Re: [arch-d] clusterpation, was Proposed IAB program: Evolvability, Deployability , & Maintainability.

[Brian E Carpenter <[e-mail]>]

Hi John,

On 22-Jul-20 16:00, John Levine wrote:
> In article <[e-mail]> you write:
>> Because (for reasons I don't understand) I'm involved in AUTH48 processing
>> for RTCWeb, I find myself wondering whether the proposed IAB program might
>> include considering ways to avoid C-238-style protocol specification
>> clusters, whether from the point of view of architectural oversight,
>> standards process oversight, or RFC Editor oversight.
> 
> I think we all agree that we don't ever want to see another document
> cluster as large or as long-lived as C238.
> 
> But I would rather not attempt to invent processes to try and guess
> how best to prevent hypothetical superclusters. I'd rather that the
> RPC and I (or a future RSE) work informally with ADs and WG chairs
> when we see a potentially large cluster forming, or worse when a
> document is about to link two normal sized clusters into a big one.

That would be wise, but by the time anybody notices, it may be too
late. I do agree that the answer is not to add more process. The 
question on the table, I think, is what factors lead to the formation
of such a cluster, and what design issues should be considered to avoid
it? (Plus the old newtrk question: how can we document clusters of
documents usefully for implementers and operators?)

    Brian