Outlier votes

As I’m in a three-way-tie (for 22nd), I’d like to ask for more digits. I can take being 24th if that’s the price of eliminating the uncertainty.


I have good news – and I have bad news. :smiley:

You don’t end up in the 24th place. But you still end up in a tie, only now it’s a two-way-tie. :face_with_diagonal_mouth:

With three decimals, the scores are:

  • 6.429: Crash (22nd place, tie)
  • 6.429: Lazy Wizard’s Guide (22nd place, tie)
  • 6.425: Cannelé & Nomnom – Defective Agency (24th place)

OK, how about four or more decimals? Actually, it doesn’t matter. The scores for the first two games are exactly 45/7 = 6³⁄₇ ≈ 6.428571.

The last score, for Cannelé & Nomnom, is actually a bit interesting from a numerical point of view. The average rating is exactly 6.425. How do we round this to two decimal places? Different software use different rounding rules when the last digits is 5. Most use IEEE 754’s default ruleround to even (also called the banker’s rule). Then the score would be 6.42. And the game would end up in 24th place. So even with the use of only two decimals, there would be a two-way tie instead of a three-way tie for these games.

But the IFComp scoring algorithm seems to round up instead. And that’s why we got a three-way tie. :smiley:


Wow, I haven’t played Lazy Wizard’s Guide yet. I need to, so I can see exactly what it’s like to play my game.


like skating, throw out the highest and lowest

1 Like