[Eligibility-discuss] Re: [Gendispatch] Re: Re: Re: Comments on Composition and Comportment of the IETF Nominating Committee draft

[Joel Halpern <[e-mail]>]

May I suggest structuring the algorithm around drawing from the common 
pool , noting how many people from the alternate pool are selected.  
Stop drawing from the common pool when the number of remaining seats to 
be filled is equal to the number of additional alternate pool 
participants that are needed.

For example, using the old "we need 1" goal, one would fill the first 9 
slots from the common pool.  If noone from the alternate pool was 
selected among those 9, then the last one is from the alternate pool 
rather than the general pool

Yours,

Joel

On 8/18/2026 4:20 PM, Tara Tarakiyee wrote:
> No, that is not what I was expecting. Thank you for pointing it out, 
> the overlap between the two pools is doing something I did not intend. 
> I will look into correcting the formula.
>
> Best,
> Tara
>
> On 8/18/26 9:42 PM, Salz, Rich wrote:
>>
>>  *
>>     Joel: on the upper end, the 5-slot case only arises when the pool
>>     is at or near parity, and at that point 5 is approximately the
>>     expected value of the flat draw, so the formula is not
>>     intervening at all.
>>
>>
>> I don’t think that’s accurate. Replace ’n’ with ’10’ and simplify:
>>
>>       r = min(p, max(3, min(5, floor(10 * p / t))))      
>>
>> So the range is [3..5] or one-third to one-half depending on the size 
>> of the opt-in pool as you say. But then we fill up the rest of the 
>> seats, five in this case, and the total count increases  (call t=100 
>> and p=50) then the true representation will be 5 + [2,3]. Is that 
>> what you were expecting or want?
>>
>>
>