I think the question you should really ask here is what your friend means by "working". In a fully trusted and firewalled network, installing a mail server is

I think the question you should really ask here is what your friend means by "working". In a fully trusted and firewalled network, installing a mail server is probably as simple as `sudo apt-get install <mail server>` and accepting whatever debconf prompts you with. But if you want an Internet-connected system and you want to send mail to users on outlook.com and gmail.com, then it's a different beast entirely, for one simple reason: spam. Two concerns: incoming spam, and outgoing spam. Dealing with spam is an arms race; in the 90s, a simple bayesian filter on the contents of the email (such as POPFile) would be sufficient for most users. Spammers have gotten more sophisticated, and so tuning the rules so that you're not inundated with spam takes quite a bit of time. On the flip side, other people have pretty sophisticated email filters that look at things like email signatures (DKIM), DNS records (reverse DNS, SPF), whether you're running a server off your home ISP connection (DNSBL), and all sorts of other shenanigans. Configuring a mail server to pass all these checks is supposed to be hard enough that you'd only want to do it once -- if it wasn't, then the spammers would be able to do it quickly and users would be getting a lot more spam because every time you block one server, the spammers would just set up another one.