Re: List of volunteers for the 2020-2021 NomCom

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 6/30/2020 12:54 AM, Toerless Eckert wrote:
Thanks for the analysis.

Inline.

On Tue, Jun 30, 2020 at 12:12:59AM -0400, Michael StJohns wrote:
To put some real numbers behind this I grabbed the list, and did some
consolidation and sorting.  There were basically only a few companies where
I ended up merging things:  Anything with "Huawei" in the name was Huawei.
(Huawei, Huawei Technologies, Huawei Tech Dusseldorf, Beijing Huawei, Ditto
for Google (Chrome, Jigsaw). The remainder of the list were slight variants
(Cisco vs Cisco Systems, etc) and easy to merge.  I wasn't sure of whether
Futurewei should be grouped with Huawei, so I did the calculations both
ways.

Taking only the groups with 3 or more volunteers you get:

Cisco - 13 - 8.1% of the pool - 24.2% chance of having 1, 65.6% of 2
Ericsson - 6 -   3.8% of the pool 37.5/26.9
Futurewei - 10 - 6.3% of pool 31.5/51.5
Google - 4  2.5% 34.9/14.6
Huawei - 28 17.5%  3.2/95.9
Juniper - 10  6.3% 31.5/51.5
Mozilla - 3 1.9%, 30.9, 9.1
Nokia - 3  1.9%  30.9/9.1
ZTE - 6 - 3.8% 37.5/26.9

(Huawei + Future Wei ) 23.8% .6/95.6

Using the current model, Huawei with 17.5% of the pool has only a (100 -
95.9 - 3.2)  .6% chance of having no volunteers selected, a 3.2% chance of
having 1, and a 95.9% chance of having 2.   If Futurewei and Huawei are
counted as separate companies, together they have about a 50% chance of
having 4 members between them (estimate rather than doing the calculations).

Cisco and Juniper each have more than a 50/50 chance of having two members
and a >80% chance of having at least one each.

The above numbers were calculated using the BINOMDIST function in Excel.
Whats the probability of a single individual from  e.g: Huawei+Futurewei
to be elected to NomCom. vs. the probability of an individual from
a company where he/she is the only nomcom volunteer ?

The raw probabilities are the same.  E.g. Ptot ~=  1 - ((1-1/160)^10) or about 6% chance for any individual in the pool being selected.  Its approximate because the pool actually shrinks with each pick changing the probability upwards for each subsequent pick.  For Huawei+Futurewei, let's assume they end up with two members (as that's a near 100% chance).  The Ptot for each would be ~= 1-((1-1/38)^2) or about 5%.  Not really a great difference.  (It's actually %5.19% * 95.6% or 4.96% plus ..6% * 2.6% - call it 5%).

Note:  Figure the probability of not being selected N times (which is (1-Psel)^N) and subtract that from 1 to get the probability of being selected at least once.



I saw some folks without listed company affiliations that have i think
known company affiliations. Whats the process ? Is NomCom chair
having to try to figure that out, best effort ?
I saw three I think.  One from Cisco, one from Huawei and one from Cloudflare.  I expect those will be filled in.

Obvious, and this is the root of what i think is our problem dealing with
candidate bias: There are no simple critera to demonstrate other than
company affiliation clusters of bias that would imact the quality of
the nomcom results. So big companies are the easy pinata.

Cheers
     Toerless

Mike






[Index of Archives]     [IETF Annoucements]     [IETF]     [IP Storage]     [Yosemite News]     [Linux SCTP]     [Linux Newbies]     [Mhonarc]     [Fedora Users]

  Powered by Linux