incubator-couchdb-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Randall Leeds <randall.le...@gmail.com>
Subject Re: [DISCUSS] dont't abuse of "lazy concensus" on mail tagged [DISCUSS]
Date Fri, 10 May 2013 19:39:15 GMT
On Fri, May 10, 2013 at 12:34 PM, Noah Slater <nslater@apache.org> wrote:
> Randall, I think I understand your argument. Would it be true to say that
> you think that "DISCUSS" threads should be reserved for actual discussion,
> and that we need a new tag for the threads which give notice about lazy
> consensus?

Or perhaps no tag, since my understanding of lazy consensus is that
one needn't call for it, one surmises it through general awareness and
sensitivity to what is and is not likely to be controversial or
require input.

In my description of lazy consensus I was sort of saying that every
commit in CTR is an instance of lazy consensus wherein the committer
has taken action they believe would be representative of a consensus
opinion without actually asking for it, i.e. lazily.

So I'd prefer to see [DISCUSS] and [URGENT] used and just be mindful
that discussion suggests ample allotment of time for discussion. I
don't believe we need a specific tag for lazy consensus because I
agree with you that it's the default operating mode.

>
>
> On 10 May 2013 20:00, Randall Leeds <randall.leeds@gmail.com> wrote:
>
>> I'll hop in to repeat my observation one more time and offer evidence of
>> the behavior which caused me to empathize with Benoit.
>>
>> In the recent email about old releases tagged [DISCUSS] Noah said:
>>
>> "If nobody objects in 72 hours, I will assume lazy consensus and proceed."
>>
>> I like lazy consensus and consider it rolling and ubiquitous in the actions
>> of committers and in play even as we make decisions with zero discussion.
>> We act because we *believe* we would have consensus. In every case where
>> there is no formal discussion I believe I am representing my best guess at
>> what *would be* explicit consensus if it were discussed. This is my
>> understanding of lazy consensus but I'm happy to be corrected.
>>
>> However, 72 hours seems antithetical to discussion. If you call for
>> discussion because you want feedback, please give some time, especially
>> when it's not urgent. I'm not sure it is necessary to say exactly how long.
>>
>> That's a concrete recommendation from me. I hope that is constructive and
>> can help resolve this discussion.
>> On May 10, 2013 11:50 AM, "Noah Slater" <nslater@apache.org> wrote:
>>
>> > It's also perfectly fine to respond saying "woah there cowboy, we need to
>> > discuss this first."
>> >
>> >
>> > On 10 May 2013 19:47, Jan Lehnardt <jan@apache.org> wrote:
>> >
>> > > Maybe what is missing from this is that lazy consensus leads to things
>> > > that can never every be changed again. It is just a tool to keep a
>> > > distributed team going. If we do a thing and it gets lazy consesus’d
>> > > and implemented and even shipped, we can still *at any time* realise
>> > > it was a mistake, make a course correction or revert and move on.
>> > >
>> > > Jan
>> > > --
>> > >
>> > >
>> > >
>> > > On May 10, 2013, at 19:30 , Benoit Chesneau <bchesneau@gmail.com>
>> wrote:
>> > >
>> > > > I'm starting to think you don't read me carefully enough.
>> > > >
>> > > > I don't care about giving any evidence. The topic is about giving
>> more
>> > > > time to the discussion. The principle of using *by default* lazy
>> > > > consensus is what I consider an abuse. I explained it why third time
>> > > > in that thread. And already did it before that mail. But you refuse
>> to
>> > > > take my arguments in consideration keeping to ask me to show you how
>> > > > thing turned out to be wrong. Which is not the topic.
>> > > >
>> > > > The problem by using lazily consensus over a shot time is that you
>> > > > don't let people think about it much. Which wouldn't be a problem
if
>> > > > there was an intense communication between people. But this isn't
the
>> > > > case today. Some ideas are still coming from nowhere without
>> > > > preparation. Don't get me wrong I don't say that these ideas are bad
>> > > > or that there wasn't any thinking behind them. No the problem is you
>> > > > expect that people are able to answer it in 72 h or so. your time.
>> > > > Which don't let  sometime the time to think much about it and give
>> > > > your opinion or possible changes to it. Sometimes you really want
to
>> > > > tell a thing but finally can't do it because of timing issues.
>> > > > (Sometimes yes, you 3 days are really short). Maybe it could be just
>> > > > by saying it (like "hey I really want to answer but i don't have the
>> > > > time") which I think could work. But I clearly think that in that
>> case
>> > > > just giving more time or simply not using lazy consensus could just
>> > > > work. This is why I propose to adapt the time asked for a lazy
>> > > > consensus depending on the context, ie. not using 72 h by
>> convenience.
>> > > > The delays proposed were just some suggestions.
>> > > >
>> > > > To be clear, I strongly disagree to use the lazy consensus as *the
>> > > > default* way to take decisions. The apache way considers it as an
>> > > > important and main way to build (some kind of) consensus.  But main
>> !=
>> > > > default . It is also saying that we should try to build a consensus
>> > > > first. But not it is not saying that *lazy* consensus must be used
by
>> > > > *default*. By culture I don't like anything that is lazy by default
>> > > > but I can accept its use.
>> > > >
>> > > > All the rest is out of topic. Though the thing wasn't a question of
>> > > > ego. You missed the point. The problem was the lack of communication.
>> > > > But this is out of topic and I won't answer to that here.
>> > > >
>> > > > To make it more clear since you asked it. This discussion is about
>> > > > discussing the use of the lazy consensus *by default* and for me it
>> > > > should be just an option, not something use for anything. It all
>> > > > depends on the context. And in any case think more about the delay
>> you
>> > > > give depending on the importance of the decision or the urgency.
>> > > >
>> > > > To say it another way: this discussion is about the proposed policy
>> to
>> > > > use the lazy consensus *by default*. I hope it's clear now. And this
>> > > > discussion is perfectly legal imo.
>> > > >
>> > > > Voila.
>> > > >
>> > > > - benoit
>> > > >
>> > > > On Fri, May 10, 2013 at 4:48 PM, Noah Slater <nslater@apache.org>
>> > wrote:
>> > > >> On 10 May 2013 09:39, Benoit Chesneau <bchesneau@gmail.com>
wrote:
>> > > >>
>> > > >>> Though I failed in this bad (imo) habit we took recently to
>> > > >>> propose decisions before discussing the foundations of this
>> > > >>> discussion.
>> > > >>
>> > > >>
>> > > >> Not everything needs to be discussed.
>> > > >>
>> > > >>
>> > > >>> All I wanted is discussing what I considered an abuse and
>> > > >>> make some proposals.
>> > > >>>
>> > > >>
>> > > >> Sure. I've invited you to make your proposals. I really hope you
do!
>> > > >>
>> > > >>
>> > > >>> Also I don't have to give concrete examples since the problem
I
>> > > >>> describe " use lazy-consensus all the time and only  propose
72
>> hours
>> > > >>> to react" is the abuse. You may disagree with that but this
is
>> what I
>> > > >>> call an abuse.
>> > > >>
>> > > >>
>> > > >> I am asking you to provide specific examples. We can't talk about
>> this
>> > > >> meaningfully with them.
>> > > >>
>> > > >> Not only the problem is that some proposed threads didn't have
>> > > >>> discussions at all
>> > > >>
>> > > >>
>> > > >> Decision making does not require discussion. Sometimes discussion
is
>> > > good.
>> > > >> Sometimes it is needless.
>> > > >>
>> > > >>
>> > > >>> either purely or violently objected or simply ignored
>> > > >>
>> > > >>
>> > > >> Third time you say this without any evidence. Please provide
>> evidence.
>> > > >>
>> > > >>
>> > > >>> Worst case an idea/code from an ignored thread came 1 year
or
>> > > >>> 2 year after is  presented as a new thing.
>> > > >>>
>> > > >>
>> > > >> Why is that a bad thing? Stuff gets recycled. I'm grateful that
>> things
>> > > are
>> > > >> picked up eventually.(Unless your problem is with the credit.
Which
>> I
>> > > don't
>> > > >> give two shits about. That's some meaningless ego thing.)
>> > > >>
>> > > >>
>> > > >>> The problem is not to force decisions (yes I call it forcing)
by
>> > using
>> > > >>> lazy consensus without prior discussions
>> > > >>
>> > > >>
>> > > >> One of three things must be the case:
>> > > >>
>> > > >> 1) You don't understand how lazy consensus works, and so you
>> perceive
>> > it
>> > > >> as a way to force through decisions without discussion.
>> > > >>
>> > > >> 2) You understand how lazy consensus works, but you disagree with
it
>> > on
>> > > >> principal, because you believe _all decisions_ require discussion.
>> > > (Please
>> > > >> note how broad the category of "all" is in this context.)
>> > > >>
>> > > >> 3) You understand how lazy consensus works, and can see it has
>> useful
>> > > >> application, but you believe that somebody on this project used
lazy
>> > > >> consensus to ram through a decision which should have been handled
>> > with
>> > > a
>> > > >> discussion.
>> > > >>
>> > > >> Please clarify which one of these is the case, and if it is 3,
>> please
>> > > >> provide a reference to the thread where you believe this happened.
>> > > >>
>> > > >>
>> > > >>> working on taking all new ideas in a positive
>> > > >>> manner, and being open even if the idea sounds stupid at first.
>> Also
>> > > >>> listening about differences. Something that we still have
to work
>> on
>> > > >>> imo.
>> > > >>
>> > > >>
>> > > >> Agree. It would be good if we got better at this.
>> > > >>
>> > > >> That exactly my thinking about the lazy concensus *by default*:
a
>> > > >>> buraucratic crap and a way to  not share the control with
the
>> > > >>> community or make it harder to do it.
>> > > >>>
>> > > >>
>> > > >> Then I think you must misunderstand what "bureaucratic" means.
>> > > >>
>> > > >> Two possible definitions:
>> > > >>
>> > > >> 1) Making it harder for people to do things by imposing rules,
and
>> > > policy,
>> > > >> adding additional steps you must go through to get anything done.
>> > > >>
>> > > >> 2) Making it easier for people to do things by simplifying rules,
>> and
>> > > >> streamlining policy, and removing steps you must go through to
get
>> > > anything
>> > > >> done.
>> > > >>
>> > > >> Most people would say "bureaucratic" means 1. And I think most
>> people
>> > > would
>> > > >> say that imposing the requirement of discussion, followed by a
1
>> month
>> > > wait
>> > > >> period before _any_ decision can be made qualifies. And I think
most
>> > > people
>> > > >> would say that lazy consensus is more along the lines of 2.
>> > > >>
>> > > >> And this discussion make me think that my next proposal to go
to a
>> RTC
>> > > >>> policy [1] will have the same kind of reaction.
>> > > >>
>> > > >>
>> > > >> I expect so. We have version control for a reason. And from what
I
>> > have
>> > > >> seen across the rest of the foundation, RTC is imposed by sclerotic
>> > > >> projects paralysed by their fear.
>> > > >>
>> > > >> I am open to having this conversation, but I am requesting that
you
>> > make
>> > > >> things more concrete.
>> > > >>
>> > > >> Specifically:
>> > > >>
>> > > >> 1) Provided references for your statements about "certain" threads
>> > where
>> > > >> this abuse is happening.
>> > > >>
>> > > >> 2) Draft a set of by-laws that we can debate.
>> > > >>
>> > > >> --
>> > > >> NS
>> > >
>> > >
>> >
>> >
>> > --
>> > NS
>> >
>>
>
>
>
> --
> NS

Mime
View raw message