The IETF is not the ‘standards police’

By on 19 May 2021

Category: Tech matters

Tags: ,

Blog home

“You can’t do that, the IETF said so” is something that nobody who actually knows the IETF would agree with.

The problem is, outside of the IETF, a lot of people think this is how it works. 

Why do I bring this up, at this time? Well, a recent IETF draft has some rather unorthodox instructions, yet it makes an excellent point. 

This draft explores an idea. The idea is that you can say anything, at any time, in an IETF draft. A draft in the IETF, isn’t a “normative” document, which has binding force or intent, because it’s what the word says — it’s being ‘drafted‘.

It consists of nothing but ‘ideas’. Normative language is how the IETF tries to formalize the expression of things, using all uppercase words like MUST, MAY, SHOULD and MUST NOT specially, in quite carefully checked ways.

Anyway, this latest fascinating draft has been authored by Warren Kumari from Google. Give it a read, it just might surprise you.

The IETF holds onto its history forever, drafts and all

It’s important to understand the IETF document archive holds on to drafts, forever. It’s a (semi) open publishing space, which aside from obvious abuse or harm, doesn’t delete documents if they conform structurally to the style of writing. It makes a virtue out of this; you can use the data tracker to follow the edit history and path a document takes through the IETF process.

Warren is trying to help everyone understand that no matter how much IETF-watchers try to tell people “drafts don’t tell you what to do, or even represent a consensus view in the IETF”, they aren’t going to listen. And, because of how the IETF document archive functions (things last forever), it’s very likely that anything submitted is going to persist in the document archive forever.

IETF drafts are just attempts at seeking consensus

People who work in the RIR community are used to saying “we’re not the routing police” with regard to the abuse of Internet number resources. In a similar manner, the IETF is not really the ‘standards police’ that some people think it is. Some people think the IETF can somehow make things happen or not happen, can make the Internet stop or start.

But, in reality, the IETF is more an indication of technical (mainly) consensus, among people who volunteer their time. It isn’t regulation as much as it is expression of desire. If it’s in draft, it is still an idea in formation. If it becomes an RFC, it’s reflecting ‘rough consensus’ and has more force of meaning — more people think this is a good way to describe things.

Read more: The making of an RFC in today’s IETF

If it has BCP (Best Common Practice) or if it has STD (STANDARD) status, it has moved further along the chain — if you find someone departing from this expected behaviour, it’s very likely causing problems and you can expect people to wake up to this and require (within the limits of their authority) the wayward entity to do what the STD document says.

For a BCP, it’s more that there are expectations here, but they’re still more aspirational than a requirement for conformance. 

So, if you see something in another system, or on-the-wire in the Internet, it’s tempting to go into IETF draft documents and find statements that support your view that something on the Internet is ‘wrong’ using the implication that the IETF said so.

Ceci n’est pas une draft (this is not a draft)

Warren Kumari has some status in the IETF (can anyone be said to be ‘in’ the IETF?) as an IETF Area Director. His document really tries to get at the key point that a draft isn’t an expression of the IETF’s ‘view’ on anything.

Within the limits of belief, Warren actually did discuss and submit 13 revisions of this document, but it would be wrong to even ascribe meaning to the words in the document that refer to the process stages to get there.

You would be better off using the IETF document revision information, which is metadata.

The draft (these documents are often called ID for Internet Draft) is called draft-wkumari-not-a-draft-13. The title tells you that it’s a draft, who authored it, and the 13 refers to the number of revisions. 

The most important thing this document can do, is say one thing simply:

“Drivel which endures remains drivel”

“All too often one reads something in the press, or some ravings on a mailing list that reference some Internet Draft, that claim that “the IETF thinks that XXX” or that the ID is an IETF document, and so represents support by the IETF.

Repeatedly pointing at the RFC Editor page, carefully explaining what an ID is (and isn’t), describing how consensus is reached, detailing the Independent Stream, etc doesn’t seems to accomplish much.

So, here is an Internet Draft. I wrote it. It’s full of nonsense. It doesn’t represent the “IETF’s views”; it doesn’t mean that the IETF, the IESG, the RFC editor, any IETF participant, my auntie on my father’s side twice removed, me, or anyone else believes any of the drivel in it.

In addition, the fact that a draft has been around for a long time, or has received many revisions doesn’t add anything to the authority – drivel which endures remains drivel.”

There’s a lot more entertaining nonsense in that draft, so it’s well worth a read. But the compelling message here?

As Warren says, “drivel which endures remains drivel”.

Bad ideas, published in draft, aren’t going away.

Rate this article

The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.

Leave a Reply

Your email address will not be published. Required fields are marked *