Gunnar Wolf <gw...@debian.org> writes: > If this is right, Sam, let me politely ask you to unbundle. Not only due > to Martin's argument (the scar of "editorial changes" we all had to > endure and understand a little too late), but also to keep each of the > choices as simple and clear as possible -- and to avoid > combinatorics. And even clarity!
> Say, now that Judit added a ballot option that works targets only one of > your concerns (voter secrecy). This option does not consider 3 and 4. > Suppose no other options are present. Judit's option wins, yours is > second, and NotA is third. A simplistic reading would mean, "merge > Judit's proposed changes in the constitution". However, more people > voted 3 and 4 above NotA -- Shouldn't they also be included? Do they > warrant a separate GR now? > Or should the GR have now four options? (Sam's original, Sam's minus 3 > and 4, Judit's original, and Judit's plus 3 and 4) I believe that the combinatorics (putting each possible combination on the ballot) is the correct approach given our voting system and given the range of possible opinions. To see why, suppose there is a voter who is happy with private votes provided that the secretary decisions can be overridden, but if secretary decisions cannot be overridden, they do not want private votes. If the two votes are unbundled, that voter cannot vote their preference, and their only option is to either add a new ballot option on one of the votes to do both at once or vote both below NotA. If all three options (secretary changes only, private vote only, secretary changes plus private vote) are on the same ballot, that voter can then accurately vote their opinion by putting the last above NotA and the second below NotA. The point of using a clone-proof voting system is to allow us to capture preferences like that by feeling free to add additional options to the ballot. I completely agree with separating *unrelated* changes, but the whole point of this discussion is that some folks believe the changes are closely related, to the extent that one of the changes may not be desirable unless the other one is made at the same time. -- Russ Allbery (r...@debian.org) <https://www.eyrie.org/~eagle/>