Rules for Online Competition

[Javaness2]

To remove some of the traffic from that other thread. Imagine you have a professional match which is played online. That is, from the private houses of the player's themselves. How would you verify that each game would be conducted correctly? There are two different scenarios you can plan for.

• Directly moving on a go server.
• Relaying moves to a go server.

I think that in the first, you can have
Nominated Observer: (at least 1) whose job is to ensure a player does not cheat, and to watch the clock during byoyomi.
Referee: To be available to handle disputes.
Appeals Committee: To be available at a later date in case of protest at any decision by Referee.
No undo. Since nobody can verify if a misclick is genuine or not.
Possibility to claim lag during byoyomi. Claims must be made immediately. Acceptance will equate to a restoration of the period.

[EdLee]

In chess, do they play pro games online ?

They have a head start with super-human bots (a long time before AlphaZero-chess);
they know a thing or two about online chess games of any 'significance'.

Observer ...to ensure a player does not cheat

How ? Fly a (noisy) drone 360-degrees in each player's room for the entire game?
Battery is already an issue, then (+ the noise).
(Remember that even in face-to-face pro chess tourneys, people tried hand signals, a phone in the bathroom, tiny ear pieces, etc.)

[Laman]

Javaness: mostly as you say...

for a serious online tournament, you need a trusted observer on site to prevent cheating. then it is not difficult to use a software client keeping each player's time locally. this means lags handled automatically. and it is nice to the players to alleviate their technical responsibility for the match. any issues still arising go to a referee who must be reachable during the time of the game.

i can see one loose end: how to handle long disconnections (and in a similar vein are players coming late for the game). i'd suggest forfeiting the game after 15 mins offline, probably cumulative. ideally measured automatically by the software and confirmed by the referee, alternatively measured and decided by the observer on site.

No undo. Since nobody can verify if a misclick is genuine or not.

this, or the KGS style - a player can ask for undo, the opponent has to confirm.

i think my questions regarding the observer's job raised in the previous thread are also relevant here

[gowan]

Allowing undo under certain circumstances would be OK. Often a misclick results in an obviously misclicked move, especially for pros. Perhaps misclicks should be claimed immediately (5 seconds?) and be judged by the referee.

Would Canadian overtime reduce the number of problems with overtime? I like it better in general since it allows a player to accumulated some time for thinking during overtime.

[Javaness2]

On KGS you have the option to add 1 minute to your opponent's time.
In a competition where you are using 1 minute byoyomi periods, this enables you to restore a lost byoyomi period.
The exception would be the final byoyomi period, because there KGS will say your opponent has lost.
If Skype was open, it would allow player's to shout LAAAAAAAAAG and have that verified by the Observer.

If you are using Canadian time, you can not restore a byoyomi period which your opponent has lost via lag.

[jlt]

If Skype was open, it would allow player's to shout LAAAAAAAAAG and have that verified by the Observer.

Unless Skype itself lags.

[Laman]

The exception would be the final byoyomi period, because there KGS will say your opponent has lost.

which is also the most important period

If Skype was open, it would allow player's to shout LAAAAAAAAAG and have that verified by the Observer.

this is a solution i certainly wouldn't like to see in a professional competition . better to let them lose on time.

[Javaness2]

Agree. Losing by lag in the last period is a catastrophe. On the upside the probability of losing a period due to a lag is low. MS lost 2 out of hundreds. Non a negligible number, but small. I can't see a more practical approach to take.

[Uberdude]

Without going fully "playing on real boards and clocks and just relay to KGS" I would suggest a halfway-house like so:
- Main time is played on KGS and KGS keeps the time. So if you lose a few seconds here and there then you will get less main time than your opponent, maybe 58 minutes vs 1 hour. A little unfair but it'll do. Could increase the main time for both as well, 1 hour is pretty short for a game of this importance (compare to EGC main tournament games with 2+ hours each), though I appreciate with timezones there's a limit of reasonable total game length.
- Once player enters byo-yomi a local timekeeper hits a real clock with 5x60s to start ticking when opponent moves appears, presses it again when local player clicks on board. If lag means KGS thinks they lost a period but local clock says within 60s then opponent adds time on KGS to give period back (this is a bit annoying as interruption to concentration, communication is via proctor/timekeepers and referee)
- Player loses on time only if all 5 of the 60s on local clock are gone. If KGS loses on time due to lag in final period then instant resume (review mode, set players enable this from last position).

[hyperpape]

For professional play, I would suggest a trusted observer, and a purpose built server/client that uses the players' computer to keep track of time. The standard arguments against keeping time on a player's computer are a normally issue because there is a lack of trust--given a server like KGS, there will be cheaters. But with a professional player and an observer, I wouldn't worry about it.

Of course, it would take some time to create such a server/client. Not a tremendous amount for the backend, but getting a reasonable timekeeper and board for the client seems like a bit of work.

[Bill Spight]

I know that I have expressed my opinion elsewhere, but speaking just for myself, for an amateur tournament I would say, "Whatever. Let's have fun." If I were a professional, though, I would not accept conditions where net lag was a potential problem.

[sybob]

(A) Separate line(s) of communication between all involved.
To be agreed by all on beforehand.
Phone/mobile/chat/skype/messenger, I don't mind, but outside of the go server where the game is played.
To immediately report any questions or disputes.

[vash3g]

The Pandanet AGA City League has many professional matches, each month for seven months. We have AGA and Chinese pros playing. Our rules have been updated each year to fix issues with the initial ruleset. We have played on Pandanet and advocating the GoPanda2 client since the tournament started. I have been the TD for the tournament online and face-to-face at the finals in person. I have dealt with the issues of different times and clocks, cheating issues, and player/captain complaints.

In the coming year we will be mandating a video connection for the players to cut down on possible cheating attempts. This is in line with the AGA online tournaments that have happened in the past. With the inclusion of very strong AI, I think this has been a detriment to amateur play. It is very difficult to load a game into a Lizzie/LeelaZero based application and perform analysis. It may be possible when consulting with a professional player but there are even fewer very strong players who are TDs. The issue with using this type of system is when one player starts to play music from their PC and it bleeds through to the other side. The same could be said for playing in a public area with that transmitting to an opponent.

The only actual way to verify a professional match is in person. If it is online then it does not matter what platform is used to play on. There is an understandable level of risk that must be accepted with each platform. A proctor is a good thing in each place as they should verify the player is not cheating. When using game recorders in the AGA at go congress they are instructed that they are impartial observers. They are not allowed to touch the board/clock/players during the game. If the proctor is replaying the moves then there needs to be additional time controls given when there are fast sequences in the game, a proctor will not play the same speed as if they played the opponent directly. This also introduces an opportunity for error with misclicks from the proctor.

In the City League there is a TD and any issues that I need assistance with go to the AGA National Tournament Director and generally others involved with the tournament are consulted. Generally the issues do not get that high except for FYI emails and policy questions.

[Elom]

I haven't much knowledge of the other thread , but it just happened that this situation occured when I was worldbuilding to the principle of going back to go forward. The thought was that in a couple of decades, serious online tournaments may occur in 'Rooms of Profound Darkness' (the name of the nihon-kiin's online sever!) inspired by anechoic chambers (these would be an-EM), perhaps preceded now by proctors. An online game would be treated as two simultaneous boards; simply speaking, an issue on the device of one would often result in a win and a draw (0.5 and 1 maybe added together for a 1/3 and 2/3 result).

On the note of this being the first run, I still feel you'd maybe want as many issues and altercations to occur in earlier events as possible; so if there are any problems that could occur, it would be great if they do so in the first two and brilliant if in run number one. So this is perhaps best interpreted as positive development; our tendencies to argue mean if go was more developed and we hadn't practiced this ruleset it could've been quite catastrophic. Counter fire with water and the like, always, and never with fire unless your burning soil and hay so wildfires can't spread in the future.

You might assume that in the first run of an online tournament, issues would occur that were not anticipated. This could mean that rulings would lean towards generalisation; if a witness was there for one job but it was discovered that they could do well for another, the rule can be generalised. It seems rather harsh on the organisers and volunteers to have exacting job descriptions on the very first run of a tournament they've never tried before. . . Is it not best to treat this as an experimental trial considering the circumstances? Or weight the reputation of the individuals more (this couldn't have been the worst choice of person for this to happen to )? (But it makes sense; considering how quickly the EGF could be critised even after working hard to organise this tournament for us, they probably subconsciously leaned towards easing responsibility away from themselves; hopefully it has not backfired. Helpful ideas beat critique, everyone). However, it is odd that of all things, lag was not anticipated for, so this is an unexpected expected and according to science this is then categorised as funny; so it won't likely happen again! And again, their missing something simple likely came from focusing so much on delivering this for us. . . So thank you to them, may they welcome this pain .

[Elom]

Along with the AGA and EGF theme tune, this altercation could herald the start of the Eumericas Cup.