I'm writing a master's thesis about the consistency-responsiveness trade-off problems related to multiplayer online games. Thesis has now proceeded into the user testing phase, where potential testers are needed. So I decided to try whether I could reach a good amount testers via this forum.

The test is implemented as a simple ping-pong game, where each player plays 4 rounds. Each round ends when either the player receives 10 points or the time limit of one minute is reached. Each round uses a different consistency maintenance technique, which typically affect the game somehow. After each round, the user will be queried about how she/he noticed these possible game-play anomalies.

Testing will take approximately eight to ten minutes. Each player may participate the test with their own computers by using their web browsers to navigate into the URL http://cr-tradeoff.cs.hut.fi . No additional software installations should be required. Players may use a keyboard to control their avatar and to provide answers to the queries. It is therefore preferred to use an PC while participating.

Server will be approximately online due to next Wednesday (13.01.2016) and it should be reachable around the clock.There's no need to register for the test and all players may participate anonymously.

Hope that as many of you as possible would take a part of this test! Thanks!

BR,

JT

Interesting! Would love to see the results for the different compensation models.

A bit more about the test implementation...

When a new game session is created, the server assigns a session delay from the 4 available delays (0ms, 25ms, 50ms, 125ms). This delay is used to hold incoming and outgoing messages for the specified time before they are processed. Same delay will affect the whole game session i.e. until all rounds has been played.

Techniques are selected in a random order from the four available techniques (pessimistic-presence, pessimistic-delay, optimistic-presence and optimistic-delay). Optimistic techniques are able to provide instant feedback to users while pessimistic techniques use either event serialization or event delaying before the feedback is provided.

As swiftcoder said, some of these techniques may become quite challenging to play with high latency values. For example, the implemented serialization will move the ball into new direction in a warping manner as the delta time from the previous update iteration may grow quite high. These warps may easily make a real-time simulation game to became a bit annoying if not impossible to play. However, I'll try to post some of the test results here as well when the testing part is over.

It would be great if we could get more people to participate as the current amount of results is still quite low.

Server online time is now extended to the end of the month.

That's a good note from Waterlimon. The ball and the controlled (i.e. owned) paddle could be considered as the primary focus points within the scene as they are the only entities which can be directly interacted by the player. This may make player to easily ignore corrections that are applied to the remote paddle positions. It could be avoided by creating a game where the players may interact directly with each other, like with a small FPS or such. However, this is something that is also considered when the results of the testing phase are studied.