ietf
[Top] [All Lists]

Re: 6tsch BoF

2013-08-01 06:17:45
On Thu, Aug 1, 2013 at 3:04 AM, manning bill <bmanning(_at_)isi(_dot_)edu> 
wrote:
we have never voted at IETFs.
"we believe in rough consensus and running code"


We are not voting.
We are expressing agreement with a specific assertion.
That is true whether the agreement is expressed via vocalization
or motion of limbs.

Humming allows the chairs to be totally biased and interpret the hum volume
in a way that aligns with their preference for the outcome.

However, I have seen many times where the chairs clearly miscounted
the number of raised hands in order to make the outcome align
with their preference.  However this is much harder to pull off than comparing
2 similar hum volumes.

The chairs can pick however they want to measure agreement.
Many chairs ask for a show of hands.  I prefer that method.
There is still a judgement call when the numbers for and against
are not significantly different.

/bill

Andy


On 1August2013Thursday, at 2:14, A ndy Bierman wrote:

Hi,

Isn't it obvious why humming is flawed and raising hands works?
(Analog vs. digital).  A hand is either raised or it isn't.
The sum of all hands raised is comparable across tests.
The sum of the amplitude of all hums is not.


Andy

On Thu, Aug 1, 2013 at 1:50 AM, Ralph Droms 
<rdroms(_dot_)ietf(_at_)gmail(_dot_)com> wrote:

I found the process in the 6tsch BoF (Tue 1520) for asking about taking on 
the work discussed in the BoF to be thought-provoking.

Toward the end of the BoF, the chairs asked the question "1. Is this a 
topic that the IETF should address?"  First, the chairs asked for a hum.  
From my vantage point (middle of the room), the hum was pretty close to 
equal, for/against.  I reviewed the audio 
(http://www.ietf.org/audio/ietf87/ietf87-bellevue-20130730-1520-pm2.mp3, 
starting about 1:22), and heard a slightly louder hum "for".  The BoF 
chairs decided they needed more information than they could extract from 
the hum, so they asked for a show of hands.  From the audio record, there 
were "a lot" for (which matches my recollection) and "a handful" against 
(my memory is that no hands showed against).  There was a request to ask 
for a show of hands for "how many don't know".  The question was asked, and 
the record shows "a dozen".

So, there was apparently a complete change in the answer to the question 
based on humming versus voting.  There may also have been some effect from 
asking, after the fact, for a show of hands for "don't know".

I'm really curious about the results, which indicate that, at least in this 
case, the response to the question is heavily dependent on the on the mode 
used to obtain the answers to the question (which we all know is possible). 
 In particular, the effect of humming versus show of hands was pretty 
obvious.  draft-resnick-on-consensus gives several reasons why humming is 
preferred over a show of hands.  From this example, it seems to me to be 
worth considering whether a more honest and accurate result is obtained 
through humming rather than a show of hands.

The other question raised in my mind is why the initial result from the 
hum, which did not have a consensus either way, was not sufficient.  
"Roughly the same response" for/against the question would seem to me to be 
as valid a result as explicit consensus one way or the other, and the act 
of taking a show of hands to survey the appeared to treat the hum as 
irrelevant, rather than highly significant.

- Ralph



<Prev in Thread] Current Thread [Next in Thread>