Print

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

[Dubdubdubdub]

Just started playing the adventures. The first couple of levels are of course very simple - there's not really any way to lose. But getting a three star rating is a problem
True, I'm not playing optimally, jerking around with elaborate baseset engines instead of something simpler. But I consistently beat the bots with like 44 to 5 or something, still getting 1 star. So what is this rating based on? Mainly on number of turns? *only* on number of turns?

[Captain_Frisk]

It's covered elsewhere, but at least originally, you need to get a certain total # of points - and its well over 40.  So if you really want the stars, you need to make sure you drain the duchies too, not just win convincingly and quickly.

[ftl]

They have changed around the number of points required, it's almost impossible to get 3 stars now...

[Captain_Frisk]

They have changed around the number of points required, it's almost impossible to get 3 stars now...

Maybe you should buy m0hr zaps?

[Captain_Frisk]

Better as meme

[Dubdubdubdub]

Right, so the exact opposite: be more elaborate and drain the vp before finishing. Got it. Not doing that.

[jsh357]

Yeah, this is one of my chief complaints about Adventure mode.  Takes all the finesse out of winning if you have to point gouge.

[Ozle]

Have you tried Up, Up, Down, Down, left, right, left, right, B, A, Start?

[Donald X.]

I recommend suggesting what the actual formula for stars should be. Who knows, they might change it.

[Captain_Frisk]

Easy (ish)

Fewer turns + Less Zaps == More stars

They'll have to tune the curves on a board by board basis... a sea hag game is going to take foverer, while wharf + fishing village will not.

Obviously, if you use 0 zaps, your games are going to be longer.  If you use the full compliment of 10-20 zaps, then the game should be much faster.

Score shouldn't factor into it because it actively discourages smart dominion play (ENDING THE GAME WHEN YOU CAN WIN)

[Captain_Frisk]

But - I don't expect this to be used - because it discourages zapping.

[Ozle]

Less Turns = More Stars.

Not quite sure on the exactness, would have to be tailored by level, but shouldnt be too hard.

Get the Goko team to play it a few times (with X zaps each, lets say 3), lowest number of turns sets the three stars level.

This has the double impact of them finding out how hard/easy it is and setting the time.

it also encourages using zaps to get a quicker turn time, thus making the $$$ as well.

[Ozle]

(Doesn't of course have to be the Goko team, I would get some of the top players in, give them 3 zaps and best score becomes the default for that level.)

That means the average schmuck like me will need to play really well, get luck, or...buy stars to get those 3 stars.

You could even reward the players by naming the bot opponent after them.

Everybody is happy!
(well probably not everybody...)

[Captain_Frisk]

Here's a gut feel for what i'd expect it to look like for a board with smithy - and no significant starting penalties



Note that sometimes when the opponent is penalized, it actually takes longer.  The games where the enemy starts with 5 coppers and 5 curses?  They take forever because the opponent can't buy provinces and you need to buy them all.

[Donald X.]

But - I don't expect this to be used - because it discourages zapping.

I don't imagine they will make the system flat-out discourage zap use, although I wouldn't rule it out, they might if it sounded good enough. But I think they have come around to my argument that they are not going to make money from zaps and should stop trying in favor of making the campaigns fun; the dominion coins product will be campaigns themselves rather than zaps.

At the moment it seems like they will use my campaign levels, where zaps can be a crutch but shouldn't be needed (except the free zaps, you need those in acts 2-3), except for level 20s, which they changed to require more zaps, I guess thinking, wait this is where you get a pile of VP tokens. And from my perspective that makes no money but makes level 20 bad, because I mean, okay, I can buy some zaps with dominion coins which were free, to joylessly beat the level 20's. But uh that's how it looks. Nineteen playable levels per campaign, well it's a big improvement.

Again I don't know when this stuff will go up, because one thing was, they plan on adding more things you can muck with in campaign levels, based on my suggestions, and so I used a little of that stuff in my levels, and so they won't go up until that stuff is programmed.

Anyway uh it's possible they are still clinging to the idea that someone will buy up zaps to get three stars. I haven't discussed the stars thing with them. Number of turns sounds like a reasonable measure if what you want is something very simple, which possibly you do, but it just has different bad cases than score. You cite the case where I end the game on piles and win before we've gotten into the Provinces, but what about say a Duke strategy? I lock up the game by getting eight Duchies before the bot is anywhere near to cleaning out the Provinces, but I don't get three stars because this strategy is too slow.

And the game is about scoring more when it's over - it's not about ending the game quickly. It's more natural to use score than turns then, right? And for some games score does tell me how well I did. OTOH, a lot of games will end with close scores, and the point to stars is to say how well you did vs. other instances of this game being played, and well can score do that?

I guess it could be, well your graph with score instead of zaps. Reward either having a high score or short game. And try to factor in Goons / Witch / etc. into the defaults.

[Captain_Frisk]

Anyway uh it's possible they are still clinging to the idea that someone will buy up zaps to get three stars.

Dear Goko,

The impression that this is your goal is one of the reasons why I have paid you as much as I have.  My contribution to your coffers is in line with my appreciation for this business model.

Regards,

Captain Frisk

[Schneau]

They should do something statistical like baseball's "wins against replacement player", where, after every turn of the game, it calculates the bot's chances of winning if both players' current decks were played by a mediocre replacement algorithm. So, basically, if you've gotten your deck to a place where a decent automated player could win a high percentage of the time vs. the same automated player playing your opponent's deck, then you've done a good job. You could implement this by simulating a large number of games starting from that turn, and see what percent of the time your deck wins. This would be similar to performing simulations with Geronimoo's simulator, except it would be starting for a different game state, namely the one resulting after each turn. If the average chance of winning for the opponent over the full game is below a certain level (say, 20%), you get 3 stars. If it is below 45%, you get 2 stars. If you win at all, you get one star.

Of course, this is likely unrealistic, since it presumes they care enough about this kind of thing to implement a large amount of simulation when they could just say you have to win by more than X points or under Y turns. But, I think it more realistically approximates what Donald X. and Captain_Frisk are trying to come up with, which is a measurement of how well you have played the current challenge.

[sandstorm]

Anyway uh it's possible they are still clinging to the idea that someone will buy up zaps to get three stars.

Dear Goko,

The impression that this is your goal is one of the reasons why I have paid you as much as I have.  My contribution to your coffers is in line with my appreciation for this business model.

Regards,

Captain Frisk

I can see where you are coming from with this statement.  But the thing is Goko can't possible be hoping to make money off of zaps.  10 zaps costs 50 dominion coins.  You get dominion coins for doing anything pretty much (you get 10 coins for every match you win and 2 for a loss whether playing against bots or other humans).  Are there really people out there who want to get 3 stars so badly that they aren't willing to play a few multiplayer matches to afford the zaps?

The other problem is that the campaign mode is such a drag after awhile.  Every level is essentially the same.  I can't see many people willing to plod through every level over and over again just to get 3 stars.  But again even when they improve the campaign I don't see anyone spending real money to buy zaps when you can acquire them for free.  If that is seriously there business model then they are doomed.

[blueblimp]

If there needs to be one rule for stars across all levels, "fewest turns" seems best. It means it's always good to end the game with a win when possible, which reduces boredom (plus more resembles actual play).

If rules can vary, then maybe Duke/Goons/etc. levels can have stars depend on some combination of turns and points. But for typical boards, getting a win in few turns seems plenty sufficient to me.

Edit: Basically where I'm coming from is: the one time I played adventure mode, I gave up after two levels because it was boring. "Fewest turns" seems like the least boring metric to me, so that's why I like it. (To clarify, at that time it was when the second kingdom only changes up like 2 cards, and the new ones you get are Spy and Moat, so you end up playing the identical strategy to the first kingdom, AND need to clear all 8 provinces because the bot is stupid and nerfed.)

[Lashof]

One thing I'd really like to see is for the levels (maybe at the end) to tell you what you need to do for 3 stars.

Maybe give bonus points for quick ending, then have (point differential + bonus) must reach a certain number?  Then, when someone gets 1 star, you can say "Try again and get at least X points for 2 stars!"

[GendoIkari]

I recommend suggesting what the actual formula for stars should be. Who knows, they might change it.

I'm quite sure that this suggestion won't be followed, but my opinion is... completely get rid of the notion of different amount of stars; just make it "win" or "lose." There's a very good reason that both Isotropic and Goko leaderboards don't account for anything like number of points or length of game. Because the game of Dominion itself doesn't make any sense to count those things as any sort of measure of skill.

I love games that allow multiple stars for each level in general. I'm the kind of guy who will not be satisfied with beating all the levels of Angry Birds; but will want 3 stars on each level. Those are fun, and a great way to increase playability in such games. So I'm not against this concept in general.

But when it comes to Dominion, it flat-out doesn't work. If I beat someone by 1 point in 26 turns, that doesn't mean that I played worse than if I beat someone by 72 points in 16 turns. The stated object of Dominion is to have more points than your opponent when the game ends. That's it. The object is not to score the most points possible; it is not to end the game as quickly as possible. There is just as much skill in finding a way to end a game while 1 point ahead as there is in finding a strategy that allows a megaturn where you buy all 8 Provinces at once. It makes no sense at all to reward one particular strategy over another, when within the object of the game of Dominion, each strategy may be equally good. I do not believe there is any way to assign stars to games that doesn't completely change the stated objective of a game of Dominion.

[jsh357]

I don't see why the star ratings can't just be based on number of turns.  In general, the faster you end a game of Dominion on a win, the better you played it.  I guess you could add score in to that mix, but doing stuff like winning by 6 Provinces seems unnecessary.  I do think scoring should be based on the difference in VP rather than the total VP collected if anything.

[Ozle]

I would think a quick poll of the people here would say they would rather finish in 13 turns and win by 1 point, than finish on turn 20 by 16 points.

[ftl]

I don't see why the star ratings can't just be based on number of turns.  In general, the faster you end a game of Dominion on a win, the better you played it.

Because it depends as much on what your opponent does as on what you do. If you he mirrors you, going for the same VP cards, then the game will take roughly half as long as when you go for different VP card piles. Regardless of how well you play.

It'll reward mirroring your opponent's strategy for a faster end EVEN IF that strategy is suboptimal and you have a higher chance of winning by not mirroring it.

[Ozle]

I don't see why the star ratings can't just be based on number of turns.  In general, the faster you end a game of Dominion on a win, the better you played it.

Because it depends as much on what your opponent does as on what you do. If you he mirrors you, going for the same VP cards, then the game will take roughly half as long as when you go for different VP card piles. Regardless of how well you play.

It'll reward mirroring your opponent's strategy for a faster end EVEN IF that strategy is suboptimal and you have a higher chance of winning by not mirroring it.

Except you are playing bots, so they know roughly in advance what the Bot is going to do and can adjust it accordingly.
Its part of the puzzle. Whats the Bot going for, how can I beat that strat in the fastest number of turns.