Thursday, April 23, 2015

The Power Grid problem.

Since I first played it maybe 7 or 8 years ago, Power Grid has been one of my favorite games.  I have all the expansion maps, the alternate power plant deck, and even the new deluxe version. We played just this past Tuesday at our weekly game night - this time using the India map, the first time I've played on that map.  It was a nice tight 4 player game...however the fashion in which it ended is something I've seen on multiple occasions, and it leads me to wonder if Power Grid is ultimately a flawed game. 

Here was the situation - my friend Brian and I were each at a power plan capacity of 14 cities.  He had built up to 14, I had previously built up to 15.  The other 2 players were behind us at 12ish cities.  We had recently turned to Step 3, so all 6 plants in the market were available.  The other 2 players had already purchased plants, so it was down to Brian and I, and since he was one city behind me, I had to nominate first.  The 36 plant came up (3 coal to power 7 houses).  This would push both Brian and to 18 capacity.  We basically knew that whoever got the plant stood a good chance of winning?  We bid it up to 73 euros, at which point Brian relented, and I won the plant.  We promptly flipped up the 38 plant, which takes 3 garbage to power 7 cities.  Brian, of course, was able to buy it for face value, and have enough cash to actually build up to 18 cities, while I only had enough cash to build to 17.

I have seen this on several occasions, where there's a tight bidding war...and the "loser" really ends up winning by getting a lucky draw on the replacement plant.  Now, perhaps in our situation, "better" players would have remembered all the plants that had been previously placed under the Step 3 card, and knew there was a decent chance to flip up another 7 plant.  (We looked at the remaining cards...there was less than a 50% chance - counting the plant Brian got, there was 2 "7" plants out of the remaining 6 or 7 cards (I don't remember the exact count).  So, he did get a bit lucky.  It should be noted, that most of the remaining plants were "6" values, so he likely could have won anyway, since that would have put him at a capacity of 17 vice my 18, but I couldn't build the 18th he would likely have won on tie-breaker with more cash.

Of course, I have seen the opposite, where someone gives up on a bid, and then a terrible plant is flipped, and the loser is set back to the point where he can't get back into the game.  In fact, when I really think about it, I think someone gets screwed one or the other by the plant flip in every game.  Perhaps with the exception of the China map, where they come out in numerical order.

Folks who have the entire deck memorized have and advantage, I think (especially on the China map).  I don't like that, primarily because I'll likely never have the deck memorized, but perhaps that sour grapes on my part.  So, I this a flaw in the game, or should you account for this in your strategy somehow?


  1. I gave up on Power Grid many years ago, after about 4-5 games.

    You raise a good point

    One solution is to have the ratio of 'cards in the deck' to 'cards drawn during the game' to be very high, such that memorizing the deck becomes meaningless. Note in some games that ratio is 'cards in the deck for round n' to 'cards drawn during round n'.

    Most gamers don't seem to like this approach, I think because they like the idea of only a certain number of possible cards, within the possibility of memorizing.

    As always, what is the game trying to model?

    Using card draw as a pseudo-random event is a poor choice.
    If a player rolls a 1d6: 3,1,2,5,4, that does not mean the sixth roll will be a 6 because it's 'still in the deck'. It's still a 1 in 6 chance.
    But if a player draws cards two, one, four, five, and three out of a six card deck, you can bet the bank the next card is six. To make card draw pseudo-random, a few hundred cards would help.

    Possible solutions include: 1) roll dice, or spin a spinner for each draw, and reference a pre-existing table for the factory. This means the same factory may come up many times, and some factories will never come up. Players will need to have some kind of duplicate cards to put in front of them.
    2) Greatly increase (x4? x8? ) the number of available cards in the deck.
    3) Put all discarded factories back into the deck and shuffle before every draw.
    4) Draw twice (3x?) as many cards so there is a plethora of available factories to choose from.

    Re: Strategy: A "good" game -- IMHO -- is one where the player's strategy is mighty similar to the strategy of the person he represents, and makes similar decisions, in whatever the game models. So in this case, the player's strategy should be similar to the strategy of the CEO of a power company. Does the CEO of a power company memorize the deck of future available factories?

    1. Well, I disagree that every game needs to "simulate" all the decisions one would make in "real life", but I get your point.

      Having multiple of the same factory is an intriguing thought, as that could lead to more direct competition in the resource market.

  2. Occurs to me some of my favorite games are card, and handle this in a different manner.

    All of these are 2-player, and each player shares the same deck.

    Up Front is the classic.

    Although only a 162 card deck, it's nearly pointless to memorize cards, with half the cards going to your opponent, and 1/3 (?) the cards being used as RNCs or RPCs to resolve combat and other mechanics of the game. As the cards are multi-use, there is no reason to expect a particular card will ever be played as its "face value".

    I suppose a munchkin could memorize the cards, and card-count as they come up, and then at least play the percentages. But as players discard -- face down -- far more cards than they play, that defeats the munchkin option.

    Maybe 1 in 15 games one player has such bad card draws, he stands zero chance of winning. No problem, the game has such great replay value, just play again. Routinely maybe 1 of 5 games goes down to the last hand, and a few times a game has gone down to literally the last card.

    As an aside, if your opponent is winning and has fire cards, he can just blast away at you to eat up the deck. [The deck is the measure of time (typical game is 3 times through the deck).] Some might call this "gaming the system" or "poor sportsmanship". To the contrary, this wonderfully models how difficult it is to accomplish even the simplest thing, when somebody keeps shooting at you.

    Hannibal: RvC is another example.

    Depending on which side you are and where in the game, maybe 20% of the cards can be used four ways, maybe 50% can be used three ways, maybe 30% can be used 'only' two ways. All cards can be used at least two ways.

    Routinely, entire game go by with a set of cards never being played as events, thus making for a different "history" each time.

    OTOH, Battle Cry and Memoir '44 (the smaller version), nearly all cards will be used at face value, yet at least each player draws from the same deck. And yes, there is some card memorization in those games.