Re: [Ohttp] Discovery

[Mark Nottingham <[e-mail]>]

On 25 Jun 2021, at 6:22 pm, Andrew Campling <[e-mail]> wrote:
> 
> On 25 Jun 2021, at 08:39, Mark Nottingham <[e-mail]> wrote:
> 
>> On 19 Jun 2021, at 12:25 am, Eliot Lear <[e-mail]> wrote:
>>> 
>>> += The working group will consider and address operational matters, so that the output does not introduce any substantial negative impact to existing deployments.
>> 
>> This is a very Russell's Teapot kind of requirement; quantifying that there is no 'substantial negative impact to existing deployments' is a massive undertaking, considering the size of the Internet.
>> 
>> It's also loading a lot into 'operational matters'; the language seems to automatically prioritise them over any other concern engaged by the work. That seems deeply problematic to me; if we're going to prioritise anything, it should not be that. 
>> 
>> Lastly, it is effectively applying the Precautionary Principle to our work. While I strongly support the idea of being more careful and thoughtful in our actions, this is a 180 degree turn for the IETF, and I don't believe it's got the balance right to say that we must be certain there's no harm to a specific constituency to proceed.
> 
> Noting RFC8890, if the Internet really is for end users then it seems perfectly valid that, at a minimum, consideration should be given to the operational impact of new protocols.  One way to address this would be to undertake some multi-stakeholder engagement to find out whether there are significant operational downsides that should be taken into account.  Whilst I understand that such an undertaking will not be attractive to everyone, it is a reasonable expectation that this is done before a new protocol is introduced if we are to avoid or at least minimise any harmful effects.  

Yes - but there's a difference between doing that work and achieving certainty that there aren't any substantial negative effects on operations.


--
Mark Nottingham   https://www.mnot.net/