Earlier quoted context omitted.
I used to run my own one-user email system. The hosting alone cost $240 a year. And I had to spend a lot of time configuring exim4 (and spamassassin and clam-av), figuring out which email indexer to use, setting up offlineimap, running cron jobs to archive mailing lists, coercing a friend to run a secondary MX for me, etc., etc. I didn't enjoy it all. And no matter what I did to reduce spam, about 10 messages a day a…
Not to mention that you have to pay to be 'recognised' so that your emails aren't autobounced by some enterprises. I have a friend that runs his own domain email for himself and friends and it's surprising to hear just how complex it can be just to get mail accepted. I can't recall if it was a whitelist or a certificate or what, but I do remember at one stage he was happy because he found a way around having to pay a…
Last year I moved and my IP address changed (ISP didn't). I expected to lose the "reputation" the previous one had built up. But I did not have any problems whatsoever. I do hear rumors like yours every now and again but I have a hard time believing them.
I think that large email providers, the likes of Gmail/Hotmail, actually look at the email addresses / domains / servers that their clients send email to. Then they assign trust to those tokens. If you're a large provider, you can do many things with the data you get from your own customer's behaviour. How about looking at accounts that have been in use for some time, seen regular web interface action, and send email to other @hotmail/gmail accounts that actually get read and not flagged spam? If those accounts send mail to my mailserver, then my mail server / domain must have something good going for it. Well that's what I would do if I were running a huge setup anyway...
As for incoming spam: I'm using various postfix tricks, greylisting, and dspam. I have no problems. I should write a howto ;-)