restricting servers: best practices

I am running into the issue that so very many SMTP servers are misconfigured
that I am actually losing customers because I block mail from idiots who
cannot follow the RFC.
Postfix is merely a spamfiltering system for me.

I block incoming mail with:
------snip----------------
smtpd_client_restrictions =
permit_mynetworks,
hash:/etc/postfix/ip_access,
reject_unauth_pipelining,
reject_unknown_client,
reject_rbl_client relays.ordb.org

smtpd_helo_restrictions =
permit_mynetworks,
check_helo_access hash:/etc/postfix/helo_access,
reject_invalid_hostname,
reject_unknown_hostname,
reject_non_fqdn_hostname,
reject_unauth_pipelining
----------snip----------

I am rejecting over 1m connections a day, processing 36k and delivering 7k
(yes, I get HUGE spam).
Of the 7k messages, I would say 65% are false positives.

Any advice/recommendations that could loosen this up but not open the
floodgates? There are cities, government agencies, etc that are connecting
with systems that don't use fqdn or have a reverse lookup for the name they
do use, or don't have reverse lookup for their Ips. When I find out about a
problem from a user, I try contacting the sending admin, but with
bureaucracy it often does no good.

Brian Andrus