|
|
|
|
|
|
by mikeash
5403 days ago
|
|
|
Maybe the randomization could be weighted by the individual percentages. For example, randomize if the candidates are within 1%, but give the higher-voted candidate an increased chance of winning weighted by how far apart they are. That way the difference kicks in slowly, reducing the incremental value of additional votes in a reasonably smooth fashion. |
|
|