[daemon]

I employ both server-side blocking (using dsbl.org and spamhaus.org) and client-side filtering (using annoyance-filter, a Bayesian Spam filter) and some Spam still gets through.

The server-side blocking checks four listings dsbl.org's primary and multi-hop open relay lists and Spamhaus's SBL and XBL lists. This imposes four DNS lookups for each legitimate e-mail I receive and fewer for caught Spam. Imagine how much overhead a similar system would add to NTL's processing of customer e-mail.

The client-side filtering is also a significant processing overhead. Each received e-mail has to be scanned for occurrences of words in the dictionary and evaluated as to whether it is legitimate e-mail or Spam. This type of processing would be costly to implement at the server and is best left to free CPU cycles of NTL's customers' machines.

So I hit Spammers hard with two techniques. At the server I implement a blocking mechanism that can inconvenience legitimate senders (if their ISP's servers get listed because a customer has been using their network to send Spam) by bouncing their e-mail. At the client I filter based on recent legitimate e-mail and Spam content. This requires periodic maintenance and continual monitoring.

In conclusion, no blocking or filtering solution is ideal. The real way to defeat Spam is through global legislation with severe penalties but as most Spammers tend towards the criminal, this is unlikely to produce tangible results.

I believe that NTL should employ server-side blocking of known Spam outfits through the use of e.g. Spamhaus lists but leave the content filtering to their customers. This will lead to users requiring not only anti-virus and firewall programs but also an e-mail content filtering program.

Regards,
Neil Darlow