Well, then, I guess the voting format remains the same as it was before. I'll keep the "Best Concept" name, too; maybe as one of additions in the Jam player, I'll add brief descriptions for best-of awards, too.
With that said, the last matter to settle before the next Jam...
...the
voting system!
And then all hell broke loose...
--------------------
The voting (tallying) system determines how the game rankings should be tallied to final results. Current system is the
normalised 1/(rank+1). It means first 3 places take 1/2, 1/3 and 1/4 score respectively, and so on until the last ranked place 1/(n+1), and all remaining games get the score of 1/(n+2). The games are then sorted by the total score.
The advantage is that the system is relatively easy to understand and check; basically, one can make a spreadsheet with scores breakdown and if someone understands how the system works, they can easily verify the results on their own.
Also, it reduces the adverse effects of people completing and voting only on a few games, including some joek game; under non-normalised 1/rank (where unranked games get no score at all) it would give considerable
boost to the joke game, likely contrary to the voter's intent who wants the joek game to score as low as possible.
The normalisation however has a drawback - it results in shorter rankings giving somewhat smaller "voting power". In particular, in rankings with top 3 entries only, the third entry gets measly 0.05 worth of score above unranked entries, in contrast to nearly 0.25 score it would get if the ranking was full. Also, the first place would get 0.3 points above unranked entries instead of nearly 0.5.
From what I recall, across the various Jams the following other systems were considered (and some actually used):
- plain "game with most 1st place votes wins"
- top-3 scores summed up (e.g. 1st place gets 7 score, 2nd gets 5 score, 3rd gets 3 score, the game with highest score wins)
- 1/rank (1st place gets 1/1 score, 2nd gets 1/2 score, 3rd gets 1/3 score and so on, unranked entries get 0)
- 1/(rank+1) (1st place gets 1/2 score, 2nd gets 1/3 score, 3rd gets 1/4 score and so on, unranked entries get 0), to soften up the effect of random rankings (e.g. giving 1st place to friend or something)
- repeated
instant-runoff voting (i.e. pick 1st place entry using IRV, add it to the end of the ranking, remove the entry from rankings, rinse and repeat on altered rankings until there are no more entries)
- mIRV - a variation on repeated IRV, but I have no idea how it works exactly (apparently it's meant to allow games with lots of 2nd place ranks but no 1st place win against game that scored 1st place in few rankings)
-
ranked pairs - it's a cool system, but tricky to understand and has fancy maths going on that goes over people's head; it promotes entries consistently scoring well rather than controversial ones
Another system I considered would be "repeated 1/rank" (or "repeated 1/(rank+1)"; heck, maybe "repeated 7/5/3" would work, too). Basically, pick 1st place using 1/rank, add it to the end of the ranking, remove the entry from rankings, rinse and repeat on altered rankings until there are no more entries.
It allows games with lots of 2nd place ranks to win against those with few 1st place ranks, and the repetition should eventually even out random votes (such as dubious 1st place vote or joek game 10th place vote, because the voter had time to play only 10 games). At the same time, it doesn't lessen the support for voter's picks if the voter ranks few entries (in contrast with the current system).
This system might be slightly damaging to controversial entries compared to single-step 1/rank variations, but not as much as ranked pairs.
The repeated IRV, mIRV, ranked pairs and repeated 1/rank systems cannot be presented in a fancy easy-to-verify spreadsheet, but it's not a problem for me personally. Also, I'm able to make and share a program automatically tallying the results using all the mentioned systems except for mIRV (because I have no idea how it works exactly).
--------------------
One more thing:
Should voting be awarded by 1st place's worth of ranking? In other words, if at least one of authors of the given entry casts their vote, should the entry get a bonus 1st-place vote to its advantage?
The rationale behind it is that since you can't vote for your own entry, the very act of voting puts you at a disadvantage. 1st-place vote award makes the voting actually beneficial and might encourage more people to vote (who generally wouldn't mind voting, but care about their rank, too)?