Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
Tom Metro asked: > So your theory is that AOL is examining Received headers and rejecting > your messages based on what it finds. What specifically would it be > looking for? It appears to be looking at one or more of the Received headers beyond the one its own incoming MTA software generates when MailHop's relay connects to it. AOL keeps its spam-control filter algorithms proprietary, for obvious reasons, so I can only point out specific instances of messages getting blocked or dropped. > the vast majority of ISPs have customers using dynamic IP > ranges connecting to a mail relay. These dynamic IP addresses almost > always appear in some of the headers. I suspect that I am not alone in this problem, though an obvious problem is that probably fewer than 1% of ISP customers fit into your description: most of us punt on having our mail servers and just use the ISP's webmail service. Those of us who have sysadmin access to dedicated corporate service can (and probably should) just use the company servers to relay through a known-reputable IP address. > How is DynDNS's MailHop different? It isn't. It takes whatever message I send, adds on its own Received header (plus a couple other customer-identifying headers), and sends it along. Two of those headers contain my Comcast dynamic IP address. > If the entire message is being subjected to Bayesian filtering, it isn't > inconceivable that subtle differences could trigger a spam threshold. > Perhaps with other ISPs the filter becomes accustomed to what netblocks > the embedded IP addresses come from, yet with MailHop they're all over > the map? But this seems unlikely. So far as I can tell, my problem of the past few months is mainly AOL. My correspondents who use other services haven't reported dropped messages and I have never seen a bounceback message for any known recipient until just today. Sure, AOL can and probably does use Bayesian among its strategies but unless any of us works in their email tools division we can only speculate. Few Bayesian techniques will trip up lengthy "hi grandma" messages of the sort I've sent unsuccessfully in recent months. > Have you discussed this with DynDNS? Waiting for a response. They could possibly be swamped with customer complaints if this is a sitewide issue for them. My options for fixing this are pretty much limited to setting up a GMail account to use for my few acquaintances who remain on AOL, at least that's all I can think of since the last time we held this discusson on BLU when folks pointed out that the inexpensive cloud-based services tend not to provide reputable static IPs. -rich
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |