What the heck is this?
You’ve found the Screenshot of the Week leaderboards. Each leaderboard ranks everybody who has entered the contest based on different criteria, such as votes, victories, and participation.
What is a member’s “formidable” rating?
The formidable rating measures Screenshot of the Week dominance. It is based on the Elo rating system and has these properties:
- Higher ratings are better.
- Newcomers start with a rating of 1,500.
- Your rating increases when you receive more votes in a contest than the algorithm expects you to receive given your rating and your opponents’ ratings. Inversely, you lose points when the algorithm thinks you underperform.
- Rating points always move from one member to another. This means that the average rating always sticks at 1,500.
- A two-to-one vote ratio corresponds to a rating difference of 500. This means that if your rating is 2,000, the algorithm expects you to receive twice as many votes as an average contestant. If your rating is 1,000, it expects you to receive half as many votes as an average contestant. (Don’t take it personally!)
- Because the algorithm only looks at votes rather than wins, it’s possible for your rating to drop despite winning a contest or rise despite finishing last.
- Ratings change quickly.
How does somebody get half of a vote, entry, win, etc.?
It’s possible for two or more members to enter a shot together. (See SotW #50 and SotW #391.) When this happens, we divide all the metrics evenly. For example, a shot submitted by two members that earns 11 votes counts as 5.5 votes for each contributor. Also, the rules of Screenshot of the Week have at times allowed ties. When a contest ends in a tie, the win is split among those who tied for the most votes.
Some members have a plus sign (+) next to their votes. Why?
The polls in Screenshots of the Week 1–4 are lost. For these contests, entrants are formally given zero votes but get preference in the leaderboards over members who would otherwise have the same number of votes.
Should I take these leaderboards seriously?
The whole idea behind this page is to offer a unique retrospective on the history of the contest for fun. There are errors on it. The metrics have flaws. The code has bugs. The answer is no: do not take these stats seriously.
How is this page generated?
We maintain text files describing contest results and member name changes. A Java applet processes the data and spits out HTML pages for you all to enjoy.
Whom should I contact with comments, questions or error reports?
|