Logo

051040043 - EPDP-Phase 2 April Team Call
Terri Agnew
39:19
Members - reminder to please select all panelists and attendees for chat option
Milton Mueller (NCSG)
42:17
not sure what is the point of the central gateway issuing a recommendation if the CP actually makes the decision
Mark Svancarek (BC)
45:56
CGM making a recommendation is the first step toward evolving CGM to perform automated decisions
Mark Svancarek (BC)
46:54
There are other benefits, such as the transaction-evaluating function of our Dynamics Fraud Prevention service, though
Brian King (IPC)
48:14
Good summary, Becky.
Brian King (IPC)
48:17
Thanks
Stephanie Perrin (NCSG)
50:57
please speak up
Milton Mueller (NCSG)
53:46
Yes, Mark, the recommendations are obviously a trojan for automated decision making by th CGM, so since we have decided on a hybrid model and NOT on a centralized model why are we bothering with it?
Hadia Elminiawi (ALAC)
54:01
So Chris can you provide examples
Berry Cobb
54:42
It may be helpful to create an example for each of the 4 areas mentioned in the definition.
Mark Svancarek (BC)
55:02
Milton, I think we already know that many CPs are not really in a position to evaluate lawfulness without assistance,
Mark Svancarek (BC)
55:32
And we have not decided that the evolution mechanism cannot ultimately include centralized decision making
Hadia Elminiawi (ALAC)
55:39
@Berry makes sense to provide an example or two for each category
Milton Mueller (NCSG)
58:53
Mark, funny I don’t see the Cps clamoring for help.
Milton Mueller (NCSG)
59:07
But it’s soooo nice of you to be solicitous of their needs
Margie Milam (BC)
01:01:13
+1 Alan G
Brian King (IPC)
01:01:17
that's a good distinction, Alan G
Mark Svancarek (BC)
01:01:40
Milton, are you absolutely opposed toward evolution toward a centralized decider, forever?
Milton Mueller (NCSG)
01:02:02
Yes
Mark Svancarek (BC)
01:02:15
dang!
Milton Mueller (NCSG)
01:02:23
LOL
Matt Serlin (RrSG)
01:02:36
I believe Alan G is correct in that a CP MAY decide to automate on their own accord
Marc Anderson (RySG)
01:02:57
+1 to Alan and Matt
Milton Mueller (NCSG)
01:03:07
We look forward to litigation around these brave CPs
Alan Greenberg (ALAC)
01:04:11
Why would we rule out the possibility of having staff involved with SSAD decisions?
Alan Greenberg (ALAC)
01:05:23
All the more so based on the B&B memo we just heard about.
Berry Cobb
01:06:31
Per Caitlins intro comment, maybe we return back to this one?
Margie Milam (BC)
01:07:30
+1 Marc A
Milton Mueller (NCSG)
01:08:30
we are NOT talking about an automated decision
Milton Mueller (NCSG)
01:08:56
if the balancing test is required by law, are you saying we shouldn’t do it, Hadia?
Hadia Elminiawi (ALAC)
01:09:38
machines can do algorithms for sure
Hadia Elminiawi (ALAC)
01:09:53
I mean can do a balancing test for sure
Alan Woods (RYSG)
01:10:19
How Hadia…. how can they possibly do that?
Alan Woods (RYSG)
01:10:34
How can a machine provide ‘meaningful human intervention’
Georgios Tselentis (GAC)
01:11:20
If we are talking about APPROVED automated disclosure what is the balancing test about?
Milton Mueller (NCSG)
01:11:25
Some of my favorite humans are machines
Mark Svancarek (BC)
01:11:28
Alan, "meaningful human intervention" is not a requirement for a balancing test
Hadia Elminiawi (ALAC)
01:11:49
@Georgeous I agree with you
Mark Svancarek (BC)
01:11:54
At least not in all cases
Owen Smigelski (RrSG) (Namecheap)
01:12:24
While it is nice to talk about how AI can aid the SSAD in learning about automating disclosure requests, I am concerned about the cost of developing such a system. This would likely require millions of dollars to develop- who will pay for it?
Matt Serlin (RrSG)
01:12:42
I think Alan G was correct…if a CP believes the request violates the law, they need the ability to reject the request…how that would work with an automated system??
Milton Mueller (NCSG)
01:12:46
you will, buddy
Matt Serlin (RrSG)
01:12:53
Lol what Volker just said :)
Olga Cavalli
01:13:24
Hi apologies for the delay
Brian King (IPC)
01:13:49
+1 MarkSv
Brian King (IPC)
01:13:58
e) should be disclosing
Alan Greenberg (ALAC)
01:14:03
A decision for disclosure by the SSAD still goes to the CP for data release and they could conceivably filter requests that are problematic.
Matt Serlin (RrSG)
01:14:21
Agree with Mark SV on changing processing to disclosing
Hadia Elminiawi (ALAC)
01:14:41
@milton balancing test is essential where required by the law we are not arguing this
Mark Svancarek (BC)
01:15:47
Milton, there was some confusing public comment on this one. I didn't understand all of it, either
Owen Smigelski (RrSG) (Namecheap)
01:17:55
@Hadia- I still cannot imagine a scenario where a machine can perform a balancing test. To have something that is effective and accurate will require substantial resources to develop, and who is going to pay for it? The cost will not be insignificant.
Hadia Elminiawi (ALAC)
01:19:58
@Owen for sure the cost has to be taken into consideration. The report clearly notes this, however we do not know yet the costs entailed with such automated proceses
Mark Svancarek (BC)
01:20:57
I think we can move on. Stop offering me more data than I requested :)
Chris Disspain (ICANN Board Liaison)
01:21:51
Becky and I have a board call coming up so will drop. Apologies…
Alan Woods (RYSG)
01:22:11
Oh god no
Alan Woods (RYSG)
01:22:38
No locking.
Matt Serlin (RrSG)
01:23:32
Yeah locking introduces a whole new host of issues to tackle…I wouldn’t support
Stephanie Perrin (NCSG)
01:25:16
I don’t believe we have asked for a legal opinion on this matter, but as I have pointed out before, many (most ) registrars have other business relations with the registrant/ As far as I am concerned, if a registrar has information that he/she fails to look at in performing a balancing test, unless there is very serious fireballing of the data collections (using that term in the metaphorical sense) if they faild to look at relevant information,( e.g. press, political, religious, women’s health) in performaing a balancing test, they would be found negligent and responsible for resultant harm, We must separate out ICANN’s remit in terms of our policy, but registrars and registries have an entirely different horizon in front of them in terms of their responsibilities to their customer.
Stephanie Perrin (NCSG)
01:25:57
that was fireballing, not fireballing
Milton Mueller (NCSG)
01:27:03
Is someone seriously suggesting that you can lock someone’s domain simply by making a disclosure request?
Brian King (IPC)
01:27:48
nobody here suggested that
Milton Mueller (NCSG)
01:29:07
“The domain name that is the subject of a disclosure request shall be locked during the pendency of a disclosure request/“
Milton Mueller (NCSG)
01:29:15
sounds like someone suggested that
Brian King (IPC)
01:29:54
These are public comments
Milton Mueller (NCSG)
01:30:55
OK, good
Hadia Elminiawi (ALAC)
01:31:10
The disclosed data should be the data against which the decision was made as Mark A said
Brian King (IPC)
01:31:12
Authoritative data at the time the request is transmitted from the SSAD shall not be considered historic data.
Alan Greenberg (ALAC)
01:31:30
That means the data that is inspected must be captured and that sent instead of the then current data. That is MUCH harder to implement.
Berry Cobb
01:34:04
More for implementation, but seems when data is disclosed, that the disclosed timestamp can be matched with the Updated Date in the query.
Stephanie Perrin (NCSG)
01:39:20
My apologies, I forgot to include all attendees in my post
Stephanie Perrin (NCSG)
01:39:31
From Me to All panelists: (10:53 AM)
I don’t believe we have asked for a legal opinion on this matter, but as I have pointed out before, many (most ) registrars have other business relations with the registrant/ As far as I am concerned, if a registrar has information that he/she fails to look at in performing a balancing test, unless there is very serious fireballing of the data collections (using that term in the metaphorical sense) if they faild to look at relevant information,( e.g. press, political, religious, women’s health) in performaing a balancing test, they would be found negligent and responsible for resultant harm, We must separate out ICANN’s remit in terms of our policy, but registrars and registries have an entirely different horizon in front of them in terms of their responsibilities to their customer.
that was fireballing, not fireballing
Stephanie Perrin (NCSG)
01:43:21
one more time….firewalling, I need new glasses, apologies
Alan Woods (RYSG)
01:45:09
As drafted is fine by me
Matt Serlin (RrSG)
01:45:25
As drafted is good yes
Milton Mueller (NCSG)
01:45:54
scroll up to G again please
Milton Mueller (NCSG)
01:46:30
agree
Georgios Tselentis (GAC)
01:46:35
yes
Milton Mueller (NCSG)
01:46:43
can’t see how it would increase liability if it is a restatement of the law
Thomas Rickert (ISPCP)
01:46:43
OK
Mark Svancarek (BC)
01:48:00
thinking
Alan Woods (RYSG)
01:49:19
Thomas said it so much better than I lol
Stephanie Perrin (NCSG)
01:49:46
+1000 Thomsa
Matt Serlin (RrSG)
01:49:53
+1 Thomas
Mark Svancarek (BC)
01:53:01
It's not clear to me that we are discussing the second...
Mark Svancarek (BC)
01:53:12
So let's clarify
Mark Svancarek (BC)
01:53:29
clearly it is ambiguous to many readers
Mark Svancarek (BC)
02:00:46
+1 Thomas
Stephanie Perrin (NCSG)
02:10:54
I agree with Laureen, I think this is a useful clarification with the bullets.
Matt Serlin (RrSG)
02:12:31
Agree with not changing the initial report language but an implementation note may be okay to address…
Stephanie Perrin (NCSG)
02:12:38
Many regulatory investigations have no link to criminal law, but are vital to public safety
Marc Anderson (RySG)
02:15:18
+1 to what Janis and Chris said
Milton Mueller (NCSG)
02:17:37
implementation
Hadia Elminiawi (ALAC)
02:17:47
There is no need to sort this out here
Matt Serlin (RrSG)
02:17:51
Punt this one to implementation
Mark Svancarek (BC)
02:17:55
implementation
Hadia Elminiawi (ALAC)
02:18:41
+1 Marc A - It is a CP decision
Milton Mueller (NCSG)
02:18:52
agree with Marc
Matt Serlin (RrSG)
02:19:00
Marc makes a good point
Alan Woods (RYSG)
02:19:05
Agreed. This is CP only.
Hadia Elminiawi (ALAC)
02:20:24
I am just finishing the EPDP calla and will send you the amendments
Hadia Elminiawi (ALAC)
02:20:56
Sorry the above was not meant for this group
Matt Serlin (RrSG)
02:21:19
Stick with the contracted party language…
Marc Anderson (RySG)
02:21:49
agreed.... keep existing language
Hadia Elminiawi (ALAC)
02:22:31
No objection - does not matter
Stephanie Perrin (NCSG)
02:24:39
I agree with Marc. There is always judgement involved.
Matt Serlin (RrSG)
02:25:21
I have to drop for another call…thanks all
Alan Woods (RYSG)
02:27:14
“Wrong” define please
Owen Smigelski (RrSG) (Namecheap)
02:27:39
Just because a party may disagree with a decision does not make it wrong
Alan Woods (RYSG)
02:30:35
Gotta drop for another call. Thanks all.
Hadia Elminiawi (ALAC)
02:32:13
Thank you all bye
Brian King (IPC)
02:32:42
thanks all
Julf Helsingius (NCSG)
02:32:48
Thank you all
Marc Anderson (RySG)
02:32:51
thanks all