Companies sensible to effective delivery of email to all free email services may have noticed problems with deliveries to Hotmail addresses. Despite the SMTP dialog ending with a successful "250" return code, recipients don't see the message. In their Guidelines, MSN require thorough compliance with IETF standards. However, it seems they have their own interpretation about provisions for Delivery Status Notifications, a.k.a. bounces, that servers must send after they have accepted responsibility for delivering the message. According to the Windows Live Hotmail Fact Sheet: "...messages delivered to the junk mail folder will not automatically show or render message content."
Dropping messages without notice is cool if the server is fairly certain about the nature of the message. Most servers drop viruses that way, as it's the only way to effectively stop their propagation. For spam, the best practice is to block messages at an early stage in the SMTP dialog by responding with a permanent failure code. The connecting MTA, normally the sender's outgoing SMTP server, should then issue a bounce. DNS Black Lists (DNSBL) are an ISP's best friends when it comes to discriminate connecting clients. Let me remark that the "DNS" acronym here refers just to the lookup protocol being used. DNSBLs' responses are based on no authoritative delegation.
Hotmail doesn't say what DNSBLs they consult, if any. They are not going to open their filtering technique, out of concern the disclosures would allow spammers to bypass the defenses. It turns out that an ISP may gain the privilege to send mail to their addresses after subscribing to Smart Network Data Services. Microsoft suggests one can use that service, as well as other services, some of which require payments, neither of which guarantees a result. With more than 280 million active accounts (according to the mentioned fact sheet) Microsoft can establish the policy they want. They are not going to own email that way: just a piece of it.
On the opposite, tiny to medium ISPs have a hard time blocking mail from large ISPs' hosts that happen to get listed, e.g., in Spamhaus SBL-XBL. Telecom Italia, for one, is a major connection provider in Italy who regularly gets blacklisted for not having an effective anti-spam policy. Users can't stand not receiving messages and don't want to know why. Therefore, small ISPs whitelist the offending IP. Again, the result is that various senders are happy with their email addresses, and don't realize they only work inside a small piece of the Internet.
That tendency toward fragmentation can be turned the other way around.
Perhaps, we've dismissed proposals like SPF somewhat hurriedly. After all, SPF syntax allows to create domain-specific white lists or to endorse existing DNSBLs. We didn't really explore the possibilities that SPF has to offer before dismissing it. Regional clustering is the easiest case because of the way IPs are assigned by IANA. However, I found out that many domain administrators didn't even think that, using CIDR notations as "?ip4:126.96.36.199/7", a whole Region of IPs can fit within a single SPF line. Why would an Italian domain administrator allow sending from any RIPE address? Because it is much less practical for him to prosecute spammers in the US or China. And then the whole Europe is just a piece of the Internet, much less than "~all". Divide et impera. Conquer or be conquered.
By Alessandro Vesely, Tiny ISP and freelance programmer
|Cybersquatting||Policy & Regulation|
|DNS Security||Registry Services|
|IP Addressing||White Space|
Afilias - Mobile & Web Services
Minds + Machines