Round 4 - Most Voted

  • 3
  • Article
  • Updated 8 years ago
Here are the currently highest voted lab submissions, with my control score.
I'm just posting these up now, so then I can measure afterwards how well my score correlates with the actual results.

Photo of Berex NZ

Berex NZ

  • 116 Posts
  • 20 Reply Likes

Posted 8 years ago

  • 3
Photo of Adrien Treuille

Adrien Treuille, Alum

  • 243 Posts
  • 33 Reply Likes
How did you calculate these values, or is it top secret?
Photo of Adrien Treuille

Adrien Treuille, Alum

  • 243 Posts
  • 33 Reply Likes
Also, since you're doing this service to the community, I would be interested to hear what you thought about our proposal for alternative voting systems.
Photo of Berex NZ

Berex NZ

  • 116 Posts
  • 20 Reply Likes
Hehe it is top secret, but very simple to reverse engineer.
Basically it measures each sequence based on nucleo and sequence number.
Comparing it against the top two designs and then averaging out a number Which can be used to score how SIMILAR it is to the two top sequences.
I could also make it measure against any number of sequences that I choose.

I don't have enough information as to how you plan to implement Elo so I can't really evaluate it. Initial impressions, it depends on how you implement the K factor. Personally if we're going for the Elo system, it might pay to look at Microsoft's TrueSkill algorithm. Its basically the same as Elo but with an overlay to ensure a more even playing field.

Are you planning to apply Elo to the players or to the submissions themselves or a mixture of them both?
Photo of Berex NZ

Berex NZ

  • 116 Posts
  • 20 Reply Likes
Ok well now that the results are out.

Here is the final tables.
The first one is in the original format.
The second one is descending by final score.





So obviously it works for some designs and it isn't perfect. But its a start :)