Search RPD Archives
Limit search to: Subject & Body Subject Author
Sort by:

[rpd] impact analysis for policies

Sylvain BAYA abscoco at
Mon Jun 24 13:11:52 UTC 2019

{Warning & apologize :-) you might found it tl;tr}

Hi all,

Le 6/23/2019 à 11:55 AM, JORDI PALET MARTINEZ via RPD a écrit :
> Hi Sylvain,
> Again thanks for your inputs. Comments below in-line.
> Regards,
> Jordi
> @jordipalet
> El 22/6/19 11:16, "Sylvain BAYA" <abscoco at
> <mailto:abscoco at>> escribió:
> [...]
> CPM section 3.4.1 : «[...]The Working Group Chair(s) may request
> AFRINIC to provide an analysis (technical, financial, legal or other),
> of the impact of the draft policy proposal.[...]»
> ...i think we must follow our rules or change them if they appear to
> be broken somewhere. Right ?
> There are two things here. One is the policy manual and the other one
> is the internal AFRINIC operational procedures.

Yes of course, i mean i **expect** the "internal AFRINIC operational
procedures" to folow the CPM (Consolidated Policy Manual). Am i wrong ?

> If you read the message from Dewole on the thread you referred to
> ( it clearly
> stated that this is the procedure:

Ok i have read it again carefully, thanks for have pointing it to me.

What i understand, from the explanation of Dewole, is that :

* The Chairs and the Staff found an arrangement regarding the production
and publication of impact analysis reports ;
* This **agreement** was concluded by them to simplify the process (PDP)
and the procedure (Operations) ; 
* The Staff has accepted to provide (in due time) an impact analysis for
every versions of all the policy proposals in discussions ;
* The Chairs will no longer have to request again for a specific impact
analysis report when needed ;
* We have a kind of automated sub-system. conclusions in correlation with the facts you provided below :

+ The Chairs have got a willing|desirable arrangment with Staff about
impact analysis publication but ;
+ The Staff has proved to be unable to provide impact analysis for all
the versions of the policy proposals in due time (they can certainly
explain this fact) ;
+ The Staff are **always** providing impact analysis report for a random
set of policy proposals and not for all versions of a given policy
proposal ;
+ Sometime they have failed to provide very needed impact analysis
reports at time ;
+ An analysis (yours) for all the policy proposals shows that : the
actual procedure the Staff is using to provide Impact Analysis report
looks as producing random results (and this is not better than the
normal Process [PDP] as provisioned by CPM section 3.4.1) ;
+ The **agreement** obtained by the Chairs is proven to be flawed,
particularly by your analysis generously (thanks) provided below ;
+ By this **agreement**, the Chairs have lost|abandoned (involuntarily
for sure) their responsibility in regard of the impact analysis report
requests ; 
+ It seems to be an opaque procedure because community was not aware
before the recent mail from Dewole (**Varyance** [CPM section 3.4.1] is
a good|useful tool, but we deserve a bit of transparency and to be
timely informed of such a **emergency* *[and **temporary**] decision).

CPM section 3.6 : «3.6   Varying the Process

The process outlined in this document may vary in the case of an
emergency. Variance is for use when a one-time waiving of some provision
of this document is required.

 1. The decision to vary the process is taken by a Working Group Chair.
 2. There must be an explanation about why the variance is needed.
 3. The review period, including the Last Call, shall not be less than
    four weeks.
 4. If there is consensus, the policy is approved and it must be
    presented at the next Public Policy Meeting.»

I prefer that we follow (or openly amend) the CPM section 3.4.1, and
hopefully we will have more chance to receive the requested impact
analysis reports in due time.
Perhaps we should also pass through a new policy proposal to ameliorate
something in regard of that...not sure what's useful to do but :-/
> “For a while now, the procedure (as part of lessons learnt) has been for
> staff to provide their analysis of draft proposals once they are
> received so it's not really on the author. Co-chairs and staff have long
> agreed that this makes sense.”

...IMHO it was a very good deal for the community, but it is now proven
to badly functioning since then. 

I guess we want to fix. Then what to do ?

Your comments on the disponibility of impact analysis reports for all
the actual policy proposals seems to confirm a broken procedure ; where
the PDP is not applyed (with this agreement the Chairs are not doing
their duty : ask for a given **needed** impact analysis report).
> I fully agree with the rest of the text, that mentions that if a new
> version is sent, for example one week before the meeting, it may not
> be feasible. However, my point is that, proposals that have been
> received several weeks before the meeting AND had already a previous
> impact analysis, AND the authors just updated the proposal to comply
> with the impact analysis, in my opinion, results in a few hours (not
> to say minutes) to rework the previous impact analysis.

...useful insights for the Staff, i think ?
> Chairs and staff know, that at least in my case, if I get an email
> from them with questions about my proposal version or impact analysis,
> I do my best to respond in a matter of few hours, often during the
> next hour or so, unless it falls during the time I’m traveling or
> sleeping, but most of the time I respond within the same day. I don’t
> break during the weekends, don’t do holidays or take days off, and
> average sleep 5 hours per day, so it is difficult to believe that I’m
> not responding “quickly”, unless by chance an email is classified as
> spam and filtered (I don’t recall it happened with emails from staff
> and co-chairs, but just in case).
> Also take the considerations from SM about the existence of similar
> policies and impact analysis in other RIRs, which may be at least a
> starting point (

...thanks Jordi.
You must know that you are helping me a lot to quickly accomodate to the
PDP and particulary some hidden (but don't get me wrong it's just about
facts we are still discussing), and **policy-free**, practices in this

I learnt and concluded two things while reading, again, the mentioned
SM's mail carefully :

+ To ameliorate the, timely, delivering of impact analysis reports, we
need (amongst other actions) to figure out : "How long does it usually
take from the time the PDWG is asked to provide feedback [...] to when
the analysis of the proposal is available ? 

You did a bit of analysis, of the situation, below and it was really
instructive to me. Thanks once more. Hope we use it to fix the issue.

+ While "all the analysis work is being left to Afrinic Ltd." Are we
expecting to have an optimal PDP ?
> So, if we analyze **all** the policy proposals on the table (so no
> just “defending” my own position, but also other authors of proposals):

Thanks for sharing your thoughts (analysis) about the disponibility of
impact analysis for all the policy proposals on the table.

Hopefuly it will help the staff to quickly provide the missing ones...
>  1. AFPUB-2019-GEN-001-DRAFT01: published on 17^th May. This is the
>     only proposal, which is quite complex, that I agree may take
>     longer than 4 weeks for the impact analysis, even if there is
>     already an existing impact analysis in other RIRs.
>  2. AFPUB-2019-IPv4-001-DRAFT01 and AFPUB-2019-IPv4-002-DRAFT01:
>     published on 13^th May. However, those proposals **share** the
>     same impact analysis as the one already done for
>     AFPUB-2018-GEN-003-DRAFT01 (1^st November 2018) and they are
>     **simpler** than that one, take only on IPv4 (instead of
>     IPv4+IPv6+ASN), impact analysis already available also from other
>     RIRs.
>  3. AFPUB-2019-V6-001-DRAFT01: published on 14^th May. Impact analysis
>     already published for the previous version. It is true that this
>     policy proposal is a bit more complex than the previous versions,
>     but I think it was clear in the meeting (where it reached
>     consensus), that the staff don’t have any “impacting”
>     considerations, so what they said in the meeting, could had been
>     published before.
>  4. AFPUB-2019-ASN-DRAFT02: published on 11^th April. My god! How come
>     we can’t have in more than 2 months an impact analysis, which was
>     already published for the previous version, while this version is
>     actually precisely, much simpler, considered the impact analysis
>     to correct all the points, and **reached consensus** during the
>     discussion in the list? I’m tempted to appeal the chairs decisions
>     on it (no-consensus), because if we had the impact analysis and
>     the chairs, really considered the list, there was overwhelming
>     support and, in my opinion, none of the comments in the meeting
>     was actually sufficiently substantiated as something **against**
>     the arguments in favor. I’m not going to do so, because I support
>     the previous co-chairs decisions, and unfortunately (and this is
>     BIG BIG BIG issue), some of the inputs of the people in the room
>     where from folks that either a) they are not in the mailing list
>     (I already suggested them to join and participate), b) some folks,
>     which are in the list, which are **actual** authors of other
>     policy proposals, didn’t brought their inputs in the list, they
>     have actually brought them in the meeting, just for the **insane
>     sake of overthrow this proposal** and exclusively for their own
>     interest (as I’m not supporting their own policy proposals. I
>     those co-authors of other policy proposals are really looking for
>     the well of the community, they **MUST** participate in the list
>     **not only** to defend their own policy proposals, but also to
>     improve the ones from other authors (as I do). If they provided
>     those inputs in the list, then I could have made a new version in
>     time for resolving their issues and reaching consensus.
>  5. AFPUB-2018-GEN-004-DRAFT02: published on 18^th April. Received
>     substantial inputs in the list, but it was withdrawn for author
>     personal reasons (I know the details, but I will suggest him to
>     explain to the list ASAP). However, despite having a previous
>     impact analysis, we got nothing this time. How come?
>  6. AFPUB-2016-GEN-001-DRAFT08: published on 7^th June. There was a
>     previous version impact analysis. The proposal didn’t change so
>     much that that impact analysis could have been reviewed, at least
>     in draft, to have “something” before the meeting. So again, why
>     not we had something?
>  7. AFPUB-2018-GEN-001-DRAFT03: published on 5^th June. Again, there
>     was a previous version impact analysis, and the changes were to
>     accommodate to them, making the proposal much simpler, and
>     ensuring that those details on the impact analysis are resolved.
>     So how come we didn’t have one at least in draft? This one also
>     has been adopted in other regions, and of course, we have impact
>     analysis in others.
>  8. AFPUB-2018-v4-001-DRAFT-01: published on 28^th October (2018), is
>     the only one that had an impact analysis. Is this meaning that
>     staff needs 8 months to complete an impact analysis? **by the
>     way** I don’t know when it was completed, because a) I can’t find
>     any message announcing it to the list, b) the impact analysis
>     doesn’t say the date! Is that right?
>  9. AFPUB-2017-GEN-002-DRAFT-05: published on 9^th June. There is a
>     previous version impact analysis, and again, there are minor
>     changes, so I could have expected at least a draft impact analysis
>     considering those minor changes.
...randomized delivering ?

Is it intended that it walks like that  ?

> Note that sometimes the publication is delayed several days, so the
> staff has got in fact **more** days to start the impact analysis.
> Note also, that in some cases, if an impact analysis can’t be
> completed in time, it is good enough to have the same “draft” inputs
> that the staff is providing in the meeting, **but** authors need that
> **before** the meeting, so we can work out a clearer explanation in
> case of discrepancies.

:-) Jordy, reassures me that we can correct it...and please tell me also
how we can succeed ?

Any other voice out there ? :'-(
>     I may agree that if a proposal has been submitted only 10 days
>     before the meeting, it may be difficult to make it in time, but
>     not for a “new” version of an existing proposal, which already had
>     the analysis impact.
> Ok if it's unpractical, we should try to modify the CPM, at section
> 3.4.2, to fix|ameliorate what seems to be broken...
> CPM section 3.4.2 : «[...]No change can be made to a draft policy
> within one week of the meeting. This is so that a stable version of
> the draft policy can be considered at the meeting.[...]»
> As my analysis above shows, I don’t think anything is broken. We can
> always improve the PDP, but if something is already part of the
> operational process, it is just a clarification, and there are more
> important policy proposals in the table than sending on for just
> resolving a non-problem. I will be in favor of this if we have 2 days
> policy-day next time!
> *** Chairs and staff must agree on how much an impact analysis may
> take in every policy proposal case. Some proposals may take hours.
> Very few may take weeks or even more than a month. However, it may be
> a **management** or even **board** problem to ensure that the staff
> has available “human resources” to do their work timely. May be the
> folks in the staff that are responsible for making the impact analysis
> (I know it implies different departments and even CEO approval), need
> to prioritize, up-front of a meeting, what of their tasks are more
> important for a successful policy-day.

I agree that the BoD could help here; but i think we must first clearly
state the problem to solve.

First, our Chairs **agreement** with the Staff, creates an unexpected
situation. (this is fact)

Then, to handle this fact, IMHO we should stop this **agreement** and if
we want to ameliorate things in that regard, it will be possible to try
something new (BoD reaction, PDP revision, other possibility)...
>     And of course, I’ve asked the staff several times for it, during
>     the last weeks. Co-chairs were copied.
> Ok, i personaly think that it is legitimate to want to move things
> forward ; but i also understand (policily speaking) that the staff was
> in their rights to not considere your *direct* request as a legitimate
> one. Even though i guess they were just busy :-)
> Agree (see previous comment), but one of the **most** important things
> that a RIR must do is PDP! This is top priority.

Ok, hope that the coming facts will convince us of that *top* priority...
>     Precisely, most of the newer versions, at least in the case of my
>     proposals, where to address the points raised in the impact
>     analysis. So, it should be easy to get the previous impact
>     analysis, compare the differences of the old and new versions of
>     the policy proposal, against the impact analysis, and raise if
>     something is missing.
> :-) i'm sure that at least one of them (policy liaison?) might be
> reading...
> Do you think we should also (policily speaking) ask one of them (say
> the same staff member i mentioned above) to follow|read the
> discussions and react when we need|seek their expertise ? :-D
> I know by the fact, they are reading, that’s why I post those messages
> here when private ones didn’t work as expected! Again, not blaming
> them, just makes sure we improve things.

...again, not blaming anyone too, i just want to understand how things
walk ; in order to see if we can ameliorate something.

Then, tell me please : why (despite the provision of CPM section 3.4.0)
we don't get any response since we ping them ? [1]
> Staff and chairs know very well that I send them emails any time I can
> offer suggestions, ideas to improve participation, requests for
> anything which is good for them and for the community. I always try my
> best, because when you contribute, you discover issues to resolve.
> Nobody is perfect, this is about **collaboration from all**.
> For example, I discovered a few weeks ago, that the numbering system
> of the policy proposal is flawed so suggested it to get changed and
> don’t differentiate them, just “year-number-version”. The point is
> that because the policy proposals get “classified” as “general, IPv4,
> IPv6 or ASN”, if during the discussion, the policy proposal change its
> scope, which happens sometimes, then you need to withdraw the policy
> proposal and re-start it. This was the reason instead of just updating
> my last year Intra-RIR policy proposal (which was “GEN”), turned into
> a new one.

Thank you for sharing your findings with us.
When i see the provision from the CPM section 3.4.1 then i confirm that
the Staff has the full rights to change it right now :-)

CPM section 3.4.1 : «[...] Each draft policy is assigned a unique
identifier by AFRINIC and the AFRINIC website shall also contain the
version history and the status of all proposals. [...]»

> Hopefully this is considered by the staff soon and we remove this
> classification soon, and we get all the policy proposal in the table
> renumbered to the “new” system.

Many thanks for your endeavor. I look forward to see that improvement.
>     Believe me, in most of the proposals, this is a matter of few
>     hours of dedication. My only concern is if staff is overbooked,
>     which is something that should be evaluated by the organization
>     (not authors).
> First, have we already tried to fix the problem with our means (the
> policy) ?
> As said before, there are two parts. PDP (policies) and procedures.
> Not all requires policies. If we micro-manage, it becomes cumbersome.
> In my thinking, we should use policies only when procedures don’t
> improve by themselves (or the community radically disagree with the
> procedures), or don’t exist at all. PDP is heavy, its slow, it
> requires a lot of community effort, which sometimes is not needed if
> just improving procedures is sufficient. Just consider the time it
> requires to have 10 policy proposals presented in the meeting. How if
> we try to present 15 or 20?

Thanks, i will remember this PDP picture you painted for us (me). :-)
> [...]
> seems as there is something to fix :-)
> Considering what i have said above, do you admit that, as it is, the
> process is flaw ?
> Please, consider my point about the difference between operational
> procedures and policies. Micro-management is not good, even if
> sometimes is needed.

...i understand it thanks. 

However, my will is not to micro-manage operational tasks but to find
appropriate way forward to deal with what seems to be flawed.

But if you know how to fix it without any PDP alteration. Fix it please.

I really appreciate your support, i see you trying to help me :-)

...just remember that i'm still learning the PDP and logically, i first
rely on the CPM (v.1.3).

If some operational procedure doesn't follow the PDP, it shall be more
difficult for me to understand how things walk :'-(
>         The impact for this is really big. Right now, we have 8 policy
>         proposals that didn't reached consensus and only considering
>         new proposal that I've in mind to submit (5), it means that
>         for the next meeting we will have *at least* 13 policy
>         proposals in the table.
>     ...other might also propose drafts for new policy proposals :'-(
>     I hope so, more people contributing, and yes, that’s why I said
>     “at least” :-)
>     :-) but please withdraw one of your two similar policy proposals
>     about out of AFRINIC region resource (IPv4) transfers. That action
>     could diminish the number of policy proposals ;-)
>     Remember that I presented them jointly, with the same time for two
>     proposals, that for a single one, so actually that’s not taking
>     time.  Anyway, if we can have a “clear” decision in the list from
>     now to the next meeting, I’m happy to withdraw one of them, but
>     this is only going to work if we have 200 voices in the list
>     supporting one of those. Otherwise, I can withdraw the wrong one
>     and then in the meeting, the participants decision withdraws the
>     other one, so we have nothing. You see the point?
> you don't want to take the *risk* (i don't actualy see a risk,
> though) ? :-)
> I may change my mind on this depending on how the discussion goes in
> the next weeks … Let’t talk that in the relevant thread!
> Thanks again!
> please, it's me :-)

> [...]


Sylvain B.
Website : <>
Wiki : <>
Surveys : <>
Subscribe to Mailing List : <>
Mailing List's Archives : <>
Last Event's Feed : <>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0x0387408365AC8594.asc
Type: application/pgp-keys
Size: 4826 bytes
Desc: not available
URL: <>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 819 bytes
Desc: OpenPGP digital signature
URL: <>

More information about the RPD mailing list