Andre writes:
> probably, yes. if ai is advanced enough to deal with incoming
> communications in an acceptable fashion, this will be just fine.
> - your trust in your ai would be most commendable and as imho, ai will
> be running everything in a few years anyway, this is perfectly
> acceptable
Andre, please accept my apologies for writing about the "Turing test",
it was meant lightheartedly but has confused the issue. It is my fault
for being flippant.
I didn't envisage that a true AI would handle would such enquiries. I
was pointing out two areas that are problematic, separately:
1. The RIPE NCC may not recognise an auto-responder when they see it, if
it has been tailored to them specifically (and to achieving a formal but
spurious compliance); and
2. A response by a real human being may be just useless as a bounce
message, if constrained by policy. That is to say, if all the human is
permitted to do is choose from a small range of stock responses (such as
directing the person making the enquiry to some pre-written FAQ) then
this is equivalent to an autoresponder, so why prohibit (deem
non-compliant) only the automated response?
However from Hervé's reply I see that a less ambitious bar is being set,
and an autoresponder is acceptable, whether a human autoresponder or a
software one. I have my doubts that this really achieves anything
useful, but at least it is clear.
On 25/09/2017 11:34, ox wrote:
> On Mon, 25 Sep 2017 10:55:09 +0100
> Malcolm Hutty <malc...@linx.net> wrote:
>> Scenario 1: An LIR directs e-mail sent to their abuse-cc: address to
>> an auto-responder that says "This mailbox is not monitored by a human
>> being", and advises on alternate "support services" (e.g. a FAQ, a
>> webform that feeds a ticketing system etc). Is RIPE NCC intended to
>> mark the attribute as invalid in this scenario?
>>
> there is no point to have an email address that does not exist or is not
> monitored.
>
> if or when email ever stops working and is replaced by alternate "support
> services"
> this will be a good timeTM to accept non monitored email addresses
>
> but to granularly define generally accepted forms of communications,
> is pointless as there will always be a good reason for whatever form of
> communication, to not be suitable or acceptable to someone.
>
> take mobile, or phone numbers, it can easily be argued "but i am not
> available to take calls" or whatever... - everything always has
> exceptions, it is whether those exceptions are generally reasonable or not
> and/or generally acceptable.
>
>
>> Scenario 2: An LIR filters incoming e-mail sent to their abuse-cc:
>> address. Email from RIPE NCC gets "priority treatment", i.e. is
>> directed to someone who passes a Turing test administered by the NCC.
>> E-mail from anyone else gets the same treatment as in scenario 1.
>>
>> Is Scenario 2 compliant with the policy? If not, how is RIPE NCC
>> supposed to know to mark the attribute as invalid? What tests are the
>> NCC supposed to administer? And what must an LIR do to pass them?
>>
> probably, yes. if ai is advanced enough to deal with incoming
> communications in an acceptable fashion, this will be just fine.
>
> ianal but, there are legal implications, if your ai receives x notice,
> replies, etc.
>
> - your trust in your ai would be most commendable and as imho, ai will
> be running everything in a few years anyway, this is perfectly
> acceptable :)
>
>
> Andre
>
--
Malcolm Hutty | tel: +44 20 7645 3523
Head of Public Affairs | Read the LINX Public Affairs blog
London Internet Exchange | http://publicaffairs.linx.net/
London Internet Exchange Ltd
Monument Place, 24 Monument Street London EC3R 8AJ
Company Registered in England No. 3137929
Trinity Court, Trinity Street, Peterborough PE1 1DA
