Power Grid design review

Although I am a fan of Power Grid, I have been fairly critical of the design of the power plant market mechanic. It is too fiddley; there are too many things to do and too many special cases. It is not elegant.

• Specific initial plants 3-10 (13), but shuffled deck.
• Special rule if no power plant is purchased.
• Special rule if power plant numbers are less than or equal to leading player's cities.
• Special action at end of phase 5 to bury highest power plant.
• Special action at step 1/2 transition.
• Special but different action at step 2/3 transition depending on phase (2/4/5).
• Change in power plant layout during step 3.
• Special action in phase 5 during step 3.

In my personal attempt to come to terms with all of this, I am going to analyze the intent of the mechanic and how all the above craziness came to be. This is purely speculation; I do not know Friedmann Friese, and have not interviewed him.

Let's start with the basic premise: players need to buy power plants to power their cities. Since the number of cities will grow as the game progresses, players will need plants capable of powering more cities later in the game. So a good starting place is to put all the plants in numerical order:

3 4 5 6 7 ... 40 42 44 46 50

Well, this works, but it's not very interesting. The plant order is always the same, and it's predictable. The game could be completely analyzed leading to specific opening plays with expected value for each plant based on the number of players, etc.

So we want some randomness. Let's shuffle the deck of plants.

31 11 44 26 19 ... 4 40 28 7 33

Hmmm. This causes some problems. Some starting plants cannot be afforded early in the game (purchase price + cost of fuel), so players will get nothing. And near the end of the game there will still be many low-numbered plants that no one wants any more. Completely random is worse than completely sorted.

How about sorted with a small amount of localized randomness? Sort the deck in numberical order, then change each plants position by a small but random amount up or down in the deck.

5 6 3 11 4 9 14 ... 37 36 44 50 42 35 46

This works very nicely, but the effort required to do it is prohibitive. If Power Grid was only a computer game, this might be the way to go as it would be exceedingly simple to "shuffle" in this way. So how do we get a result similar to this without all the work?

Let's start with Power Grid's initial setup: remove the first 8 plants (and 13), and shuffle the rest of the deck.

3 4 5 6 7 8 9 10 (13) ...

The power plant market is a moving "window" on the shuffled deck of cards. As plants are bought, a new card is added, moving this window 1 place to the right.

3 4 5 6 7 8 9 10 13 27 42 30 15 39 21 ...
......|_________________|

The fixed initial power plants guarantee that there is something affordable and appropriate for all players. However, since the rest of the deck is random, unwanted plants (too low and/or too high) will ocassionally fall into the window. What are the issues at this point?

Players will each be positions to want different sizes and types of plants over the course of play. One player may want a plant in the 20's and another in the 30's. So in this sense, having a nice random selection of plants is a good thing.

... 27 42 11 30 15 39 21 46 ...
....|_____________________|

There will be plants that no players want any more. These will be the plants that have too small an output. These are essentially useless in the game at this point. The rules allow for this by removing these plants from play.

... 27 42 (11) 30 15 39 21 46 38 ...
....|_______________________|

(goodbye 11, hello 38)

There will also be plants that no players want YET. These will be the plants that are too expensive right now, but may be desired in a later turn. The rules allow for this by placing these plants under the draw pile until step 3.

... 27 42 30 15 39 21 (46) 38 23 ...
....|_______________________|

(goodbye 46, hello 23)

All of the above serves two purposes: it de-randomizes the power plant order enough so that appropriate plants are available when needed, and it moves the "window" along at approximately the same pace as the players' actions. Since this changes from game to game (and with number of players), the progression of the window must adjust itself accordingly.

So it really works?

Yes.

Then why does it bug me so much?

There's eight special rules to manage this window, and it doesn't scale.

Doesn't scale?

If you used the same mechanism for a longer game or with 8 players, it would fall apart. A good analogy would be to look at a graph of the function y = x^2. It's a curve, but when you look at certain local areas up close, you might mistake it for a line. This is what Power Grid does. The power plant mechanics work just good enough for 2-6 and for the number of turns that typical games take that it gives the illusion of "being a line". But I can see the hidden curve on either end.

Can I think of something better?

Maybe. Some day. If I do I will present it here as a possible variant.