Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Qvist]

What can you say about translations? When will they be released? What languages are currently planned? Will the interface be translated too or only the cards itself?
Ted: We have the cards translated in German.  However, we need to translate the buttons and UI elements.  This will probably happen once we get all the cards released so we can do it in one effort.

If you need help for that, just let me know. 

If you're offering, I can send you the button and status line text that I need translated!

Yes, go ahead. I'm glad to help you with that.

[Ted Griggs]

What can you say about translations? When will they be released? What languages are currently planned? Will the interface be translated too or only the cards itself?
Ted: We have the cards translated in German.  However, we need to translate the buttons and UI elements.  This will probably happen once we get all the cards released so we can do it in one effort.

If you need help for that, just let me know. 

If you're offering, I can send you the button and status line text that I need translated!

Yes, go ahead. I'm glad to help you with that.

Awesome.  I just need to pull in the Alchemy data and I'll send it out to you!

[Tables]

Wait wait when did this turn into the Goko volunteer recruitment thread ?

[Watno]

How do you plan to display the game on smaller phone screens?

[Ted Griggs]

How do you plan to display the game on smaller phone screens?

Sorry, I thought I had answered this earlier, but it didn't post.

In any case, I think this is important and we'll do something after the first version for iOS and Android comes out.  The folks here seem to like playing with the current UI, but I think can be improved!  We did add things like the ability to buy a card while zoomed, increasing the touch area around the +, etc.

Suggestions (other than text mode!) are welcome!

[trisha_brooke]

Thank you, everyone!

Ted and John really enjoyed answering questions. If you have any more questions, you can send them to help@goko.com. Or let me know, and I can forward them on. :-) As always, you are awesome.

And as before: Thank you, Great Expectations for gathering all the questions and putting them in an easy to read format, and thank you Theory for your help on pinning the message and your input.

Thanks!

[WanderingWinder]

Or even better, the formula itself?

Here's a very basic formula I asked the engineer who programed the stats:

rating = μ - (2 * σ)
Where μ is the mean of your estimated skill level and σ is one standard deviation from that mean.

Dominion's initial values for μ and σ are:
μ = 5500
σ = 2250

Great, TrueSkill.

That does not mean TrueSkill. I mean, without any method of how you come up with those things, it could be basically anything. Actually 'Elo-like' is more information than this, and more or less satisfying - I wouldn't expect them to give us the exact algorithm, though of course it would be nice.

[WanderingWinder]

Can you provide us some assurance that you are staying ahead of new security concerns? For instance, do you have a third party doing periodic checks on things?
JQS: We believe we’re up to speed and respond to anything that looks like a security issue quickly. So far we’re doing well. As mentioned in the thread, we don’t touch anything to do with credit cards or payment systems (most web sites don’t). We will have periodic reviews but that doesn’t really mean nothing will happen in the future... even large entities like Yahoo/Google/Microsoft despite their large dedicated security teams still have to address problems when they arise.

This is really the only issue for me, well not the only issue but the rest is gravy, and of course I understand that you can't guarantee foolproof security. Actually it's heartening to hear you say that, because you can never ever do that. Fine, good. But I'm still feeling like your only assurance is 'trust us', and well, that's just not all that reassuring. Those other companies, for one thing are much larger, and there comes some reassurance from that, but more than that, they have lots of stuff out there; you guys have... 'trust us', and 'we think we're doing well'. The problem is, everyone thinks they're doing well. You thought you were doing well before, and you had an enormous problem.

tl;dr if you could give some description of what you mean by "We will have periodic reviews", that would most likely help a lot.

[Rabid]

From the previous rating information below I think most people have assuming a Trueskill like rating system.

http://forum.dominionstrategy.com/index.php?topic=4321.msg95643#msg95643
How are ratings calculated? - Each player gets a rating between zero and ten thousand. All registered players are rated the same way, including the bots. The system for calculating ratings is pretty complicated and dependent on quite a few factors, so I'm going to kind of go through some of those factors. First the ratings are just based on what place you came in rather than how many victory points you won by. If the game is resigned it's viewed as a loss. It doesn't matter if you came in at a hundred points or two points, its all rated the same. Ratings are also adjusted based on the ratings of your opponents, so beating a higher rated opponent will help ratings much more than beating a lower rated opponent. The system is conservative in it's calculations, so it is reasonably sure you deserve the rating work before it awards it. In other words, your rating may be low when you haven't played many games and rise as the system gains more confidence that you deserve the higher rating. Your rating will fluctuate more when the system has little data about how skillful you are. So the first couple games you play your ratings are going to be all over the place. As you play more games your rating is only going to slowly move up and down. Unexpected results, such as a win over a higher rated opponent, affects the rating most, so play someone that's much better than you and win and your rating might go pretty high. Your rating will fluctuate more with results against players with more established ratings. So pretty much the more you play, the more established your rating is going to get, and playing against the other players with established ratings is going to be the best way to improve your score.


http://forum.dominionstrategy.com/index.php?topic=4895.msg115363#msg115363
+4 - How does the rating system actually work? Numbers and formulas please.
Our rating system function very similarly to Isotropic’s. However, players in Goko’s Dominion start at 1000 points instead of 0. As you play more, your actual rating becomes more clear to our formula. So your rating will start to fluctuate less. Your rating, a number from 0 to 10,000, is shown in the Rating column of the leaderboard.  A number in parentheses indicates the number of points rating points you gained or lost in your last game.

A "P" indicates that your rating is "provisional", i.e. the system does not yet have a good estimate of your ability.  While you are provisional, you'll see your rating in the leaderboard, but nobody else will. We do this because the initial standard deviation is so high.

Ratings are adjusted based on the following:
* What place you come in and how many players are in the game.
* The ratings of your opponents (The higher the level of your opponent the more you’ll move up)

Ratings are not adjusted based on the following:
* How many points you win by
* Number of turns

[Polk5440]

Thanks for the time you took to answer all of the questions. It is much appreciated.

*What is your attitude towards Dominion players/community helping you to develop the site?
In the future, we’d like to publish things like how to create your own AI. Help with ideas to draw new players in. We love that the community is creating adventures and that kind of input is exactly what we’re looking for.

Wow! Program your own AI? That's cool. Would there be an ability to run simulations (AIs against each other, essentially), too?

Will Dominion/Goko support pure webkit based browsers like that on the Wii U?
Ted: HTML5 has come a long way.  The issue isn’t compatibility of the browser but things like hardware acceleration and other things that modern games need. The game is playable on Wii U, but sluggish.

This makes me sad. Goko Dominion is too much for the Wii U to handle? What is going on under the hood that requires all this high tech stuff?

For auto-match, is it foreseeable to use the most expansions between the people auto-matched to prevent people from gaming certain setups they are good at?
Ted: In professional setting it uses the cards of the table owner, but doesn’t allow them to pick the cards they want to play.

Too bad one cannot use the union of the sets owned.

The issue is that when we release new versions, which we do every few days, the sessions are invalidated by virtue of the servers restarting.

I just played a game, and I noticed it runs much better than the last time I tried. The updates must be doing something on the performance end, too. Never mind. A game just failed to load. Maybe next month. I figured out the problem -- I can't navigate to the game client through history in Firefox because then the first game I play doesn't load. Weird, but easily avoided, I guess. 

[hsiale]

Just let us know how you'd like tournaments run.  I can see a number of ways, but if there is a best/favorite way, then we should go with that first.

I see two tournament formats that would be good to have:

1. Big and long tournament. Happening probably once a year (Goko Dominion Championships?). Single or double elimination bracket, seeded according to Goko pro rating. One round per week, players have to agree on when to play, first to 4 wins wins the round and advances. First player: random or higher seed first game, loser of latest game in consecutive games, random after a draw.

2. Evening tournament. Held on times when there are many people online, if interest is sufficient even daily. People sign up and at a set hour they are paired for first round. Swiss, 4-5 rounds depending on number of players, during the round each table plays the same cards (to make the games last similar amount of time). First player decided so that everyone starts first more or less the same amount of time. Some of those tournaments could be 3- or 4-player games.

[Toskk]

Here's a very basic formula I asked the engineer who programed the stats:

rating = μ - (2 * σ)
Where μ is the mean of your estimated skill level and σ is one standard deviation from that mean.

Dominion's initial values for μ and σ are:
μ = 5500
σ = 2250

That does not mean TrueSkill. I mean, without any method of how you come up with those things, it could be basically anything. Actually 'Elo-like' is more information than this, and more or less satisfying - I wouldn't expect them to give us the exact algorithm, though of course it would be nice.

I agree, knowing they're using an Elo derivative is helpful, although there's still a fair amount of incomplete information even then (for example what method they used to modify Elo for multi-player matches). The item I'm especially interested in, however, is how they dealt with the problem of luck in Dominion matches. For example, the Chess Elo system is partly based around the statistical observation that "White has a 100% expected score with a 390-point (or more) rating advantage, and a 0% expected score with a 460-point (or more) rating disadvantage."

http://www.chessbase.com/newsdetail.asp?newsid=562

i.e. at a 390-point rating advantage, the player will have an effective 100% win rate.. something that is absolutely untrue in Dominion, even with a 10,000-point rating advantage. I really hope their engineer has run a heck of a lot of statistics on matches, to determine a more accurate win/loss ratio-per-rating-point-difference.

[WanderingWinder]

Here's a very basic formula I asked the engineer who programed the stats:

rating = μ - (2 * σ)
Where μ is the mean of your estimated skill level and σ is one standard deviation from that mean.

Dominion's initial values for μ and σ are:
μ = 5500
σ = 2250

That does not mean TrueSkill. I mean, without any method of how you come up with those things, it could be basically anything. Actually 'Elo-like' is more information than this, and more or less satisfying - I wouldn't expect them to give us the exact algorithm, though of course it would be nice.

I agree, knowing they're using an Elo derivative is helpful, although there's still a fair amount of incomplete information even then (for example what method they used to modify Elo for multi-player matches). The item I'm especially interested in, however, is how they dealt with the problem of luck in Dominion matches. For example, the Chess Elo system is partly based around the statistical observation that "White has a 100% expected score with a 390-point (or more) rating advantage, and a 0% expected score with a 460-point (or more) rating disadvantage."

http://www.chessbase.com/newsdetail.asp?newsid=562

i.e. at a 390-point rating advantage, the player will have an effective 100% win rate.. something that is absolutely untrue in Dominion, even with a 10,000-point rating advantage. I really hope their engineer has run a heck of a lot of statistics on matches, to determine a more accurate win/loss ratio-per-rating-point-difference.

That's actually not true - if you look at the actual formulas, it only asymptotically approaches 100% - the number Sonas uses in that article is only where it goes over a certain threshold (iirc, 95%). More important, FIDE actually caps the expected percentage after a certain amount, not letting it go over something like 92%. Anyhow, if your point is correct, it would just mean a tighter bunching of players; what 1 point is worth is totally arbitrary. But there are considerations to take into account here - simply Elo-like is quite vague, I agree.