First, let's consider that "the goal is to prevent an eavesdropper on the local end from noticing Tor usage". For argument's sake, let's say that it's easier, at any particular time, to identify all VPN service entry servers than it is to identify all Tor bridges. And let's say that all Internet traffic is logged, and at least metadata (such as IP addresses, times and so on) is retained indefinitely.

Although Tor bridges may well be "harder to monitor", it's arguable that that's only so transiently. That is, virtually all Tor bridges are sooner or later identified, and are replaced with new bridges. Is that correct?

Consider the implications for observers of my ISP-assigned IP address. If I were using Tor bridges, all of my prior traffic to a particular Tor bridge would be readily found once that bridge had been identified. And my Tor use would be apparent, given that all Tor bridges (I assume) lead to Tor.

If I were instead accessing Tor through a VPN service, identifying my Tor use would require either traffic analysis based on intercepts, or access to VPN connection logs. In contrast to the Tor-bridge scenario, identification of past Tor use would arguably become harder over time.

But of course, if I were truly concerned about hiding my Tor use, I would use both VPN services (in nested chains) and Tor bridges with Obfsproxy.

Second, regarding the "benefit to stringing together multiple VPNs that cannot be achieved by increasing the number of relays per circuit", the most obvious one is usability. On a good day, I can get at least 5 Mbps each way with a nested chain of three VPN services, with latency under 400 msec. With the standard Tor configuration, I'm lucky to get 2-3 Mbps each way, and latency is typically 1000-2000 msec.

Another benefit of using multiple VPNs instead of Tor bridges is that Tor traffic is further disassociated from users. Even if one could chain multiple Tor bridges, all prior Tor use would be revealed as soon as the initial bridge had been identified.

Identifying Tor users in VPN traffic, especially retrospectively, is arguably harder than searching based on known Tor bridges. Many people use VPN services, far more than use Tor. BitTorrent use and HD video streaming account for a substantial and increasing share of Internet throughput overall. And, in response to efforts to block "piracy", that traffic is increasingly migrating to VPN services.

First, let's consider that "the goal is to prevent an eavesdropper on the local end from noticing Tor usage". For argument's sake, let's say that it's easier, at any particular time, to identify all VPN service entry servers than it is to identify all Tor bridges. And let's say that all Internet traffic is logged, and at least metadata (such as IP addresses, times and so on) is retained indefinitely.

Although Tor bridges may well be "harder to monitor", it's arguable that this is so only transiently. That is, sooner or later, virtually all Tor bridges are identified, and replaced with new bridges.

If that assessment is accurate, consider the implications for observers of my ISP-assigned IP address. If I were using Tor bridges, all of my prior traffic to a particular Tor bridge would be readily found once that bridge had been identified. And my Tor use would be apparent, given that all Tor bridges (I assume) lead to Tor.

If I were instead accessing Tor through a VPN service, identifying my Tor use would require either traffic analysis based on intercepts, or access to VPN connection logs. In contrast to the Tor-bridge scenario, identification of past Tor use would arguably become harder over time.

But of course, if I were truly concerned about hiding my Tor use, I would use both VPN services (in nested chains) and Tor bridges with Obfsproxy.

Second, regarding the "benefit to stringing together multiple VPNs that cannot be achieved by increasing the number of relays per circuit", the most obvious one is usability. On a good day, I can get at least 5 Mbps each way with a nested chain of three VPN services, with latency under 400 msec. With the standard Tor configuration, I'm lucky to get 2-3 Mbps each way, and latency is typically 1000-2000 msec.

Another benefit of using multiple VPNs instead of Tor bridges is that Tor traffic is further disassociated from users. Even if one could chain multiple Tor bridges, all prior Tor use would be revealed as soon as the initial bridge had been identified.

Identifying Tor users in VPN traffic, especially retrospectively, is arguably harder than searching based on known Tor bridges. Many people use VPN services, far more than use Tor. BitTorrent use and HD video streaming account for a substantial and increasing share of Internet throughput overall. And, in response to efforts to block "piracy", that traffic is increasingly migrating to VPN services.

OK, let's consider that "the goal is to prevent an eavesdropper on the local end from noticing Tor usage". For argument's sake, let's say that it's easier, at any particular time, to identify all VPN service entry servers than it is to identify all Tor bridges. And let's say that all Internet traffic is logged, and at metadata (such as IP addresses, times and so on) is retained indefinitely.

Although Tor bridges may well be "harder to monitor", it's arguable that that's only so transiently. That is, virtually all Tor bridges are sooner or later identified, and are replaced with new bridges. Is that correct?

Consider the implications for observers of my ISP-assigned IP address. If I were using Tor bridges, all of my prior traffic to a particular Tor bridge would be readily found once that bridge had been identified. And my Tor use would be apparent, given that all Tor bridges (I assume) lead to Tor.

If I were instead accessing Tor through a VPN service, identifying my Tor use would require either traffic analysis based on intercepts, or access to VPN connection logs. In contrast to the Tor-bridge scenario, identification of past Tor use would arguably become harder over time.

But of course, if I were truly concerned about hiding my Tor use, I would use both VPN services (in nested chains) and Tor bridges with Obfsproxy.

Regarding the "benefit to stringing together multiple VPNs that cannot be achieved by increasing the number of relays per circuit", the most obvious one is usability. On a good day, I can get at least 5 Mbps each way with a nested chain of three VPN services, with latency under 400 msec. With the standard Tor configuration, I'm lucky to get 2-3 Mbps each way, and latency is typically 1000-2000 msec.

Another benefit of using multiple VPNs instead of Tor bridges is that Tor traffic is further disassociated from users. Even if one could chain multiple Tor bridges, all prior Tor use would be revealed as soon as the initial bridge had been identified.

Identifying Tor users in VPN traffic, especially retrospectively, is arguably harder than searching based on known Tor bridges. Many people use VPN services, far more than use Tor. BitTorrent use and HD video streaming account for a substantial and increasing share of Internet throughput overall. And, in response to efforts to block "piracy", that traffic is increasingly migrating to VPN services.

First, let's consider that "the goal is to prevent an eavesdropper on the local end from noticing Tor usage". For argument's sake, let's say that it's easier, at any particular time, to identify all VPN service entry servers than it is to identify all Tor bridges. And let's say that all Internet traffic is logged, and at least metadata (such as IP addresses, times and so on) is retained indefinitely.

Although Tor bridges may well be "harder to monitor", it's arguable that that's only so transiently. That is, virtually all Tor bridges are sooner or later identified, and are replaced with new bridges. Is that correct?

Consider the implications for observers of my ISP-assigned IP address. If I were using Tor bridges, all of my prior traffic to a particular Tor bridge would be readily found once that bridge had been identified. And my Tor use would be apparent, given that all Tor bridges (I assume) lead to Tor.

If I were instead accessing Tor through a VPN service, identifying my Tor use would require either traffic analysis based on intercepts, or access to VPN connection logs. In contrast to the Tor-bridge scenario, identification of past Tor use would arguably become harder over time.

But of course, if I were truly concerned about hiding my Tor use, I would use both VPN services (in nested chains) and Tor bridges with Obfsproxy.

Second, regarding the "benefit to stringing together multiple VPNs that cannot be achieved by increasing the number of relays per circuit", the most obvious one is usability. On a good day, I can get at least 5 Mbps each way with a nested chain of three VPN services, with latency under 400 msec. With the standard Tor configuration, I'm lucky to get 2-3 Mbps each way, and latency is typically 1000-2000 msec.

Another benefit of using multiple VPNs instead of Tor bridges is that Tor traffic is further disassociated from users. Even if one could chain multiple Tor bridges, all prior Tor use would be revealed as soon as the initial bridge had been identified.

Identifying Tor users in VPN traffic, especially retrospectively, is arguably harder than searching based on known Tor bridges. Many people use VPN services, far more than use Tor. BitTorrent use and HD video streaming account for a substantial and increasing share of Internet throughput overall. And, in response to efforts to block "piracy", that traffic is increasingly migrating to VPN services.