What is the ChessTweets Experiment?
When Kasparov faced the world in 1999, Kasparov was considered a "prohibitive favorite" because previous examples of majority Internet voting had produced mediocre competition. The reason for these poor performances is readily apparent: the most popular chess move for a given board state is not necessarily the best. The world team overcame this handicap against Kasparov by appointing four grandmasters to guide the popular vote and enforce vetos when necessary.
The ChessTweets Experiment attempts to expand on the collaborative ideas intended by this famous chess game by creating the world's first automated and objective hive-mind machine. With the advent of Twitter, developing such a machine has become both readily-possible and irresistible. Can such a machine learn to compete with the best of the best? The ChessTweets Experiment intends to find out.
By asking its participants to give their input in community games and without preselecting grandmasters, ChessTweets will examine each participant's relative skill and apply a formulated weight to each and every suggested move such that every mind plays an important and unique role.
As more and more information about its participants becomes available and as tiers of skill-levels distiguish themselves, the hive-mind will grow and learn in an artificial-intelligence-type manner. If the machine develops with any measure of success, it will beg a greater question: when given a method for weighting an individual's performance at answering closed-information questions (those whose answers depend on hidden or unknown information), can a similar hive-mind machine produce optimal results?
How is relative skill calculated?
A great question! Although we have several thoughts of our own on the matter, we don't yet know. For the time being, ChessTweets is excited to be collecting participant data. A major part of the ChessTweets Experiment will be determining how to best manage the data we are collecting. If you have any thoughts or ideas for how an individual's move suggestion should be weighted based on the following information, please contact us at :
- Number of times participant has suggested a move in a community game that was both selected and (eventually) led to a win.
- Number of times participant has suggested a move in a community game that was both selected and (eventually) led to a loss.
- A pariticipant's Elo Rating in private games.
A note on Elo Rating: Although Elo Rating has proved an effective means to rate chess players for competitive play, ChessTweets is brainstorming a new rating system for its purposes, specifically one that doesn't incorporate diminishing returns against weaker opponents. Because the quest for data is ultimately more important than any individual participant's concern for his own rating, a rating system that encourages all types of pairings and community participation may be better suited.
What is ChessTweets' stance on the use of Computer Chess Programs?
As soon as Elo Rating became an important factor for rating chess players, individuals began seeking better ratings through the use of computer chess programs. While ChessTweets doesn't forbid (or for that matter care about) the use of such programs, it is important for the ChessTweets community to have a consistent participant base.
We want players to enjoy participating in the ChessTweets project, and we want everyone to continue their participation. With that in mind, ChessTweets has the following suggestions to offer for both optimal enjoyment and personal growth:
- Separate yourself from any numerical rankings ChessTweets may assign you. Members of all skill levels are very important to the ChessTweets Experiment.
- Avoid computer chess programs in favor of a natural study of the game.
- Challange friends (and strangers!) to personal games regardless of separation in skill level.
If you can gain more enjoyment out of the ChessTweets experiment by following a different set of guidelines, by all means do! We only ask the following: Be consistent in your play. Participants who play at largely variant skill levels because half the time they consult a chess program and half the time they play at a beginner's level are most detrimental to the experiment.
How long do I have to make a move?
Currently, moves in public games are tallied manually on a rough 24-hour basis. Once we hit a critical mass, moves will be tallied by some automated time schedule, whether it be exactly once every 24 hours or after x valid move votes have been submited. Our current goal is to grow to a point such that public moves are made every 100 vote submissions.Inactive private games are kept on ChessTweets for 10 days. If a game is inactive for longer than this period the game will be resolved in one of two ways:
- If the game has progressed less than 5 full turns (10 valid moves by both White and Black), the game will be deleted with no stat collection or elo rating changes.
- If the game has progressed more than 5 full turns (10 valid moves by both White and Black), the inactive player will be forced to resign, and the corresponding stat collection and elo rating changes will take place.