On 7/26/2020 12:12 PM, Dave Crocker wrote:>
My comment was about observable behavior, not philosophical anything.
The issue isn't about was is available to users, but what they
actually do. Behavior.
Dave, for a number of years of practice, depending in the system or
service, users have been provided with trust-related decisions . Do
you need real examples?
Gmail moves spam into Spam box.
If you user do not look into their spam box, their inaction is used to
give spam weight to false positives. If the user is looking for
something, finds it in the spam box, it will see something like this:
"What is this message in spam? (specific or vague reason)"
"REPORT NOT SPAM"
Clicking "REPORT NOT SPAM" it doesn't guarantee it will be moved to an
area you can see it. It's been reported as lost, you just can't find
it. So it is sent again. The next time another message of the same
type is received, it will also be placed in the spam box but this time
it says:
"What is this message in spam? it is similar to messages
identified as spam in past"
"REPORT NOT SPAM"
Clicking "REPORT NOT SPAM" it doesn't guarantee the next transaction
will not be moved into the spam box the next time. There are
properly other behaviors the user must make before the message source
is "white listed."
I am not sure if we can model this behavior based on the different
ways it may be done, but let me outline what I have observed with users:
- May never check spam box
- May check if expecting a message unexpectedly delayed.
- Ignore it, it gets weighted as spam.
- Click Report not Spam. Cross finger you can see it in-box.
- Next similar messages are still moved to spam box
It is the last behavior that is concerning because when you are
dealing with layman users who may see an message in a spam box but
leave it there, it appears GMAIL will keep adding weight to it as a
spam, negatively impacting the source identities of the mail.
In the end, the positive of SPAM box are lost because now you are
forcing the user to scrutinize the Spam Box opening the door to
reading and accidental clicks. if REPORT NOT SPAM does not clear it
up, there is even more confusion for the purpose of that button or any
similar AI learning logic that may be employed.
-- Hector Santos,
https://secure.santronics.com
https://twitter.com/hectorsantos
_______________________________________________
dmarc mailing list
dmarc@ietf.org
https://www.ietf.org/mailman/listinfo/dmarc
