Multiplayer Score Adjustment

UPDATED on Jan. 22 with details!

We’ve had several requests from our top players to decrease the gaps between scores. Right now, our top players simply can’t find enough similarly-ranked opponents to play against.

To refresh your memory, SparkChess uses an Elo-based system for score, but we tweaked its parameters a bit to reward players more and to encourage higher-ranked players to play against lower-ranked ones. The unfortunate side-effect is score inflation (even the official chess ratings are susceptible to this, albeit to a lesser degree.)

We’ve run a number of simulations and found a solution that would reduce the reduce the scores while keeping rankings and relative differences intact.

Have a look at the chart below:

The chart shows in blue the distribution of all scores above 1000. With orange you have the proposed score adjustment.

So, a score of 5600 would become 3100; a score of 3000 becomes 2020; a score of 2000 becomes 1550; finally, a score of 1000 remains unchanged.

This change will not affect rankings – a player currently ranked 10 will be ranked 10 after the change as well.

We don’t anticipate any downsides to this. We expect to implement this change by January 22 and it should involve no downtime.

UPDATE: The change was made on January 22 at 7 AM GMT. There was a one-minute downtime to reset games in progress to make sure all scores are consistent. We applied a more aggressive curve – the highest score went down from 5600 to 2600, then the change was gradually smaller until 1000.

We hope you’ll enjoy the added competition!

Armand Niculescu is a full-stack developer and creator of SparkChess. He has two BDs in engineering and marketing and a MSc in Management. He's been programming since the age of 12 and he also loves design, photography, and - obviously - chess.