I.e. if people can only cast votes for a spec to be included and can’t vote against a spec being included, how would that work? edburns: I was thinking the ballot should consist of all the possible choices of {MVC, Data, JNoSQL, and none}, similar to thedistinguishable balls in indistinguishable boxes problem. Voters would then rank their choices and the winner would be selected. For example, {none, none, none} is a valid vote. Yes, the ballot would be long, but that’s not a problem. So the intent would be to have a single ranked-choice vote with these options? Data_yes JNoSQL_yes MVC_yes Data_yes JNoSQL_yes MVC_no Data_yes JNoSQL_no MVC_yes Data_yes JNoSQL_no MVC_no Data_no JNoSQL_yes MVC_yes Data_no JNoSQL_yes MVC_no Data_no JNoSQL_no MVC_yes Data_no JNoSQL_no MVC_no (making sure I understand before commenting) Edburns: Yes, that is correct. Then, we would run a separate ballot to decide what platform variant would have those specs.
I did an experiment ranked-choice vote with 6 votes cast, allocated as follows:
1 Data_yes JNoSQL_yes MVC_yes 1 Data_yes JNoSQL_yes MVC_no 1 Data_yes JNoSQL_no MVC_yes 1 Data_yes JNoSQL_no MVC_no 2 Data_no JNoSQL_yes MVC_yes 0 Data_no JNoSQL_yes MVC_no 0 Data_no JNoSQL_no MVC_yes 0 Data_no JNoSQL_no MVC_no
In this scenario Data would not go into Jakarta EE 11 despite 4 clear votes for it to be included and 2 votes for it not to be included. Those 4 voters simply did not agree on what other specs should or should not be included. 2 people did agree Data should not be included and did agree on the other results, so they won the vote overall.
I’m definitely a fan of ranked choice voting and *very* glad we’re discussing it. Not sure it works here as our “boxes” are effectively zero to unlimited, so the math of ranked-choices against a fixed number of boxes doesn’t really apply.
If we needed to chose say 2 specs out of 3, I’d also be insisting we use it.
Thoughts?
-David
|