a way that is decoupled from user identifying information, whereas applications
616+
that do not use OHTTP to interact with the Target Resource do not risk this type of disclosure.
617+
618+
1. Centralization. Depending on the protocol and system, as well as the desired privacy properties, the
619+
use of partitioning may inherently force centralization to a select set of trusted participants.
620+
As an example, the impact of OHTTP on end user privacy generally increases proportionally to the
621+
number of users that exist behind a given Oblivious Relay. That is, the probability of an Oblivious
622+
Gateway determining the client associated with a request forwarded through an Oblivious Relay decreases
623+
as the number of possible clients behind the Oblivious Relay increases. This tradeoff encourages
624+
centralization of the Oblivious Relays.
625+
606626
# Security Considerations
607627

608628
{{limits}} discusses some of the limitations of privacy partitioning in practice. In general,

0 commit comments

Comments
 (0)