Above and beyond the obvious moral debate over whether or not a filtering option should be offered, whether it can be effective, or whether it should be opt-in

Above and beyond the obvious moral debate over whether or not a filtering option should be offered, whether it can be effective, or whether it should be opt-in or opt-out, a number of practical things worry me about having any such architecture in place: 1. Once you classify web activity it becomes more economical, and arguably even justified in order to carry the cost of maintaining the filter, to make the move to offering a tiered internet service (i.e. having to pay more for access to porn) 2. Not everyone in a household may have the same preferences, so you'll inevitably want to to start tracking web activity on a per-person basis 3. Your preferences will be kept by your ISP, probably their partners, and definitely made available to authorities on-demand. "The defendant opted-in to viewing violent pornography Your Honour". 4. To simplify the design the system will most likely require that all 'hits' in any given category be routed to a web proxy system, even while you're opted-out. You can bet that this will be logged meticulously by someone. Hell, some ISPs may even be brazen enough to offer the 'feature' of alerting parents to their kids activities. So it's potentially, "The defendant opted-in to viewing violent pornography Your Honour, and did so on a regular basis." 5. Massive potential impact to the robustness and performance of the web. More middle-boxes. Less flexibility to replace protocols, even HTTP 2.0, in the future.