Jump to content


Photo

Looking Forward to an Improved Singleplayer Game!


  • This topic is locked This topic is locked
102 replies to this topic

#41 maxroelofs

maxroelofs

    Major

  • Dutch Tournament Manager
  • 1,042 posts
  • Coat of arms
  • Platinum Marshal

Posted 03 December 2014 - 08:18 PM

All the progress sounds good so far! I have two think to mention tough:

 

1: Vincent de Boer once wrote a program for a computer stratego player. It was able to beat mid ranked players. Because he has so much trouble with this, i doubt it that the programmers on this site have been able to create a even stronger computer player, which is difficult for all.

 

2:

 

''Do not worry Gary! We've ensured that everyone on the Stratego-team is now an excellent Stratego-player and knows the game inside out.''

 

How long has the team been working on this? I think it is not much longer than a month or so. With all respect, i think you are underestimating the game then. It takes years to become an ''excellent'' stratego player, and there are only a few on this site, i want to mention satan, nortrom and sohal here. With a month of training you will probably not even reach the silver league. Don't want to sound to harsh, but i think it is underestimated how complicated the game is.


  • aSymmeTric likes this
To watch stratego videos: https://www.youtube....HOHXWONQMsVcOLA

#42 GaryLShelton

GaryLShelton

    Marshal

  • Moderators
  • 4,448 posts
  • Coat of arms
  • Gold Miner

Posted 03 December 2014 - 11:50 PM

I really think the only way to know anything is to put it out in the field and see what happens. I wish the new programmers much luck and I urge them on, but wrapping the brain around all the situations of this game in a theoretical way is a super tough proposition. I just can't wait to see what they come up with!

Vincent de Boer is a genius in my book.
  • vegas likes this

The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...604#entry339604

#43 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 04 December 2014 - 09:28 AM

All the progress sounds good so far! I have two think to mention tough:

 

1: Vincent de Boer once wrote a program for a computer stratego player. It was able to beat mid ranked players. Because he has so much trouble with this, i doubt it that the programmers on this site have been able to create a even stronger computer player, which is difficult for all.

 

 

We know and we used that as a basis to start with. That AI was designed to beat the strongest player. We designed it so it can be tuned to casual and hardcore players.

 

 

2:

 

''Do not worry Gary! We've ensured that everyone on the Stratego-team is now an excellent Stratego-player and knows the game inside out.''

 

How long has the team been working on this? I think it is not much longer than a month or so. With all respect, i think you are underestimating the game then. It takes years to become an ''excellent'' stratego player, and there are only a few on this site, i want to mention satan, nortrom and sohal here. With a month of training you will probably not even reach the silver league. Don't want to sound to harsh, but i think it is underestimated how complicated the game is.

 

We have been working on it for longer than a month, I can assure you that. 

We are not underestimating the game at all. I think you're underestimating us ;)

 

It's our job to design games and think for and like the players. We can handle this. It's our job after all!

 

 

I really think the only way to know anything is to put it out in the field and see what happens. I wish the new programmers much luck and I urge them on, but wrapping the brain around all the situations of this game in a theoretical way is a super tough proposition. I just can't wait to see what they come up with!

Vincent de Boer is a genius in my book.

 

Thank you for your support :D


  • maxroelofs likes this
Posted Image

#44 Lonello

Lonello

    General

  • Moderators
  • 2,020 posts
  • Coat of arms
  • Gold Marshal

Posted 04 December 2014 - 09:55 AM

Yes, well I gotta say it does all sound really promising, we would have something here non existing anywhere on the world now :).


Lo

#45 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 04 December 2014 - 11:18 AM

Yes, well I gotta say it does all sound really promising, we would have something here non existing anywhere on the world now :).

 

That's why we're gonna blow your mind and take the world over by storm!


Posted Image

#46 Napoleon 1er

Napoleon 1er

    General

  • Moderators
  • 2,042 posts
  • Coat of arms
  • Platinum Marshal

Posted 04 December 2014 - 12:46 PM

...looks like this singleplayer version might be so much appealing that players may even prefer it to playing against other players!!!...lol... hope this is also taken into consideration...stratego is a 2 players game where behaviour plays also a role ... like a little hesitation of a player before playing may provide info to his opponent ... i see that difficult to implement in a software ... only "bluffing" hesitations could possibly be implemented...

...but we are all keen to see it coming soon...

If you don't know where you go ... you have a lot of chance to arrive elsewhere ...


#47 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 04 December 2014 - 03:47 PM

Well, the singleplayer-experience will not and cannot be the same as the multiplayer-experience. But the singleplayer will offer much more than any stratego game has ever given.

Let's hope people prefer playing this. We're aiming for that!

 

Some people will always prefer to play against real people.

 

For those people, we will take a look at more stratego games after launching the Singleplayer ;)


Posted Image

#48 maxroelofs

maxroelofs

    Major

  • Dutch Tournament Manager
  • 1,042 posts
  • Coat of arms
  • Platinum Marshal

Posted 05 December 2014 - 05:33 PM

 

Let's hope people prefer playing this. We're aiming for that!

 

I really hope they don't! Would be the end of all the beautiful Dutch live tournaments then.  ;)

 

Looking forward to it though.


To watch stratego videos: https://www.youtube....HOHXWONQMsVcOLA

#49 The Prof

The Prof

    Major

  • NASF Committee
  • 1,467 posts
  • Coat of arms
  • Platinum Scout

Posted 05 December 2014 - 08:22 PM

I'd like to suggest an objective way to measure how good this new single-player AI will be.  Each of the three levels of the AI should have an ELO rating that rises and falls with its wins and losses just as human players have.  I am not saying that single player games should count as ranked game for the human player, but rather only for the computer player.  The AI will be playing hundreds of games a day, and so it will quickly reach an ELO rating that reflects its true skill level.  Without this, evidence of how good it is will be purely anecdotal.  There will be people making posts in the Forum saying things like "The AI sucks, I just easily beat the hardest level".   However, if the AI has an ELO rating that reaches the silver ranks on the leader board then no one could argue with the accomplishment of its designers.  Allowing the new bot to have a rating would be a strong sign that its creators really believe in their new product.


  • Lonello, Napoleon 1er, papillon and 1 other like this

#50 Desert Oasis

Desert Oasis

    Miner

  • Members
  • PipPipPipPipPip
  • 160 posts
  • Coat of arms
  • Silver Colonel

Posted 05 December 2014 - 08:54 PM

Good idea Prof

#51 tableplay

tableplay

    Scout

  • Members
  • PipPipPipPip
  • 147 posts
  • Coat of arms
  • Platinum Marshal

Posted 06 December 2014 - 07:49 AM

There's a Fields Medal or Nobel Prize (or whatever the AI equivalent of this is) here if this bot can beat an 800 plus ELO player.

  However, if the AI has an ELO rating that reaches the silver ranks on the leader board

 

We know and we used that as a basis to start with. That AI was designed to beat the strongest player. We designed it so it can be tuned to casual and hardcore players.

 

 


#52 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 08 December 2014 - 09:33 AM

I'd like to suggest an objective way to measure how good this new single-player AI will be.  Each of the three levels of the AI should have an ELO rating that rises and falls with its wins and losses just as human players have.  I am not saying that single player games should count as ranked game for the human player, but rather only for the computer player.  The AI will be playing hundreds of games a day, and so it will quickly reach an ELO rating that reflects its true skill level.  Without this, evidence of how good it is will be purely anecdotal.  There will be people making posts in the Forum saying things like "The AI sucks, I just easily beat the hardest level".   However, if the AI has an ELO rating that reaches the silver ranks on the leader board then no one could argue with the accomplishment of its designers.  Allowing the new bot to have a rating would be a strong sign that its creators really believe in their new product.

 

This is purely based on skepticism.

The upcoming AI will feature 3 difficulties, that will each be attuned to a different player.

Our goal is not to make the most difficult AI that even the best Stratego-player can beat; our goal is to make a great AI that is fun to play against for everyone, ranging from a beginner to an expert.

 

Implementing such a rating-calculator to win over the faith of a few skeptics is not something we think deserves our time.


Posted Image

#53 Lonello

Lonello

    General

  • Moderators
  • 2,020 posts
  • Coat of arms
  • Gold Marshal

Posted 08 December 2014 - 11:13 AM

Implementing such a rating-calculator

 

I think Prof worded it quite difficult but I imagine the 3 AI's will automatically get an Elo, just as everyone has. Not to win over faith, not to take any time as to create a rating-calculator or anything like that, but simply have it the same way as it is now. The AI's will have accurate Elo's fast as Prof points out: the AI's will play a vast amount of games in a short while. So anyone will know their strengths in a heartbeat. No fuss about it?!

 

I think the programmers will be curious to know the strength as well. You can even learn the evolution of the AI's then. Maybe the humans will learn how to deal with the AI's which make their Elo's drop, then the programmers may from time to time adjust the AI's parameters which will surprise the humans and the Elo's will rise again. Fun to watch.


Lo

#54 Desert Oasis

Desert Oasis

    Miner

  • Members
  • PipPipPipPipPip
  • 160 posts
  • Coat of arms
  • Silver Colonel

Posted 08 December 2014 - 06:31 PM

Yes my interest would be out of curiousity as to how the AI matches up to the players. The most honest way is through the ELO rating.
  • Kernel Mustard likes this

#55 The Prof

The Prof

    Major

  • NASF Committee
  • 1,467 posts
  • Coat of arms
  • Platinum Scout

Posted 08 December 2014 - 10:02 PM

This is purely based on skepticism.

The upcoming AI will feature 3 difficulties, that will each be attuned to a different player.

Our goal is not to make the most difficult AI that even the best Stratego-player can beat; our goal is to make a great AI that is fun to play against for everyone, ranging from a beginner to an expert.

 

Implementing such a rating-calculator to win over the faith of a few skeptics is not something we think deserves our time.

 

My motivation was not skepticism.  The paper by Vincent De Boer convinced me that a bot can indeed be programmed to play at a high level.  However, my point was that by not having an objective measure, I believe you will be inviting skepticism.  I completely agree that the goal is to have fun, and the concept of three difficulty levels is a good one.  For a silver player, what will lead to the most fun is a bot that plays as good as as possible, while for others it makes sense to have "easy" and "medium" levels.  My suggestion is a transparent way to indicate the skill level of the three different AI options.  If a new player is looking for a competitive single player match, it would be helpful to know if a bot has a 200 rating or a 400 rating, which seems to be more meaningful that choosing between “easy” or “medium”.  As Lonello said, there is no need to make any new rating calculator.  Just use the same formula that already applies to every human player. 



#56 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 09 December 2014 - 09:24 AM

As I said, our developers are currently occupied with more important matters.


Posted Image

#57 Lonello

Lonello

    General

  • Moderators
  • 2,020 posts
  • Coat of arms
  • Gold Marshal

Posted 09 December 2014 - 10:51 AM

 developers are currently occupied with more important matters

 

I'm not sure what you're saying here Mick. You get an Achievement when you beat the AI don't you? A point, a streak, or whatever. In fact, Stratego.com rules in that. You get all kind of Achievements which are all fun to reach. That is what makes you come back. Commercially the best idea the site had when started 2 years ago. Giving away free points, doesn't cost a thing, but it makes people come back. Hell, you even did the same here on the Forum, I'm now a Miner I see ;)

 

The Elo-template is there for the taking too, like Prof said. Ofcourse you can also give the AI a point when it wins, or a strike, a bomb, or whatever. Resulting into a Leaderboard. Players come to see where they stand, and look where the AI's are. Find themselves really good when they for example see they are above AI3 and go to beat the AI2.

 

So that's all I'm saying... ranks are involved in the Single game aren't they?


Lo

#58 Desert Oasis

Desert Oasis

    Miner

  • Members
  • PipPipPipPipPip
  • 160 posts
  • Coat of arms
  • Silver Colonel

Posted 09 December 2014 - 07:51 PM

It is about control. The admins have an agenda. They have things they want to do with the site. They don't want to be told what to do. Implementing a forum post ranking system took time, which no one asked for, but a ranking sytem for the AI is not their idea. Hence the lack of interest. I imagine Mick and his crew are even less open to ideas considering the amount of criticism they received when they announced a new single player format and ignored the demand for an improved multiplayer.
  • Luckypapa likes this

#59 GaryLShelton

GaryLShelton

    Marshal

  • Moderators
  • 4,448 posts
  • Coat of arms
  • Gold Miner

Posted 09 December 2014 - 09:58 PM

As I said, our developers are currently occupied with more important matters.

 

Mick, you've said your focus is on the Singleplayer game.  If that's true then The Prof's idea of giving the AI an ELO rating is a solid contribution to the discussion about Singleplayer.   What more important matters could there be?  It would absolutely be open and transparent to have everyone be able to see that the AI is rated 200, 300, 400, or whatever.   It would be playing hundreds of games a week, as The Prof said, and its progress would be fun to follow, like Lo said. :)   DeKaeneas-Spy could even make a betting event on when the AI beat its first silver player, for example.  :D  

 

Also, names and ELO ranking are not mutually exclusive propositions for the AI.  Each AI would need a name, just like any other player, and "Easy, Medium, and Hard" all would still work fine as descriptive monikers.  It's just that each AI would simultaneously have its own ELO ranking for all to see.

 

Moreover, I really think this would be a popular thing with new Stratego players to the site.  Having an ELO ranking on the AI would help them get up to speed on how good they are playing before venturing over to the ranked player side, or anytime they wanted.

 

If it is not to happen this way sooner, it would be a fine idea for later.



The complete GS&F Rules can be found here: http://forum.strateg...rum-rules-2016/

Draw Refusal Rules, specifically, can be read here: http://forum.strateg...604#entry339604

#60 MickM

MickM

    sr. Game Artist

  • Administrators
  • 337 posts
  • Coat of arms
  • Bronze Spy

Posted 10 December 2014 - 11:50 AM

Again, implementing an ELO-system costs time which we'd rather spend on other things.

You have no idea how much time this costs :)


Posted Image




1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users