[vote] Update OpenWrt rules
David Woodhouse
dwmw2 at infradead.org
Tue Nov 4 07:24:56 PST 2025
On Tue, 2025-11-04 at 15:14 +0000, Daniel Golle wrote:
>
> Example:
>
> 9 votes for (A)
> 4 votes for (B)
> 8 votes neutral
>
> In the currently proposed model, this would result in
>
> (A) 9 + 0.5 * 8 = 13 (61.9%)
> (B) 4 + 0.5 * 8 = 8 (38.1%)
>
> While the ratio of the non-neutral votes was actually
>
> 69% (A) vs. 31% (B)
>
> If this had been a decission which required a 2/3rd majority, then
> the neutral votes would have changed the outcome in this case.
>
> What we can do is simply not count them but assume quorum is
> reached once enough members indicated their mere awareness of the
> voting process.
>
> What Adrian probably means is that a "neutral" vote should be
> counted as 0.69 for (A) and 0.31 for (B) rather than 0.5 for each.
> In this way it would not change the outcome at all.
Yes. But I don't think we can use (A) and (B); it isn't multiple
choice. I think you meant *yes* where you said (A) and *no* where you
said (B). The two-thirds threshold is important in the yes/no model.
We should *discount* the neutral votes and count that the 9 yes votes
are 9/13 == 69.2% of the total yes/no votes.
The more confusing way of doing it is to keep the neutral votes, and
count them as 2/3 of a vote. So it's 9 + (8 * ⅔) == 14⅓ out of 21,
which is 68.3%.
You don't have to count the neutral vote as *precisely* the same as the
ratio between yes/no, because you don't have to avoid affecting the
*numeric* result. You just have to avoid taking it across the 2/3
threshold in either direction, so a neutral vote should mathematically
be 2/3.
Congratulations, you found an even *more* confusing way to do it :)
We should just discount the neutral votes when evaluating the yes/no
against the 2/3 threshold :)
-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 5069 bytes
Desc: not available
URL: <http://lists.openwrt.org/pipermail/openwrt-adm/attachments/20251104/f505c564/attachment.p7s>
More information about the openwrt-adm
mailing list