3 ms·

I ran a set up similar to this for many years. It's not that hard, for those with a little unix experience. As moxie mentions, email is very forgiving—you have

by cwp 13y ago

I ran a set up similar to this for many years. It's not that hard, for those with a little unix experience. As moxie mentions, email is very forgiving—you have to break it badly and leave it broken for a long time before you start to lose messages. What eventually drove me to GMail was spam. I tried a bunch of different filters, and never found one with good-enough accuracy. Finally I decided that the independence and privacy wasn't worth the time I spent fiddling with filters and dealing with misclassified messages. As far as I can tell, Gmail is 100% accurate. Problem solved.

cabalamat 13y ago
> As far as I can tell, Gmail is 100% accurate. My experience is I've had a few false negatives and false positives. Gmail is still very good though, and any replacement, if it is to be widely used, needs to solve the spam problem. If each message had to be separately encrypted for each receiver, would this add much to a spammer's costs? I'm guessing it would, but not by enough to make spam uneconomic. A better solution might be to require that the first email someone sends someone else, unless they've been OK'd by a third party, contains bitcoins to the value of $0.01.
- comex 13y ago
  Fun fact: Bitcoin's proof-of-work algorithm is based on Hashcash, which was an anti-spam measure along those lines: a message contains a partial hash brute-force that takes the sender about a second of CPU time to compute (and depends on the recipient). But it didn't take off, and is problematic when spammers have vast amounts of CPU resources in the form of botnets.
bifrost 13y ago
> As far as I can tell, Gmail is 100% accurate. I can confirm that that is not correct. Google regularly flags internal ticket email as spam for one of my larger customers (mail that is sent within GApps). I also see about a 10-20% false positive rate on our various mailings (account setups, billing notices, etc).

by cwp 13y ago

cabalamat 13y ago

> As far as I can tell, Gmail is 100% accurate. My experience is I've had a few false negatives and false positives. Gmail is still very good though, and any replacement, if it is to be widely used, needs to solve the spam problem. If each message had to be separately encrypted for each receiver, would this add much to a spammer's costs? I'm guessing it would, but not by enough to make spam uneconomic. A better solution might be to require that the first email someone sends someone else, unless they've been OK'd by a third party, contains bitcoins to the value of $0.01.

comex 13y ago
Fun fact: Bitcoin's proof-of-work algorithm is based on Hashcash, which was an anti-spam measure along those lines: a message contains a partial hash brute-force that takes the sender about a second of CPU time to compute (and depends on the recipient). But it didn't take off, and is problematic when spammers have vast amounts of CPU resources in the form of botnets.

bifrost 13y ago

> As far as I can tell, Gmail is 100% accurate. I can confirm that that is not correct. Google regularly flags internal ticket email as spam for one of my larger customers (mail that is sent within GApps). I also see about a 10-20% false positive rate on our various mailings (account setups, billing notices, etc).