> On Jun 28, 2026, at 8:18 AM, Alessandro Vesely <[email protected]> wrote: > > On 27/06/2026 20:11, John Levine wrote: >> It appears that Dan Mahoney <[email protected]> said: >> >>> In the top 10,000 u1m domains, I found about 30 with pct=0, including >>> several major corporate ones (ciscospark.com, several shopify domains) and >>> a NY state university (sunysb.edu) that have non-p=none policies. Is this >>> a small amount? Probably. But it's nonzero. > > > I'd be curious to know if any domain was found with both pct= and t=.
My limited grep shows just one (because grepping t= also matches pct=), I had to do a couple things to pull this out: 2026-06-25 partyslate.com 25 - y - - v=DMARC1; p=quarantine; pct=25; t=y; rua=mailto:[email protected],mailto:[email protected]; sp=none; np=reject; aspf=r; ...so they're aware, ostensibly, of the new specs, *and* set a non-0-or-100 value in pct=. Fascinating, captain. > Tolerating pct= for some time seems reasonable. However, new implementation > will likely drop it. Thus, support for pct= is bound to become more and more > deceptive. > > Perhaps we need a campaign, similar to what was done for the DNS Flag Day? My survey scripts were inspired by exactly that. (my dayjob has something to do with DNS Flag Day, I just don't post with my work address here). Unfortunately, the spec doesn't actually give a way to have the dmarc record return a contact record for an actual human with a clue (and the SOA record is just as broken). I strongly suspect many of these records were pasted in once by someone who used an online dmarc record generator who then set-it-and-forget-it. "Okay, we're compliant, job done!" -Dan _______________________________________________ dmarc mailing list -- [email protected] To unsubscribe send an email to [email protected]
