I realize that there are many on this list who find data collection to
be pointless, but Kee Hinckley has shown this to be incorrect.
While it's always nice to see some numbers, I'm not sure what
these figures are actually measuring. For instance, yahoo will give a
positive reply for a random mailbox in RCPT TO, delaying rejection until
after DATA. I think the response after data will indicate if the account
doesn't exist or has been (recently) terminated.
I suspect hotmail will fail an address at RCPT TO even if the address is
associated with a registered account which has become "dormant". (please
correct me if I'm wrong on this one)
I suppose it might be possible to assume that these two biases will cancel
out, but that's quite a large assumption to make.
Vernon
Schryver's assertions were useless (even if correct) without hard
evidence, and Kee's data is insufficient without wider deployment.
Likewise, Vernon's followup that Kee is analyzing a different statement
than Vernon asserted is a legitimate concern. The data analysis
methodology should be publicly vetted to ensure that it is providing
meaningful and acurate data.
The primary issue is not deployment. We're not at that stage. One needs to
define what a "valid" address is and how to determine an address is
valid by that definition.
Once that's done, one needs to decide if the data collection is feasible
and useful. Then someone can write and review some scripts to analyse some
logs.
(Incidentally - which MTAs logs are you going to analyse? Sendmail? qmail?
exim? all of them? Standard / non-standard log formats?)
Data collection can indeed be pointless, particularly if the information
content is low.
All the same, it's good of KH to take the trouble to introduce some facts
into this discussion (seriously).
--
_______________________________________________
Asrg mailing list
Asrg(_at_)ietf(_dot_)org
https://www1.ietf.org/mailman/listinfo/asrg