I think there should be two options added to the game. A draw button (so kestas doesn't have to do them manually) and a end game button (for the scenario described above)

I would be very strongly opposed to having a computer enter moves, however. First off, I believe it goes directly against the rules (although I'm not positive) secondly, it adds an uncomfortably large portion of luck to a game that should be based on skill.

If a player does go CD that soon in a game, i'd suggest holding until a replacement could be found.

The idea of the artificial intelligence is something I would like to see developed so that players who have a legitimate NMR are not destroyed by it.
For example when I ran a 7x Gunboat tournament three of my players wound up in emergency rooms and hospitals overnight on some rather serious issues/auto accidents and the like. I run with the Easy Does It System where if there is a NMR I put a set of orders in for them and then find out what happened. A machine A/I is being developed so that does not have to be done in the future. However, it really saved the game for everyone involved, the players of course being a Gunboat game had no idea that the orders were done by "A/I" and all three players were very appreciative that their whole position did not crumble and the other players did not feel like the air was suddenly sucked out of the intensity of the game.

Personally for 1901 problems I strongly suggest that the game simply be stopped and restarted with a new player BEFORE any orders are processed.

I agree that the game should be stopped if someone goes CD that soon, but as that is not an option (that I know of) the best thing to do is hold.

I can see where you're coming from with AI if you're playing in a more timing consuming event, but it's so easy to create games here, that I think it would take away from the game by adding an AI. Also, you may run into a problem with people who understand the algorithm for the AI better than others having an advantage.

The whole point is to create a philosophy of play...maybe an ethic of play is better... wherein we develop the concept that the play comes before the game. We play to have fun and make it fun for others. When the game is broken from the start, we put the pieces away and start another game.

We've talked about setting moves similar to those for CD countries. Or at the very least setting up support holds as much as possible.

Of course, when last discussed the adjudicator wasn't 100% compliant - so Kestas had plenty of higher priorities.

OK, so I just learned what a gunboat game is. In that case, I definitely agree with having an AI stand in. Otherwise, I'd be against it.

These things are to be expected in low-points games, the idea is that once that happens people can gain points and make it to higher points games.
The higher the game you enter, the less likely it is you're playing with someone who's going to leave; people who leave go down, people who don't go up. This is true whether or not people find points important

So unfortunately you do have to endure games with players who are new to the game or likely to leave to make it to the better games, but that's the cost of being such an accessible site

Regarding AI: Diplomacy AI works like Chess AI; orders are generated and adjudicated in a "brute-force" sort of way, and after each adjudication an algorithm determines whether the units would be in a more favorable position. This process takes a very long time computationally, especially across so many games, which we can't afford to do on shared hosting, especially using PHP. It's also complicated to program, and AI is only a reasonably good substitute in gunboat games

So the problem of early drops has been around.
The idea that having to endure the early drop outs to get points to get to the higher point games is a valid approach. How has it worked out over all?
If we took games with players with 200 then 300 then 400 points and looked at the games with that minimum do we see a significant change in the NMR and the Drop out rate of players? I don't know but it would be cool to be able to see.

What other ways have people tried to cut down on the drop out rate?
This way the hobby can avoid those poor approaches.
Where do the drop out's go? I wonder how many can be brought back to the hobby? Hmmmm....just thinking out loud.

I think it's inevitable that on a site where it is so easy to register (a good thing) and so easy to join your first game, or your first twenty games (a good thing), there will always be curious visitors joining out of idle curiosity, and then moving on when they lose interest, which might be as soon as they realise that nothing's going to happen in their games for several hours, or as soon as they realise that they're never going to pwn Europe unless they read the rules. Where do they go? I don't know.... shopping? Clubbing? www.nudenakednurses.com? Who cares? How many can be brought back to the hobby? Very few, hopefully....

I would oppose the use of AI to solve this 'problem', even if it were practicable. If a foreign power is going to just roll over and die, I say, let them.

Yes, I know that it can be frustrating to play in games where half the players immediately disappear, especially when this is to an adversary's advantage, but ultimately, it means that a dedicated newcomer can acquire a reasonable number of points just by sticking at it, and then move on to take on the big boys. I've played with at least a quarter of the sites top 100 ranked players, and I've found an enormous difference in the standard of play between them and the newbies. I can't show you statistics to prove that higher point games have a lower drop out rate, but I have no doubt that that is the case.

> How has it worked out over all?
It has worked out very well; if you think it's bad now it was terrible before the points system came in. People would join a game, hope it'd go well, and if it didn't leave and join another

The threshold for fairly reliable games is in games where the bet is over 100 points, because at that level all players have to have had at least some success in previous games
The best way to look at this is to see the difference between the total amount in the pot, and the number of bets * 7. If the amount in the pot is much larger then a large player must have been taken over from civil disorder

ROUND( AVG( g.pot ) ) ROUND( AVG( m.bet ) * 7 ) Stats
Under 100 points:
169------------------------ 129 ------------------------------Left:2% Defeated:18% Playing:80%
Over 100 points:
982--------------------------- 976 --------------------------------Left:0% Defeated:16% Playing:84%

So in the >100 points games the difference between bets and pot size is tiny, but in the <100 points games it's much larger. Even in the sub 100 points games the CD players do get taken over, leaving the average number of left players at 2% which isn't too bad.

Below 100 points it's not so clear cut. At very low points levels the left rate is higher, then it decreases, but it gets higher again when it gets to 100 as some new players bet their starting 100 points in one game

> What other ways have people tried to cut down on the drop out
> rate?
I've found the problem exists on all phpDip based sites, and less on the less accessible sites and e-mail based ones. I expect stabberfou suffers less from this problem because games are created manually on request, and e-mail based ones don't have such a problem because you generally need to own your own board and already know the rules to play.

There definitely isn't a way to eliminate the problem entirely: The pro is that we get new players here who have never heard of the game, who may develop into better players. The con is that not all players will want to develop into better players, and will leave
I don't think any technique can possibly keep one without losing the other

> Where do the drop out's go? I wonder how many can be brought
> back to the hobby?
I think most of the sign-up-&-drop-out players weren't part of the hobby to begin with; they just wander in, decide to give it a try, and don't find it to their liking. That's probably the main difference between the recent web-based sites and the older ones