3 ms·

Self-hosting your e-mail is really not much better, from a security point of view, than letting gmail have your e-mail. Google has some of the best security peo

by x1798DE 12y ago

Self-hosting your e-mail is really not much better, from a security point of view, than letting gmail have your e-mail. Google has some of the best security people in the world, if your mail is safe anywhere, it's safe with Google. The problem, of course, is that Google is definitely going to try to read your mail, use and retain your private data indefinitely. The problem is that if you host your own e-mail server, you dramatically increase the chances of some unpatched flaw or tiny messed up configuration file causing the Russian mafia to get access to everything on your e-mail server. You probably don't even particularly decrease the risk of the NSA getting your e-mails. This is a problem that could be fixed by new models for e-mail that utilize encryption so that you inherently don't need to trust third parties in order to make use of their services. Avoiding third parties entirely is very likely impractical and unwise.

__david__ 12y ago
I've seen this argument before, and, while it's true to some extent, it seems like FUD to me. The same exact argument could be used for every single server you host yourself. But how many people here have an AWS machine running HTTP and SSH? How many of them have been compromised by the Russian mafia? People are so scared of email, and I just don't understand it.
Tepix 12y ago
I disagree. Google has no protection against NSLs. Check out sovereign on github: https://github.com/al3x/sovereign https://github.com/al3x/sovereign
- x1798DE 12y ago
  Correct that Google has no protection against NSLs, but that's like saying that you should always drive your car because planes are common targets for terrorism. I'm deeply troubled by the activities of the NSA et al, but realistically you are far, far more likely to be targeted for actual harm by malicious hackers than by the NSA. Even if you are targeted by the NSA, there's a better than average chance that they'll be able to get the data you care about anyway - consider how often you send an e-mail to something that doesn't end up on a server that's vulnerable to NSLs - they can compromise you on either end of the conversation, and that's not even considering the fact that governments are out there buying zero day exploits on the open market. My point is that taking an inherently insecure (at least with respect to data privacy) protocol like the ones we use for e-mail now and putting it on your own server (and thus taking responsibility for patching, avoiding zero days, etc) is not an answer to the problem of data privacy. The way forward in my opinion is the development of communication protocols which by their very nature are trustless. If you're just interested in protecting content, then if you're using end-to-end encryption, you could use LegitimateBankSiteNumberOne.ru as your e-mail provider and it wouldn't matter. The protocols for doing this are already well-known, it's just a matter of adopting them.
wpietri 12y ago
Depends on the threat. If you host your own email, there is a whole class of legal intrusion (persuasive cops, warrants, national security letters) that you get to hear about if they want your archives or stuff that's encrypted on the wire. The same applies to corrupting employees; I'm going to know if somebody bribes my sysadmin for access, because that's me. There's also some benefit in avoiding being part of a monoculture. Big mail providers are an enormously interesting target for both spies and criminals. Breaking into random quirky personal servers, though, has a very different cost/benefit ratio.
- x1798DE 12y ago
  I heartily agree about avoiding being part of a monoculture, there are definite tradeoffs. The weirder your setup, the less likely you are to be swept up in anything bulk, but the more vulnerable you are to anything targeted. Unfortunately, it's pretty common for some quirky little open-source project to start out relying on obscurity, then become widely adopted, putting you in the absolute worst case scenario, since suddenly you're running high profile software that's never been "battle tested".

by x1798DE 12y ago

__david__ 12y ago

I've seen this argument before, and, while it's true to some extent, it seems like FUD to me. The same exact argument could be used for every single server you host yourself. But how many people here have an AWS machine running HTTP and SSH? How many of them have been compromised by the Russian mafia? People are so scared of email, and I just don't understand it.

Tepix 12y ago

I disagree. Google has no protection against NSLs. Check out sovereign on github: https://github.com/al3x/sovereign https://github.com/al3x/sovereign

x1798DE 12y ago
Correct that Google has no protection against NSLs, but that's like saying that you should always drive your car because planes are common targets for terrorism. I'm deeply troubled by the activities of the NSA et al, but realistically you are far, far more likely to be targeted for actual harm by malicious hackers than by the NSA. Even if you are targeted by the NSA, there's a better than average chance that they'll be able to get the data you care about anyway - consider how often you send an e-mail to something that doesn't end up on a server that's vulnerable to NSLs - they can compromise you on either end of the conversation, and that's not even considering the fact that governments are out there buying zero day exploits on the open market. My point is that taking an inherently insecure (at least with respect to data privacy) protocol like the ones we use for e-mail now and putting it on your own server (and thus taking responsibility for patching, avoiding zero days, etc) is not an answer to the problem of data privacy. The way forward in my opinion is the development of communication protocols which by their very nature are trustless. If you're just interested in protecting content, then if you're using end-to-end encryption, you could use LegitimateBankSiteNumberOne.ru as your e-mail provider and it wouldn't matter. The protocols for doing this are already well-known, it's just a matter of adopting them.

wpietri 12y ago

Depends on the threat. If you host your own email, there is a whole class of legal intrusion (persuasive cops, warrants, national security letters) that you get to hear about if they want your archives or stuff that's encrypted on the wire. The same applies to corrupting employees; I'm going to know if somebody bribes my sysadmin for access, because that's me. There's also some benefit in avoiding being part of a monoculture. Big mail providers are an enormously interesting target for both spies and criminals. Breaking into random quirky personal servers, though, has a very different cost/benefit ratio.

x1798DE 12y ago
I heartily agree about avoiding being part of a monoculture, there are definite tradeoffs. The weirder your setup, the less likely you are to be swept up in anything bulk, but the more vulnerable you are to anything targeted. Unfortunately, it's pretty common for some quirky little open-source project to start out relying on obscurity, then become widely adopted, putting you in the absolute worst case scenario, since suddenly you're running high profile software that's never been "battle tested".