Now that we’re into the New Year and deadline for public comment on the proposed new .CA whois policy nears and now that my term as a CIRA Director enters its home stretch, I wanted to take some time to elaborate further on my Unsanctioned Whois Concepts post from long ago and revise it somewhat.

Last summer when the Whois Policy revision came into scope I spent some time trying to ponder the dilemma that is “whois”.

I realized that the core issue of our problems is that we only have “one-way authentication”, especially in the .CA ccTLD. Under .CA every single registration is actually “vetted” by the registry and in the gTLDs there is a growing sentiment toward this given the “post 9/11” mentality and it shows up in initiatives like FOISA.

What happens in this situation is we spend a lot of effort validating what goes into the database but near zero effort controlling what gets retrieved out. This is where the problems of machine harvesting and all its attendant consequences come from.

In the real world a lot of this is mitigated by controlling who queries the data: Every time somebody accesses my credit rating a line item shows up to that effect. Every once in awhile I run my own credit check just to see who’s been running credit checks on me.

On the other hand, if I want to drive a car, I have to have a license plate and that plate is visible for all to see and it makes me accountable for pretty well anything that car does.

We have the technology to build similar real world controls and balances into the Whois database now. Last summer I wrote a paper called Balancing Privacy & Accountability within .CA Whois Reform [PDF] and in it I posited the concept of what I called “Whois/Who”.

It basically described a system where we closed the loop on both sides of the transaction, by matching the validation we already do with .CA registrants on input with validating and authenticating “Querying Parties” (QP’s) . By providing accountability in both directions you could provide digests to registrants describing who looked up their records and why. You could also trace any given email sent via a whois record lookup to the party who queried the record by using one-time email addresses and a forwarder at the Registry (the .CA registry is already thick and carries out comparable functions today).

The idea went nowhere. Maybe it’s a bad one. It seems the will of the Board at this present juncture is to fulfill the privacy requirements under PIPEDA and is happy to stop there. As a Registrar and the guy who mans the Abuse desk around easyDNS, I’m not thrilled with the proposed new system which essentially allows for anonymous registrations for “individuals”. As a domain registrant who hates getting spam and DRoC “invoices” I will probably take advantage of it for my personal .CA names.

Shortly before the holidays I came across a thread on Slashdot about i-names and it brought to mind an earlier exposure I had to sxip, both of which are Extensible Identity Protocols and both implement the framework within which all of the ideas I described under Whois/Who could be implemented. Sxip goes one further in allowing identity data to be stored anywhere, users could run their own homesites and store their own data and create their own dissemination policies.

I believe these types of protocols are where the future of Whois will (hopefully) end up. Unfortunately CIRA isn’t availing itself of the opportunity to get really ambitious with Whois Reform while it has it on the agenda this time, but in CIRA’s defense, the exigencies of bringing the entire data collection and dissemination issues of the Whois database under PIPEDA compliance was the main concern.