Killing old/slow groups - transition thinking

Marshall Rose mrose@dbc.mtview.ca.us
Sun, 8 Dec 2002 10:04:32 -0800


all - this is a very long note, sorry.    
    
> just for kicks, I revived some old scripts of mine.
> if you check out http://www.alvestrand.no/ietf/milestones/milestones.html, 
> you will find the list of all milestones in the IETF, sorted by date.
> 
> If we kill all the WGs that have more than 1 year since *promised delivery* 
> of a document, the following WGs die (in roughly chronological order):
    
first, the PACT I-D gives WGs an 18-month timer to get their first I-D
past the IESG. (personally, i think that 18 months is too long, but i
guess i'm not as kind-hearted as others.) but the idea is that a WG that
needs to be able show it can produce something in the "short-term".
    
this is different than whacking WGs that have missed their milestones.
however, now that you bring up this topic...

    
> OSPF FTPEXT IFMIB RIP VRRP MANET TN3270E PIM ISIS MSDP DNSOP DHC PPPEXT 
> SMIME IPCDN MBONED DNSEXT RMT IPSRA SYSLOG IMAPEXT MOBILEIP MPLS KRB-WG 
> WEBDAV ITRACE ACAP CALSCH ATOMMIB LDUP TSVWG CCAMP APEX IMPP OPENPGP TLS 
> MSEC STIME L2TPEXT SEAMOBY BMWG IPV6 MAGMA PILC MALLOC ZEROCONF EOS PRIM 
> MULTI6 IPSEC TEWG FAX
> 
> 53 groups, or roughly half the IETF.
> I may have missed some, or counted some twice.
    
(at least one of these is in the "waiting for the IESG to publish our
I-Ds so we can disband" state, so we can probably pair down the list a
bit, but i get your point... let's call it 40)

    
> The IETF may be better off if we do this.
> But - it's work.
> Imposing a new rule without doing something about the old groups that don't 
> feel bound by the rule will be seen as unfair.
> Imposing a new rule on old groups will require a significant amount of 
> legwork - some of which will be seen as "meaningless bureaucracy", some of 
> which will receive some degree of sympathy.
>
> How should the transition be managed?
> 
> (and yes, we have already instituted sending people warnings about their 
> expired milestones once or twice a year. Helped a little....not much.)

the way i interpret this list is that for quite some time the IESG
hasn't cared about milestones. because if it did care about milestones,
then this  list should be much, much smaller. further, this tells me
that a long time ago the IESG forgot that actions have consequences.
(afaik, there have only been three ADs in the last decade that have
whacked stalled/problematic WGs in a timely fashion: klensin, o'dell,
and rose; where timely = within a year of them missing a milestone.)
    
ideally, each WG ought to be working towards a successful conclusion in
good faith. in practice, we know this isn't often the case, and the
reasons always vary. in order to deal with this, the IESG can either
micro-manage each WG, or it can put something in place that let's each
WG manage its own behavior. when the WGs see that actions have
consequences (or, more accurately, inaction leads to a consequence),
then most are going to shape up and fly right.
    
however, because for most of the last decade the IESG has refused to
tie grossly overdue milestones to whacking WGs, the WGs know that the
milestones are meaningless...
    
intuitively, i will tell you that when klensin, o'dell or rose actually
whacked a WG in a timely fashion, the other WGs in their areas became
more productive and less stalled, because it was damn clear that actions
did have consequences and that people were paying attention...
    
at this point, there's the usual rejoinder about this being a volunteer
organization, etc., etc. to which i reply: if the organization can't
follow it's own rules, then the organization may very well be past
saving... certainly not the most pleasant outcome, but one that we're
headed towards.

    
finally, with respect to transition issues: the PACT I-D doesn't say to
whack WGs that are more than a year past their milestones. however, if
you were asking me for a concrete suggestion on how to implement that,
then i suggest the IESG make an announcement that it's going to start
enforcing milestones:
    
    1. for existing WGs, any WG not current with its milestones by June
       30th, will have to go through the process of re-chartering, with
       the likely expectation of getting back a "no".  (this is called
       zero-based accounting.)
    
    2. for new WGs, any WG missing a milestone by more than four months
       will have to get re-chartered.
    
certainly we can tinker with the numbers; and, obviously, if a WG starts
to make real progress towards the end of the deadline, the ADs should
look favorably upon it (the goal here is to get WGs productive, not
whack them on a technicality...)
    
however, none of this is worth anything unless the IESG on June 30th is
willing to start whacking WGs. when WGs see that inaction leads to a
consequence, then WGs will start being more productive, less stalled,
etc., etc.
    
and the beauty of this is that the IESG doesn't have to micro-manage
each WG. it just has to set out some general guidelines that it's
willing to enforce.
    
/mtr