Sunday, February 24, 2013

2013 MIT Mystery Hunt: "Mashup"

The final puzzle of the Hunt that we put any effort into was Mashup, which featured a laundry list of portmanteau pictures. After digging up music for a while, Steph and I decided to try our hands at something a little more visual.

What we did

It was pretty obvious that we needed to generate two words or phrases for each picture that "mashed up" or overlapped by a few letters to generate what was going on in the picture. So, the picture that showed an Ethernet cord being poured out of a bottle of Clorox was "cableach" after "cable" and "bleach". We ran through most of the pictures pretty successfully but couldn't quite figure out "decree" for "decreepers," "oblong" for "oblongboat," or either element of "rearguardrail".

Then, like so many of our previous attempts, with a good three quarters of the clues solved, the extraction method eluded us to the end. I tried any number of arbitrary extractions: alphabetizing, sorting by total number of letters, sorting by number of overlapped letters, looking for overlaps between the different clues, and nothing worked.

What we should have done

Looking for overlaps between different clues turned out to be mostly right, except that I was looking for "near overlaps" instead of exact overlaps. Therefore, "bloateddyursa" and "uralphnader" were related by "ura" except that there was an extra "s" in the overlap region. But sometimes, there were two overlaps in a given portmanteau, and other times there were none. Still with me?

Then, I was apparently intended to list all the "extra" letters to get the string "fastballscratcher". To quote the solution directly to prove I am not making this up:

 Fastball = pitch and scratcher = itcher, so the answer is PITCHER.

Sure it is.

Would we ever have figured it out?

I'm pretty sure there's no way. Getting the more obscure clues from the pictures like "oblong" and "rear guard" would have been tough enough, but having to see the "near overlap" pattern to piece together the actual solution seemed nearly impossible. And, even if that had somehow occurred to us, I'm incredibly skeptical that "fastballscratcher" ever would have led to "pitcher".

What can this puzzle teach about good design?

Puzzles need to be as internally consistent as possible. In an environment where the minutest of details--a few digits here, an alphabetical string there--can be a clue to solving the puzzle, the number of "extraneous" patterns needs to be kept to a minimum. Two minor breaks in pattern made this puzzle much more difficult to solve, perhaps entirely accidentally.

The first and less egregious was the varying number of letters in each overlap. We figured out this might be the case, but it took a few correct identifications--we got "metropical" and "cableach" early on, and that temporarily convinced us that each overlap was exactly three letters. "Pessimisty" was enough to convince us that the overlap could vary, but we did doubt our answer for a minute or two because of a pattern we thought we had identified.

Second, and worse, is the seemingly arbitrary nature of the "near-overlaps". Not only is that a pattern I never would have thought to look for, but the position of the "extra" letter varied among the second, third, and fourth letters of the overlap. And while most of the clues contained exactly one extra letter, some of the clues had two and some had zero.

Again, it's unclear whether these extra bits of difficulty were intentional--or if a more experienced puzzler would have been slowed down at all by them. Ultimately, though, the point of the puzzle, and its most interesting part, is deciphering the relationship among the pictures and mashup phrases, not out-thinking yourself in plotting patterns within patterns.

Friday, February 22, 2013

2013 MIT Mystery Hunt: "Digging Up Music"

On the Sunday of the Hunt, amidst a tragedy of a Falcons football game, Steph and I stumbled across Digging Up Music, and we knew we needed to try our hands at it. We're both music enthusiasts who consider ourselves at least modestly well-informed about symphonic music, and unlike the last puzzle we tried, we had at least an inkling of an idea of how to start this one.

What we did

Digging Up Music showed a series of thirty-six (!) music excerpts, each a few measures' worth of a reasonably well-known piece from a certain composer. Then, we got a series of one-sentence descriptions of composers, whose names ended up blessedly in alphabetical order. By the time Steph and I started working on this puzzle, our teammates back in Berkeley had made some substantial progress, having identified about three quarters of the composers and a handful of the excerpts. 

As we cleaned up the list of composers (most of which were relatively easily Googled) and hit IMSLP to track down the excerpts we didn't immediately know, one of our teammates noticed a pattern: there was exactly one wrong note in each one. Picking out the wrong notes proved a little more difficult, requiring a sharply critical look at each excerpt. By the time we stopped this one, mostly from fatigue, we had all the composers, around two-thirds to three-quarters of the excerpts, and a handful of the wrong notes. Unfortunately, we couldn't figure out how to extract a solution from there.

What we should have done

We were on exactly the right track here, having nailed the first "aha" in identifying the existence of the wrong note. We were one more "aha" away from cracking the puzzle. According to the solution, the leap we had to make was to assign each letter of the composer's name to his corresponding excerpt. Then, taking all the wrong notes and letters in order would have provided the final clue to solve the puzzle.

Would we ever have figured it out?

This one is tough to say. We came the closest on this one out of any of the puzzles we tried, and because of the "classical music" theme, we were pretty motivated and excited to solve it. But the jump from "find the wrong letter" to "extract the corresponding letter of the composer's name" is one I don't honestly see myself having made. Perhaps I'm just bad at puzzles that require seemingly arbitrary letter indexing.

What can this puzzle teach about good design?

First, a good puzzle should be instructive. It should teach interesting facts, encourage looking at data in novel ways, reward spotting non-obvious but meaningful patterns. "Digging Up Music" did really well as an instructive puzzle. Learning both facts about composers and associating pieces of music with who wrote them made solving this one feel rewarding.

Second, puzzles should belabor their points as little as possible. Once you know what you need to do, doing it forty times isn't any more fun than doing it four times--and after too much repetition, it may start feeling like busy work. "Digging Up Music" toed the line there; thirty-six excerpts felt a bit much, and that was with a few people reasonably well-educated in music solving it. Without any musical expertise at all, this puzzle would have been possible thanks to IMSLP, but it would have taken a much longer time. (Incidentally, "Digging Up Music" wasn't even the most work-intensive music puzzle in the 2013 hunt; I didn't try it myself, but word has it that one particular puzzle involved more than two hundred mp3 clips.)

Finally, there's a big difference between clever and arbitrary, especially when it comes to extracting a solution. Here, taking all the wrong notes and putting them in order would have made perfect sense. Appending the composer's name to the excerpts and assigning a letter to each wrong note was much less intuitive. (Maybe a more streamlined design would have involved playing the excerpt, having it be from an immediately recognizable piece, and having the composer be the answer?)

Sure, to make the puzzle even trickier to solve, there's no end to the number of transforms you could require for a solution. It's completely conceivable to design a variant of this puzzle where you would have to shift each wrong note by half-steps corresponding to the composer's birth month. But by that point, such a puzzle would cease being fun and would lapse into obtuse.

Mini puzzlehunt

If 16 is the root, what do the three composers above represent?

Resemblance Review: Keltis The Card Game

Keltis: The Card Game
I've had the pleasure of trying out several new games in the past month and for several of them I am apt to compare the experience to another game it brings to mind.

The sudden frequency of these comparable games in such a short period of time has enticed me to start a series of comparison reviews which may help you identify a new game with a game you may have already played.

If you haven't been paying close attention in the past five years, you may be taken aback by the sudden inundation of games with the title "Keltis" in them. This rapid proliferation transpired after the original Keltis title took home the Spiel des Jahres in 2008 and a market was born.

Although Keltis Sr. began this modern family tree, we need to go back to the original patriarch in order to fully understand the familial relationship.

Originally published in 1999, Lost Cities was Reiner Knizia's initial design that would eventually evolve into Keltis. The critically acclaimed title is part of the Kosmos 2-player series and has unofficially been labeled the definitive gateway game for couples.

In 2008, Reiner Knizia based his eventual Spiel des Jahres winner on the ideas successful ideas implemented in Lost Cities. Keltis essentially took the mechanics of Lost Cities and put it into a 2 to 4 player board game. Its success would eventually spin off several more titles including Keltis Ór (A Dice Game version), Keltis: Das Würfulspiel (Another Dice game - wait...what?), Keltis: Das Oracle and an expansion to the base game of Keltis.

Also in 2008, shortly after the publication of Keltis, Lost Cities: The Board Game was published. Lost Cities: the Board Game was rethemed in Germany to Keltis while the American Publisher kept Reiner Knizia's intended name. LC:TBG features many similarities to Keltis, they differ in scoring methods and the direction of card play.

From Lost Cities (Card Game) to Keltis (The Lost Cities Board Game) to the actual Lost Cities: The Board Game to Keltis: The Card Game. If you're still with me, we've traversed the roughest part. Keltis: Das Kartenspiel came out in 2009 and builds on the family tree with several additions and continues to accommodate up to four players.

Comparing Lost Cities to Keltis: The Card Game

Lost Cities Overview

Lost Cities has been one of my top ten games for most of the last decade. As a brief overview for those who are not familiar, each player initially draws eight cards from a deck of five colors representing five potential expeditions. Each color consists of cards numbered 1 through 10 along with 3 "investment cards".

Players play a card  either start a new expedition by playing a card onto an empty color or add to an existing expedition, so long as the card is larger than the last. Players may alternatively discard an undesirable card into a discard pile shared by their opponent but risks helping their adversary. After playing a card, players may draw either from the deck or from the top of any discard pile.

Play continues until the deck runs out upon which scoring occurs. Each expedition for a player scores the sum of the cards played minus twenty points, representing an initial fixed cost of pursuing a breakthrough discovery or long lost treasure. Any investment cards offer a multiplication effect of 2x, 3x or 4x (1/2/3 investment cards). Finally, twenty points are added to an expedition if a player is able to play eight or more cards in a single color. Games take place over three rounds.

Lost Cities Analysis

Lost Cities has a number of impressive concepts that cause it to stand out from other card games. It features the concept of sunk costs (minus 20 to start an expedition) that can be increased with the initial placement of investment cards. There is constant risk assessment done each turn as one may have to lay down a low valued card with great uncertainty if they may be able to reach a break even point of that color.

Most of the difficulty for new and improving players comes from initial starting hands, if a player holds only investment cards and low numbers they can't be certain what colors are worth pursuing and what can be safely discarded. The end game gathers a great deal of excitement as the pacing element of Lost Cities keeps everyone on their toes - aiming to run out of cards right as the deck empties by slowing the pace to get all of their cards scored.

The downside is the scoring each round can be a bit lengthy for the time it takes to play. Multiple rounds mitigate the randomness caused by luck and award the better performance.

Keltis: The Card Game Overview

Keltis: The Card Game uses the established rules of Lost Cities with a few key changes. The five normal colors have one or more of each card in the deck numbering from 0 to 10 along with three "end cards" which close out a color for a player. Cards can be played ascending or descending, allowing for 0 or 10 to be equally attractive.

There is a sixth suit of grey cards that can be placed on your turn individually for a single point or placed  on to your established runs to increase their value, but only if they match the value of the card most recently played (I can place my grey 8 on my recently played green 8 and then continue the run).

The new addition I enjoy most is that on your turn you may a numerical pair (e.g. two sixes) and receive a wishing stone of that number (e.g. six). Wishing stones increase your score by sheer quantity, and face value of each stone is irrelevant once captured. This provides some neat ways to both accelerate the game (drawing twice to replace your discarded pair) and a way to get rid of useless cards and refresh your hand with new options.

The most significant change with Keltis:TCG is the scoring. Rather than the face value of cards being used, the number of cards played in a given color determine your score. If you start a color and only place three cards, you score -2 and for each card above that your score increases in a non-linear format (four cards is worth +1 while nine or more cards nets you +10). Capturing zero wishing stones at the end of the game will score you -4 while having five will earn you +10.

Keltis: The Card Game Analysis

The initial experience with Keltis:TCG opened my eyes to new strategies, new areas of interest and a new way to manage my hand. It is a fresh take on a Knizia classic and a good game in its own right.

I love the wishing stones and grey cards which add these new scoring opportunities, but I don't care for the scoring as its a bit fiddly. In Lost Cities you celebrate when that third yellow investment card shows up after you'd been holding out for it over the last six turns, this excitement is gone from Keltis:TCG, you're only focused on color for the and card value is diminished.

The pacing element exists in two player Keltis:TCG, but it lacks the exciting tension. I feel I have time to do nearly everything I want and then some with two players. In Lost Cities I have roughly 22 cards to play and a mid-game evaluation of opening up a fourth expedition is a crucial decision as time is always running out. Three or four players eliminates the pacing strategy as if I delay the game my opponents will continue on their intended pace and I have significantly less control.

The risk assessment aspect is absent from two player, but I feel it would improve with each added player. You often can pick a color and go with it uncontested, and information is plentiful with the longer number of turns and more discarding taking place.


Keltis offers several stark improvements over Lost Cities, but overall I can't say I find the whole greater than the sum of its parts.

A hidden element of player interaction in Lost Cities that disappeared from Keltis:TCG is the ability to force sub-optimal decision making from an opponent. As much of the early to mid-game is discarding and locating cards to maximize value, one can force an opponent into a frustrating position. If I open with a red investment card and my opponent holds only a red four, they must either:

  • a). hold it for an uncertain time period while playing other cards earlier than desired 
  • b). discard the card, giving me exactly what I want when I want it
  • c). start a red expedition, taking a moderate risk that pulls attention from better color options.
All three of these happen somewhat often in Lost Cities, and imagine the impact when your opponent does this in red, green and blue? This is subtle player interaction at its finest.

Lost Cities offers indirect but substantial player interaction but Keltis feels a whole lot more like solitaire. There is often an initial surge for wishing stones as each number can only be claimed once. After than you find a color with an adequate card count you can play and start two or three runs in different colors, but these are frequently the same ones ignored by your opponent. All that remains is ensuring you discard the cards in his colors as he or she can't use them.

Keltis:TCG also has an interesting modification for two players in which 30 cards are randomly removed before the game. I appreciate this shortens the game significantly for two players (although I still feel its too long and not enough tension among possible options) but I would have preferred something like removing all of one color (as grey stays as a fifth "universal" color) and then removing 10 cards randomly. This would shorten the game, increase the player interaction and still allow adequate time for the additional strategic options offered by Keltis:TCG.

Keltis: The Card Game

Originality (0.75/1.0) - Fresh ideas rejuvenate a gold standard
Theme (0.0/0.5) - Lost Cities has a debatable theme, this one isn't even close
Pure Fun (0.5/1.0) - Good by not great, not a game I'd reach for in many situations
"Re-play-ability" (0.5/1.0) - Additional strategic options are enticing, but blend together

Strategy/Luck Ratio (0.5/0.5) - Less useful cards can be turned into scoring option
Player Scaling (0.5/0.5) - 2 and 4 players offer an enjoyable variety, but p
lays best with 3
Parity (0.5/0.5) - The structured scoring leads to more scoring parity than Lost Cities

My Rating:
Overall 3.25/5.0 = 6.5 out of 10

Lost Cities registers at an 8.5 out of 10 for me. Keltis: The Card Game is a wonderful place to visit, but Lost Cities allows one to explore the greater depths with less complexity and is an entirely more accessible culture to discover.

Thursday, February 21, 2013

2013 MIT Mystery Hunt: "A Set of Words"

Amidst trying and failing to piece together internet memes, Steph and I turned our attention to our next adventure in the Hunt, "A Set of Words". This one played out like a Boggle game from hell, each grid full of Q's, X's, Z's, and--of all things--question marks. We had a series of eight such grids with sets of three words accompanying each.

What we did

Of course, our first instinct was to try to find the listed words in each grid. That proved impossible: good luck finding the word "bend" in a grid with no B. We went on to try various horrible permutations of altering the grid: maybe we needed to set the ? to a specific letter, or maybe we were allowed to use it as a wild card, or (since there were obviously no Q's, Z's, or X's in any of the listed words) maybe we were supposed to change those into letters that made the listed words appear in the grid.

The problem with that approach was that it left each grid with too many degrees of freedom. With no grid actually "solved" and no clear approach to a solution, we gave up.

What we should have done

Apparently, the listed words were not the exact words hidden in each grid, but clues to the words hidden in the grids. And the ? tiles were indeed supposed to be set to a single letter for each grid; at the end, the ? tiles formed a hidden ninth grid from which the solution could be extracted. I didn't read much beyond that, simply because we were so far from cracking this one that reading the full solution wouldn't have been terribly instructive.

Would we ever have figured it out?

Doubtfully. Steph actually suggested that maybe we needed to find synonyms in the grid, but we agreed that the number of possible synonyms for each word was too great to actually look for them, and then there was the matter of not knowing exactly what the function of the ? was.

Even if we knew the correct approach, some of the synonyms are so obscure that we never would have figured them out. Take the first grid as an example. "Flex" is near enough to "bend" that it's plausible. "Minx" for "flirt" is a bigger stretch. But in what notion of reality is "next" a synonym for "dorm"? Is this an egregious typo, some MIT inside joke, or an obvious connection that I'm simply missing?

Update: apparently the Next House is in fact the name of a dorm at MIT. Thanks to Ben for pointing this out. Probably anybody from MIT would have known that, and it's clues like these that make us remember it's the MIT Mystery Hunt, not the "world championships of puzzlehunting that occur on completely neutral ground".

What can this puzzle teach about good design?

The point at which the puzzle produces (or requires) its "aha" moment is one of the trickiest and most subtle elements of puzzle design. Ideally, a puzzle starts with a nominally easy task (if one that might require some specialized knowledge), takes a logical next step that might involve lateral thinking to relate separate elements of the puzzle, and finally require a breakthrough of thought to extract the answer from the easier first steps.

"A Set of Words" shifted the "flow" of its solution to require the "aha" moment disproportionately early in the solution. Once you realized that you needed to find not the listed words but synonyms of the listed words in the Boggle grids, the rest of the puzzle would have come together relatively easily. But that's far from obvious by simply looking at the puzzle as it's posed.

Requiring a big leap in logic or intuition relatively late in a puzzle motivates a solver to both start the puzzle--because it begins with digestible tasks--and to push toward a solution--because by that point, you've come so far that you know the final answer can't be too far away. Pushing that big leap to the beginning of the puzzle is a lot more disheartening, since you're left with a lot of puzzle to solve and not much clue as to how to do it.

Taking a big stack of information you've already uncovered from earlier in the puzzle and piecing it together to extract a solution is fun and rewarding. Looking at a mess of Q's, X's, and Z's without any idea what to do with them is just overwhelming.

Wednesday, February 20, 2013

2013 MIT Mystery Hunt: "lol"

The first puzzle that Steph and I attempted in the Hunt was "lol". Someone on our team had labeled this one as involving puns and memes, and all it took for us to jump into it was the p-word. The puzzle presented a series of pictures, each with a phrase in the style of a particular internet meme on the outside and a seemingly incongruous image in the inside.

What we did

Our first inclination was to identify the image at the center of each: Jungle Carbine, some bird, aqueous humor, etc. We got most of them right, though Eastern Kingbird eluded us to the end. (I was particularly proud of Shropshire Sheep, which led me to List of sheep breeds, a Wikipedia page I never thought I'd see myself visiting.) Seeing no obvious patterns among the image names, we tried to identify the name of each meme "archetype". That proved more difficult. Besides not being too educated on internet memes, those that we could identify were often a little ambiguous. Without a clear sense of direction, we gave up on this one with all but one of the pictures identified and weak guesses at about half the meme archetypes. The rest of our team contributed a few more meme archetypes, but we were never seriously close to the answer.

What we should have done

Identifying both the image and the meme type was exactly the right idea. From there, we were supposed to notice that each image name and meme name had exactly the same number of letters with the same space pattern: potassium ion and chemistry cat, etc. Then, the image and meme had exactly one letter in common in the same position (S for the potassium ion and chemistry cat example). Taking this "letter in common" for all the pictures gave a clue to the final answer.

Would we ever have figured it out?

Possibly. Upon later reflection, a puzzle full of internet memes was not the best choice given my general frustration with internet subculture, but presumably some corner of the internet has a catalog of memes. The "aha" moment of noticing the letters/spaces pattern might have taken some staring-at, but it's not out of the question that we would have seen it eventually.

The bigger issue with this puzzle is that the match between phrase and meme type is not absolute. The Flying Dutchman phrase "What if I told you that using Latin phrases doesn't make you sound smarter?" was attributed to Matrix Morpheus, which makes sense now that I know of Matrix Morpheus as a meme, but our team decided it fit the Condescending Wonka meme before we even considered Mr. Morpheus, and I never thought to disagree. Similarly, "Completes Sunday NYT crossword with a pen" sure sounds like something one of those meme-y wolves might say, but who's to say it's the one with a gray background instead of any of the others?

The counterargument is that once a solver has seen the pattern, that knowledge would allow a "weeding-out" of the meme archetype possibilities to include only the ones that matched the letter/space pattern of the accompanying image. But that essentially requires solving one part of the puzzle by solving a different parallel part of the puzzle, which seems clumsy.

What can this puzzle teach about good design?

The solution to a puzzle, and to its intermediate steps, should be absolute and unambiguous. The most obvious answer that fits all the criteria of being a plausible answer should be the answer. In other words, a solution should depend only on what clues are contained in the puzzle and not on meta-factors like the nature of the puzzle itself.

Even more specifically, for a puzzle that contains an "exemplar of a category" clue, the exemplar should be absolutely and unambiguously a member of that category. Nobody could seriously propose that Claude Monet was a master of any school of painting but impressionism, and Family Feud is unequivocally a game show. But if I encountered a puzzle where I had to assign Giovanni Gabrieli into a category of composers, I would be unsure whether to answer "renaissance" or "baroque" and would probably just end up answering the likely incorrect "Venetian school".

Puzzles don't always follow the maxim of "the most obvious possible answer should be the answer." Crossword puzzles, for example, are notorious for incorporating same-letter synonyms or competing transliterations ("tsar" vs. "czar" is a favorite). But puzzlehunt puzzles would do well to eliminate judgment calls wherever possible and focus on the mechanics of the puzzles themselves.

Tuesday, February 19, 2013

Tales from the 2013 MIT Mystery Hunt

"Yes," the skinny, bespectacled kid next to me proclaimed, less excited than convinced. I say "kid," but he was probably old enough to drink. "Yes," he repeated, "it is hunt." He knew because he looked it up, though the gymnasium full of puzzle-hunters probably could have given him a pretty good idea too. And full of hunters it was: just a row behind me was a guy who'd brought his cardboard axe, perhaps so he could defend himself if the guy with the crocodile-head hat decided to eat him.

Protesters wielding cardboard signs sat on the floor of the gym, decrying the change in venue for the hunt's kickoff, ostensibly tongue-in-cheek but probably a real sticking point for at least a few of them. At last, the woman with the six-sided die hat walked in, and that was apparently the cue that the master of ceremonies needed. He launched into his presentation, complete with PowerPoint slides, filled with pop culture references that everyone but I laughed at. It was impossible to say if they were Mystery Hunt in-jokes, or MIT in-jokes, or simply the sorts of thing you would know if you went to MIT.

It was the single most idiosyncratic thing I have ever been to.

And it proved to be the start of a weekend of intensely difficult puzzling, one that was lauded (?) as the longest Mystery Hunt in recent memory, that was proclaimed both "grueling" and "impossible" by two different people in completely separate conversations overheard on the way to the airport as I left Boston, and that took a team of over 150 people to win. So much for our ragtag band of a dozen.

I was in Boston over MLK weekend to visit Steph anyway, and since I had some friends back in Berkeley working on this puzzle, I volunteered to be their "guy on the ground," getting the introduction to the hunt, trying to make it to the "event puzzles". After the surprising intensity of the kickoff, I was sorry to have missed the events, but a combination of other plans, MIT being a little out of the way, and something called "below 20 degrees" proved too much of a hindrance.

In our free time over the weekend, Steph and I did manage to work on a few puzzles. We didn't solve any. We didn't come close to solving any. But, since one of the themes of this blog is that a good design lies at the heart of a good gaming experience, I thought it would be interesting to dissect the design behind some of the puzzles I attempted--after reading through the answers, of course.

This week, I'll be writing a series of posts, each dedicated to one puzzle from the Mystery Hunt. Some of the questions I hope to discuss: What makes a good puzzle? Where's the line between clever and obscure, between challenging and arbitrary? And how does the presentation, aesthetics, flavor text, etc. contribute to a puzzle's quality?

Each post will contain spoilers for the 2013 MIT Mystery Hunt. That should be obvious, and since the Hunt was a month ago, it doesn't seem like much of an issue. But if there's one thing I learned from the cardboard-sign-wavers, it's that you never know what will turn out to be an issue for the Mystery Hunt faithful.

Saturday, February 16, 2013

Game Night Recap: the Weirdest-Ever Game of Power Grid

Making progress toward fulfilling one of my gaming resolutions for the year, I've played Power Grid twice in the past month, and yep, had a great time both times. The first was a few weeks ago; we played on the USA map, and things didn't work out so well for me. In Power Grid, it's essential that you have a place to expand your ever-growing electric empire; when you (like I did) get boxed in between the Great Lakes and the Southeast, it's nearly guaranteed that someone has taken the Rust Belt/Northeast area, and the connectionts (represented by those horrible golden circles) make the cost to get out west prohibitive in the early game.

For our more recent game, we went with Germany, which tends to play a little slower because those massive connection costs are more pervasive throughout the map. The first turn saw the purchase of a few reasonable power plants for the early game; 05 tends to go up for auction first in just about every game we play, and 07 and 08 are useful for pursuing a slightly more aggressive strategy. Then, if it's available to you, it usually behooves you to buy a card in the tens on your second or third turn: I admittedly far overvalue 13, and 11 and 15 are often compelling choices too.

Power Grid's biggest (arguably only) source of variance is which power plants come off a random stack, but there are a couple of balancing mechanisms that ensure the market won't become flooded with early-game plants late in the game (where they are no longer useful) or vice-versa (where you'd love to have them but can't afford them yet). In general, it works great: there are only a couple games of Power Grid I remember playing where a complete mismatch between the board state and the power plant market state persisted for longer than a fraction of a turn.

Then this happened.

One of the "balancing mechanisms" is that once any player has built n settlements, power plants of base cost n or lower are removed from the game immediately. For reference, the eight power plants on display at the beginning of the game are always 03-10, and the first to flip off the deck is always 13. So in this awkward configuration, we had already purchased 05, 07-10, and 13, corresponding to exactly one turn with six people buying one power plant. But nobody had yet built three or more settlements, indicating we were very early in the game.

Power plant 30, on the other hand, is a decidedly end-game plant. It has the potential to provide a lot of power at a relatively small resource cost late in the game, and I've seen it as a cornerstone of quite a few winning strategies. The only problem with 30 in the early game is that it's a massive strain on anyone's cash flow: even if you can afford it, buying it likely means you can't buy the resources necessary to use it that turn, let alone expand your network to take advantage of your new capacity.

So it left us with an interesting decision: is 30 worth buying on turn 2? You're acknowledging that it's going to cripple you for at least a turn (probably two), but with the expectations that 1) nobody is likely to contest you buying it, so you'll get it at- or near-cost, and 2) you will never have to replace it for the entirety of the game. In other words, what's the opportunity cost of punting an early-game turn or two compared with the cost of a late-game scramble to buy an adequately big plant?

Ashley decided to go in on 30, indeed buying it near-cost at 32, which made the most sense for her of anyone at the table. She had my beloved 13, meaning that she was guaranteed to power one city without the need to buy resources. So she could simply bide her time until she could afford three trash (which piled up awfully quickly in a six-person game), settle a few more cities, and fire it up. She didn't quite ride her strategy to victory, but it was an interesting strategic move that I'd never seen before, and it was a fun puzzle to think through.

Another, mostly unrelated, odd part of the game was the win condition. In a six-player game of Power Grid, the game ends on the turn where someone has built their fourteenth settlement, and whoever powers the most that turn wins. Usually, only one or two people reach the win condition, and maybe it comes down to a resource battle or a tiebreaker between them. In our game, Zach, Tom, and I all built to fifteen settlements, and we all powered all fifteen. The game was decided on a tiebreaker, in which Tom beat Zach by about ten Elektro, and Zach finished ahead of me by eight.

Has anyone else experienced a game of Power Grid where the game state was so out of sync with the power plant market? Did it spur any unusual strategy, and did that unusual strategy work?

Saturday, February 9, 2013

Strategy Spotlight Seconds: San Juan

Link to Matt's Strategy Post 
San Juan at an initial glance appears to have a myriad of strategic decisions but what is often learned after just a few plays is that it is a game of two primary strategies, with a distinct selection of sub-strategies.

The idea that there is simply two primary routes in San Juan is initially a concern for lack of variety but in reality San Juan simply has an early demand on players to pick a strategy and run with it. I have always felt San Juan requires the ability to play both the "purple buildings" strategy and the "production buildings" strategy although in order to play either optimally one must adapt to cards dealt and play whichever strategy presents itself first.

While San Juan has some awesome cards with neat abilities, it comes as no surprise it is a game about victory point and key action selection. While I can't get into everything here, I have some tips which are the primary themes while executing the expanded ideas below (focusing on three or four player):
  • We are aiming for at least 35 points. You can win San Juan with fewer and you can lose with more, but 35 points often puts you in a comfortable spot.
  • What we build needs to either be VP/cost efficiency of 2/3 or better (Archive is 1/1, Tobacco Storage is 2/3) or offer a cost savings/income increase that will pay for itself within a few rounds (Quarry or Smithy with -1 cost reduction). We have 12 investments to select and we're looking for a good return on investment and/or quick payback period.
  • Role selection is optimal early in the game in this order: Prospector, Builder, Councillor, Producer/Trader. This is probably commonly how it plays out but selecting producer/trader is drastically inefficient unless you have several silver smelters.
  • With rare exception I will usually pass on the first build opportunity, this game is not about building your 12th building, its about having the most points when someone else does. I'm aiming to finish with 10 or 11 buildings although if a good combination of cards (explained below) come in succession it is possible to finish first with a large margin of victory due to superior income flow throughout the game. Try not to fall behind by more than two buildings to anyone as it is very difficult to stay within range.
"The City Hall Strategy"

Our first strategy is the more grandiose and technical to execute, but has more potential secondary strategies available. Our key opportunity cards here are the Carpenter and Quarry

I find this to be the most desirable combo in the game, as you can build several of our other VP/cost efficient cards such as the Gold Mine, Archive and Smithy without a quantity change to your hand. This enables you to select an alternate role selection such as Prospector or Councillor and still build when others select Builder. If dealt both, pass on building until you can build one without having to discard the other. The Carpenter is slightly better than the Quarry overall, but I'll never complain about having one over the other. The Carpenter is still worth building late in the game while Quarry is not.

Ideally we can build either of the Carpenter/Quarry early (skipping the first build opportunity to preserve options) and add the other at some point later in the game. At this point we are increasing our hand by looking for the low cost, points efficient cards listed above below in order to keep up with the pace of building while positioning ourselves better than our opponents via greater options.
Gold Mine is one of the best cards in the game as it often will pay for itself several times over in the game. The Archive is nice but here for the low cost and VP ratio. I realize the Smithy seems out of place as we don't intend to build any production buildings once we've selected our purple building path, but it is cost efficient and an easy build if it comes up in your hand.
The Prefecture, Tower and Market Hall are the next layer of this strategy. The Prefecture is an excellent piece of the puzzle, as it doubles our income from one of the frequently selected roles each turn, whether we select it or (hopefully) not. The tower is included here as it is efficient and with our selective approach to building, the income will be rolling in and the last thing we want to do is discard down to seven at the end of the round (a devastating offense in San Juan).

The Chapel is one of the secondary strategies in San Juan, although it has two prerequisites to make full use of it:
  • A hand of at least 4 cards that is growing each turn even as you are building, making this an alternative to the Tower.
  • A slower game pace in which players are building less frequently and with more efficiency (this is how you can beat an opponent using the very same build order I am listing here).
Keep in mind using the Chapel comes second to actually building each round, if your hand runs low don't utilize it as you may be a card short on building next turn. Even if you don't use it, the Chapel is not technically a bad building, but there is frequently an opportunity cost involved in building it.
The Statue, Victory column and Hero are the primary elements of the Monuments secondary strategy. At this point their points efficiency should be apparent, but as we have hopefully reduced their cost they are among the elite cards we can build late in the game. I would be willing to build the Statue as early as the fourth or fifth build if necessary but would hold the Hero until at least the ninth build assuming these actions would not diminish my hand entirely. They become more powerful with the Triumphal Arch although it is not necessary if hand size is limited.
The final layer of this strategy is to build at least one of the prized "Super Six" cards. The City Hall should come as no surprise as it will be worth 9-10 points at the end of the game if just a few things fall into place. The Palace may be worth 5-7 points which isn't bad if you haven't seen a City Hall all game. The Triumphal Arch really needs at least two of the monuments to be in place to be worthwhile, but you often don't have the luxury of picking and choosing in some games and it is an okay build in many situations.

"The Guild Hall Strategy"

Our second strategy is probably the simplest to execute, and I prefer it as vanilla as the strategy can be. Our key opportunity card is a Smithy, building it immediately or passing in the first build opportunity if it is not available. As cards accumulate, assuming the City Hall strategy option isn't present, we'll build Tobacco storage, Indigo plants or in dire cases Silver Smelters until we can get a Smithy.

I'm not fond of the Sugar Mill or Coffee Roaster. While they can likely pay for their added cost over the course of the game, their VP/cost ratio is below our expectation. I'm going to be selecting Producer/Trader as little as possible and I expect other players to do the same.
If it is called the Guild Hall strategy, at some point we need a Guild Hall. I'm not opposed to building it early, especially if I have the Smithy in place as the low costs presented by the Indigo Plant and Tobacco storage will allow me rebuild my hand in the following turns after the Guild Hall wipes it out. The key here is to hang on to it once you find one, as players will put them under their Chapels as the game goes on or you may never see one again.

The Gold Mine is always an acceptable build in San Juan and with the Guild Hall strategy I would be willing to build the Chapel when income is flowing. Monuments such as the Statue or Hero probably aren't as effective as usual as the Guild Hall makes Tobacco storage and Indigo plants points monsters with the added benefit of a potential income source but I wouldn't rule them out. I would not build anything else not listed here.
Cards to Burn

The Black market and Library are the best of the remaining cards. I find the Black market marginally useful with the City Hall Strategy but not the Guild Hall strategy. The Library really dominates as the player count decreases as among the two roles strengthened most, Prospector & Builder, one will be available the majority of the time in a three player game and always in a two player game for you to select. I find the Library to be too large a cost commitment to be especially powerful with four players.

Aqueduct is in this upper category simply because it passes the points efficiency requirement. I should elaborate here as the Aqueduct is one of the most unique cards in the game as it can both eliminate the need to ever select Producer while also creating some neat synergies with Market Stand & Trading Post for when you come to the Trader role. It also has some neat uses with the Black market with the Guild Hall Strategy, allowing you to use Producer but never Trader and have an extra source of income that is inaccessible to your opponents when it comes time to build. But it doesn't fit well with our City Hall Strategy and thus I relegated it to the bottom tier of available cards.

The Poor house is probably a moderate opening card, but it is conditional in the build phase. Cards should be selected to build based on consistency and predictability rather than possible utilization.
I'll preface this by saying there is no perfect strategy in San Juan, certainly not the two I've listed above, but between these two strategies neither has a purpose for the cards below. None of these cards are VP/cost efficient and offer little utility to make up for it. 
I'm at a loss to explain my dissatisfaction for the Crane, I don't like to rebuild over buildings that have already been selected for efficiency over the course of the game. Trading post, Market stand and Well each require either sub-optimal role selection (Producer or Trader) or synergy with another card in my sub par grouping (Market stand/Trading post and Well/Aqueduct) in order to succeed.


So what strategy is better, and which route would we take if given a strong opening hand such as Quarry, Carpenter, Smithy, Tobacco storage? I may defer to purple buildings as the potential for a Chapel strategy in a slower game, although the Smithy path churning out Indigo plants and Tobacco storage is fast paced and incredibly effective if others fall behind early and a Guild Hall falls into your sights. 

I do think being a well rounded San Juan player is the most important aspect as I've rarely had the luxury of choosing, and San Juan always delivers its own twists and turns to your game plan.