Quoting David Honig <honig@sprynet.com>:
At 07:17 AM 9/16/01 +0000, Ryan Lackey wrote:
My primary concern is limiting the usability of these services for HavenCo AUP violations; specifically spam and spam-mailbox. A per-message charge or decrement would likely accomplish this, along with either payment or proof-of-work.
You could use human- and not machine- readable text that the user has to enter ---many free-hosting sites do this to prevent bots. That won't prevent emailing spam to a list, but will prevent mass mailings.
That's exactly what I meant by "proof of work". This topic has been discussed on p2p lists (mainly by the freenet people), as well as other places. Fundamentally, though, I don't like the idea of providing a service based on "burning" human or computer time; this is deoptimizing best-case performance in order to improve worst-case. And I *do* want automated programs to be able to use the remailer network, blacknet style. Other proof of work systems are entirely automated; such as adam back's hashcash. Already on the search engine sites which require "read these numbers" to enter a URL, there are other sites which advertise "all new codes, all the time". The only viable long-term solution I see for most of these things is charging per-transaction for messages. Remailers are exactly the kind of application where "postage" could first be applied. The problems seem twofold: 1) Remailer operators are exposed to new and additional legal threat if they accept payment for service (right?) 2) It's much easier to trace payment instruments than remailer messages. While a blinded cash system might only leak "this person is using the remailer payment system", and not provide linkability to given messages, it does complicate things substantially, and provides the potential for a lot of tracing. It would be an interesting experiment to simultaneously offer a "proof of human work" system, a "proof of computer work" system, and a pay per message system, both bulk and per-message, and see which is most popular. Given the existence of ~20 public free remailers, it seems unlikely people would use them, except perhaps if they had a good web-based UI, integration into common MUAs, etc. This would have the effect of monetizing cpu time and human-who-can-read-numbers time, which would be interesting. (sort of like the "get paid to surf the web" stuff). What would be really cool would be replacing "decipher these numbers" with some kind of task which produced actual value, but could be evaluated either by a machine or a horde of other humans; "prove this theorem", "sort these files", etc. I wonder if any other remailer operators be interested in such a system. It becomes more interesting as the cost of being a remailer increases, of course. -- Ryan Lackey [RL7618 RL5931-RIPE] ryan@havenco.com CTO and Co-founder, HavenCo Ltd. +44 7970 633 277 the free world just milliseconds away http://www.havenco.com/ OpenPGP 4096: B8B8 3D95 F940 9760 C64B DE90 07AD BE07 D2E0 301F