MAPS sucks

[fanf]

We use the JANET subscription to the MAPS RBL+. It is a frequent source of false positives which causes a great deal of irritation.

• Their RSS (relay spam stopper, i.e. list of open relays) does not have any re-testing nor any expiration of old entries, so is full of shockingly stale entries.
• Their DUL (dynamic user list, i.e. list of home computer IP addresses) has many errors. They do not track re-assignments of address space which caused them to list gmail's servers earlier this year.
• Spamhaus's ZEN list is just as effective but much more trouble-free. Why is JANET paying for the RBL+ and not ZEN?
• They do not accept corrections from their paying customers. WTF!

Sadly I can't easily ditch the RBL+ since using both it and ZEN adds 20-25% to our block rate compared to either alone, so performance would suffer horribly. Also, I don't want to get into the game of maintaining my own blacklist and/or whitelist - a time sink I do not need.

Comments

[mas90.livejournal.com]

Are RBL+ and ZEN the only two DNSBLs you use, or are there others?

Until just now I was using Spamhaus SBL+XBL on its own, which I've just replaced with ZEN. I've thought for a while that there might be other blacklists out there that it would be worth using in addition. (But based on your comments I think I'll avoid the RBL+...)

On a slightly tangential note, how harmful would you say it is to do sender callout verification for all sender domains, rather than just a handful on which this is known to be effective?

[fanf]

Sender callback verification suffers from a terrible false positive rate, especially for email from web sites - we had lots of problems with various Web Of Science mirrors, for example. It's also becoming less and less effective - http://fanf.livejournal.com/70432.html?thread=263968

I like sender verification callouts

[(Anonymous)]

And, if I get a false positive report, I report it to the sending site. Generally, they fix the problem, sometimes with my help.

[bens_dad]

At work, I use sender-verification callbacks on all (none whitelisted) domains, admittedly after fanf (my ISP and MX primary) has filtered everything for me.
I accept that it has significant downsides (I also fore-saw jmason's point that it encourages spammers to use read sender address) and more operator intervention than anything else I do, but the fundamental point is: why accept an email that you can't reply to ?

Exim's caching means that I'm not doubling or trebling the amount of email flying around, which would make me more than uneasy.