I'll add W3C's Related Website Sets / First-Party Sets spec to the
list of failed proposals in this area:
https://github.com/WICG/first-party-sets.

--Ben

On Mon, Oct 26, 2026 at 7:09 PM Tom Barrett
<[email protected]> wrote:
>
> All, The recent discussion of draft-manelidis-dnsop-domain-affiliation (DAC), 
> including John Levine's pointer to RDBD and the earlier DBOUND work, led me 
> to look at our own draft again against all three. The draft is 
> draft-barrett-dnsop-domain-set-00
> 
> All,
>
> The recent discussion of draft-manelidis-dnsop-domain-affiliation (DAC), 
> including John Levine's pointer to RDBD and the earlier DBOUND work, led me 
> to look at our own draft again against all three. The draft is 
> draft-barrett-dnsop-domain-set-00 
> (https://datatracker.ietf.org/doc/draft-barrett-dnsop-domain-set/). It 
> addresses the case John Klensin raised in the DAC thread, impersonation 
> look-alikes being mistaken for legitimate names. Having looked again, we 
> think it is distinct from DAC, RDBD and DBOUND in the following ways.
>
> What domain-set does: an organization publishes the set of registrable 
> domains it wants recognized as its own, in any TLD, with relation types 
> (primary, operating, redirect, defensive). Each link counts as confirmed only 
> when both domains name each other and both directions are retrieved over 
> DNSSEC, or over HTTPS from the publisher's own domain. A consumer can start 
> from one name it trusts and retrieve the organization's whole set.
>
> This is how domain-set differs from the others:
>
> - From DAC: domain-set is for discovery, not a pairwise check. A consumer 
> gets the organization's full published set, not a yes/no answer about one 
> claimed link. The organization chooses what to include. Defensive 
> registrations such as misspellings usually don't resolve and can simply be 
> left out, so publishing a set doesn't mean disclosing a whole portfolio.
>
> - From RDBD: no new RR types and no signature scheme, only TXT at an 
> underscored name. The linking mechanism is close to the simpler design John 
> Levine suggested on the dbound list in 2019, in place of RDBD's signatures: 
> each domain publishes a TXT record under an underscore label pointing at the 
> other. Links are never one-way: an assertion that the other side doesn't 
> return is reported only as "asserted" and must not be treated as evidence. 
> And a link has one defined meaning, common operation by the same 
> organization, not an unspecified relationship.
>
> - From DBOUND: domain-set doesn't deal with policy boundaries within a 
> namespace, the cut points the PSL encodes for cookies, certificate wildcards 
> and DMARC. It neither replaces nor consults the PSL. It groups unrelated 
> sibling registrable domains under one organization, and only for operators 
> who choose to publish.
>
> - From all three: domain-set is designed to carry independent verification. 
> It works for any organization with more than one public-facing domain, but it 
> is most useful when one member has been verified by a third party, such as a 
> registry that checks eligibility before issuing names. Early interest has 
> come mostly from organizations with names in .bank, .pharmacy, .brand TLDs 
> and ccTLDs that verify their registrants. These organizations often run large 
> portfolios in .com and other legacy TLDs. domain-set doesn't perform that 
> verification. It carries the verified identity, over confirmed links, to the 
> organization's other domains, so a consumer can tell a linked domain from a 
> look-alike with no link.
>
> - Consumers: browsers and, increasingly, the AI assistants and agents that 
> people ask whether a site is the real one. They need a machine-readable 
> answer that the organization itself publishes.
>
> In the next revision of domain-set we'll consider expanding the related-work 
> section to cite DAC, RDBD and DBOUND and set out these differences in more 
> detail. We'll post to the list when it's available. In the meantime, comments 
> here are welcome.
>
> Tom Barrett
> EnCirca, Inc.
>
> --
> Schedule a call with me.
> https://calendar.app.google/BrESmtfgCG4tutSu7
>
> Thomas Barrett
> President
> EnCirca, Inc
> +1.781.942.9975 (office)
> 400 W. Cummings Park, Suite 1725
> Woburn, MA 01801 USA
> _______________________________________________
> DNSOP mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
_______________________________________________
DNSOP mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to