+1
Never assume that something doesn’t matter. It all depends on what the numbers
say. The selector and other fields are features that can potentially be used to
predict malicious or spammy behavior. I’m not aware of such a pattern yet for
selectors, but it may very well be that there will be
On 11 Oct 2017, at 6:31, John Stephenson wrote:
FWIW, maybe 5 years ago, we were required to send a legally mandated
bulk
email (deserving of delivery) and when reaching out to various inbox
providers, my contact at yahoo suggested that I send this effort
through an
existing domain, but a
> The statement was the selectors do not have an effect on reputation, but
that sometimes people believe they do because they changed the selector at
the same time they changed other things.
@Laura> that too; but there were clearly a possibility to say "no we don't
use s= at all", it hasn't been
On 2017-10-10 08:20, John R Levine wrote:
On Tue, 10 Oct 2017, David Hofstee wrote:
Didn't Google mention they wanted the age of the keys to count in the
spam
score?
I'll check but I would be surprised if it made much difference.
I rotate my keys every month, which seems to be more often
Yeah, I'd echo a bunch of what Vladimir said, selectors are useful for
different mail streams from the same domain, and we've played with using it
for reputation (as a tuple with domain). That said, we don't want to
discourage rotation, especially not anything crazy like requiring senders
to ramp
> On Oct 10, 2017, at 9:25 AM, Vladimir Dubrovin via mailop
> wrote:
>
>
> I can say nothing about Google, but selectors can really have indirect impact
> on the reputation.
>
> We do not bind reputation directly to objects like domains, selectors, etc
> and use dynamic
The *.gappssmtp.com default DKIM signatures for GSuite domains are
currently all a single key, which would seem to say that we don't currently
think that blending keys is a bad thing.
That isn't to say it can't change in the future if there becomes a need, of
course.
Brandon
On Mon, Oct 9, 2017
I can say nothing about Google, but selectors can really have indirect
impact on the reputation.
We do not bind reputation directly to objects like domains, selectors,
etc and use dynamic tuples instead (that is content of this tuple is
flexible to better match specific mailing type), and in
Didn't Google mention they wanted the age of the keys to count in the spam
score?
Old keys tend to have a longer timeframe to get stolen I guess. Maybe a
frequent key changes is an indicator of having good ops practices which
result in fewer incidents? Funny enough, I have only ever met one
Hi John,
> Do you?
In the way I tried to express it, yes.
Gmail recently said that the selector, or the change of the selector, can
have a role in their anti-spam and reputation system. Just because it's an
element of the email, and that it can indicate something.
It is not used for _reputation_
In article
ISPs might consider the change of s= or key as an element being part of
their reputation systems and metrics. The consequences are however unknown
but very most probably negligeable.
Considering that d= is the important stuff is right.
Having the same public key should not have any incidence, so
To my knowledge reputation is tied to the “d=“ domain. The value of the key
is irrelevant with regards to reputation.
Using shared or unique key pairs is a balance between managebility and
security.
Maarten
On Mon, 9 Oct 2017 at 19:06, Alexander Burch
wrote:
> Do
Do major ISP check the public DKIM key for reputation metrics?
For example, an ESP might use domain1.com, domain2.com and domain3.com to
sign messages for different reputation pools.
If these domains all have the same public DKIM key will this "blend" their
reputations in any way, namely at
14 matches
Mail list logo