Full submission is available at:
• Avoid use of DPI. The use of DPI raises serious privacy concerns that have not been resolved.
• Traffic management techniques should be applied uniformly to all users. Targeting a sub-set of users can be arbitrary and therefore unfair. Application targeting is an example of this. For example, throttling all P2P file-sharing application users will also catch many users of P2P file-sharing applications that generate only a small amount of bandwidth.
• Disproportionate targeting of lower tier customers should be avoided. If, for example, a customer with a 1 Mbps line and a customer with a 10 Mbps line were, in combination, generating enough traffic to congest a link they shared, they should each be throttled in proportion to the bandwidth they have paid for.
• Targeting newly developed protocols or applications should be avoided. Such innovations may make easy targets at first, while they are only employed by a small subset of users, as interfering with such traffic will only impact on less customers. This is true of P2P throttling. Telcos/Cablecos can now say that a relatively smaller portion of users are generating a large amount of P2P traffic and so this type of traffic should be targeted. However, given the efficiency in bandwidth distribution P2P offers customers, its use is only likely to increase in the future. As P2P use becomes more ubiquitous, the rationale that a small number of users are generating large amounts of P2P traffic will be become more inaccurate. More importantly, allowing telcos/cablecos to target a newly developed application or protocol because a.) it currently has a small number of users and b.) it happens to be very effective and so generates a large amount of bandwidth traffic, is likely to hinder innovation.