Adam Back wrote: > Could you explain what you mean in 4 and 8? I probably should explain all of them to reduce unspoken assumptions. > Not sure of terminology -- which of those corresponds to my "include > an extra header in email which authenticates the update. Users send > around the latest update in emails they send, so that clients learn in > a p2p fashion from other people, from new clients (which would start > with latest version) and so on", and what the other type of broadcast > entails? that's actually type 4b. I was floating the concept of type 4a in the days of izzy and I wasn't sure if it got lost in the chatter. sounds like you came up with the same idea (and I really don't care who was first). Let's make sure it's a good one and go forward with it > Do you find the above "good enough" -- ie simple, generally likely to > be effective. don't know yet. Haven't had any time/energy to run a model. > On Tue, Apr 20, 2004 at 05:03:05PM -0400, Eric S. Johansson wrote: > >>maybe we should create a small page about the trade-offs of the four >>different techniques we know about today for managing postage? >> >>1) challenge response (nobody's favorite) you send me some amount of postage that is less than what I need (down to 0) and I send you a postage to notice asking for the new higher amount. Benefits: most direct propagation of the postage amounts. Is really annoying if you aren't using camram at all and see these postage due notices. Downsides: Joe jobbing. Spammers can send e-mail with valid return address and no postage which floods valid address with postage due notices. Worse yet, they spread the return addresses out so that there is a chance that people will just fill in the postage due notice and allow Spam through. >>2) increase by the clock as the days tick by, increase postage according to some formula. benefits: easy to implement, completely decentralized downsides: cannot adapt to attacks from spammers or changes in Moore's Law. >>4) passive broadcast (my favorite) passive broadcast attaches current local postage rates to each mail message as it is being sent. As events trigger postage increases, subsequent mail messages reflect the higher postage rate in the rate tag. benefits: the passive technique will notify more slowly and it seems that it will be a more stable/less disruptive process for propagating rate increases. downsides: the rate of propagation of information is only as fast as a given user sends e-mail to other users. Its entire possible that the flow of notifications of rate increases could slow or stall based on the users activity level. >>8) active broadcast (i.e. full and partial injection) active broadcasting is like passive broadcasting by each message has a header are indicating the current rate for the message originator. But unlike the passive broadcasting technique, the active technique sends the shortest message possible to all of the known camram users to inform them that a rate increase event happened. In the early days, this would probably be a good thing for camram because there won't be that many users but after a certain point, spammers could bring down a significant chunk of the e-mail ecology by sending out a small number of spam messages with stamps. After receipt of the spam, there would be a flurry of messages communicating the postage rate increase and much amusement would ensue (NOT!) therefore, a partial injection in the form of sending rate increase notices to a random selection of known camram users might be a reasonable compromise between the quick response of the full injection and the slow response of passive broadcast. benefits: fast response downsides: easy triggers of e-mail DOS ---eric