AlphaGo

While I have been reading instead of playing, the most exciting news in computer gaming has been Go. Chess-playing computer programs have gradually moved from “plays a standard game pretty well” to “almost competitive with a good human” to “consistently beats world champions,” reaching the end of that progression about a decade ago. Go, contrarily, has long been held out as a game at which computers will have trouble making gains because the search space is huge for a 19×19 board, moves have long-term consequences that make evaluating individual moves difficult, and play has generally been seen as more intuitive and so less open to computational brute force.

A year ago, the best Go program was competitive against a good amateur player. In the last six months, Google Deepmind’s AlphaGo has leaped to “best in the world,” beating the European champion 5-0 and now beating the world champion 3-0 with two games to go.

frame from a manga. two young men face a computer. one says, "but they say it'll be another hundred years before a computer can beat a human at go." the one at the keyboard replies, "I don't need a hundred years." There are three things I would like to note here. First, the speed of that jump is ridiculous. Go has long been one of those “at least a decade away” computing problems, like the ones that have been forecast as “20-30 years away” for the last 20-30 years and are still 20-30 years away today. AlphaGo is the first computer program to beat a professional player without a handicap, and then it went on to beat the world champion. That is going from “can’t beat a professional player” to “beats the top professional player” in one step. This is not the gradual progress we saw with computers and chess over decades, this is an escalation in power levels that would make anime blush.

Second, this is not simply a matter of Google having massive computing power to throw at the problem. The chess world champions play on supercomputers and evaluation trillions of positions per second. The world champion version of AlphaGo uses a distributed computing network, but they also have a single-computer version that beats the distributed version about a quarter of the time. We will see if the human world champion gets one win in the series, but this suggests that a much less powerful version of AlphaGo would still be a top player.

What I find most interesting is that humans seem to be fairly bad at evaluating how AlphaGo is doing. AlphaGo optimizes for probability of winning, not its current score or a projected score at the end. So the human analysts are commenting on how the computer seems to be making mistakes, that it is not capturing territory, and oh look gg the computer has somehow gotten itself into an unassailable position. One of the reasons computers have been bad at Go is that a single move now can have subtle implications 50 moves later; AlphaGo has made the jump to where its subtle moves look like mistakes to observers until it wins. It is probably not the case that the computer was playing a close game and pulled ahead in the late game. It seems more likely that the computer was steadily pulling ahead but in a way that is not obvious until the late game. Here is Eliezer Yudkowsky exploring this point at length. Bonus thought: human commentators were probably assuming that AlphaGo would lose, so odd-looking moves were probably mistakes rather than subtle brilliance; in light of consistent wins, I am curious if the human commentators will now look more closely at its moves for hidden strengths, rather than starting with the frame “this is another lousy computer Go program.”

: Zubon

Bonus thought 2: when I see Eliezer referencing “Path to Victory” in that post, I cannot help but see him referencing Worm, which he has read and commented on before.

Brutal Honesty

Playing more Coup with new players, I find that honesty is a surprisingly powerful and common strategy. I do not know if the people I was playing with were hesitant to bluff, but almost every loss of influence/death was caused by false challenging someone else’s claim. It was a game of self-inflicted injuries.

I am sure I have played with other groups that lied constantly. I wonder what the curve is: start with minimal bluffing, because you do not know the game well enough to lie believably; lie shamelessly now that you know the odds better; go back to honesty how that people are expecting you to lie shamelessly?

: Zubon

Bleak Blaughust Game Dump

It’s a weird thing: the mind. Throughout summer, the household was in a nice routine. At the end of July it was like a flip switched. School was only two weeks away, but suddenly both the kids and Mrs. Ravious wanted school to start. Stress piled on, and instead of viewing Blaugust as a fun challenge it became an obstacle. I knew that I might get the first five under my belt, and then I would spectacularly fail. So I followed Homer’s advice and took most of August off. Refresh the juices. Continue reading Bleak Blaughust Game Dump

Nevermore

Nevermore was a friend’s most cherished Gen Con acquisition this year. It is entertaining enough, with a mix of elegant and unnecessary mechanics.

Nevermore is primarily a drafting game. There are five suits of cards, and you get five cards. Pick two, pass three; you have five again, pick three, pass two; finally, pick four, pass one. This gives you a mixture of control and unpredictability. You then compare cards with the other players. Four of the five cards work the same way: whoever has the most subtracts the second place number and does that much. So if you have four attack cards, and someone else has three, 4-3=1 and you deal 1 damage to someone. The fifth card suit, ravens, is an anti-card that cancels your other cards, but it becomes powerful if you can get most of your hand as ravens. First person to six victory points or last human standing wins.

As a drafting game, it works pretty well. Because all five of your cards are up for choose/pass each round, you might change strategies completely after your first pass. Play tends to resolve pretty quickly. Our player who plays draft Magic twice a week did rather well in his first game, so I am led to believe decisions are meaningful even if there is fair amount of weakly controlled randomization. Some players complained about sitting next to (or worse, between) skilled drafters.

It gets a bit unnecessarily complicated because there are special rules for a variety of special cases. The special rule for getting a hand of 5 ravens makes sense; I have yet to see it happens, but it just fits the game for that to be something special. Then there are special rules for the unlikely circumstances of 4+ damage (not attack cards, damage after comparing to 2nd place), 3+ healing starting at full health (same healing vs. cards), and 5+ radiance cards (cards this time, not after comparing). And then there are two pages of rules relating to raven cards.

One worthwhile piece of complexity is that players turn into ravens instead of dying. They stay in the game and keep playing, with some chance of becoming human again and getting back in the game. Raven players have slightly differing rules for play.

Gameplay is mostly quick, the rules are mostly short and simple. The special rules for a half-dozen unlikely things feel like unnecessary cruft, but it was fun to play.

: Zubon

Tokaido

I tried Tokaido this weekend. It was an appallingly awful experience, but I am led to believe it could be better.

I played a five-player game using the Crossroads expansion. Those links will explain why that is a bad idea. Between the two, you pretty much have Candyland with vastly more complexity and a small bit of strategy.

After playing, I looked at reviews of Tokaido, and almost every one quickly said that the game is not Candyland. That is a lot of smoke for there not to be fire. In the game I played, the range of sane decisions was small, the randomness of the results was large for most squares, and most players were new so we did not even know the range within which randomness occurred. My random character was from the expansion and depended on that randomness, and then I got boxed out of even using most of it due to pawn placement. Such is life. I just repeated as a mantra, “it’s Candyland,” because how much can you care when your decisions have almost no impact on your outcomes?

So pausing here, I would be interested in playing again, but with no more than three players, not using an expansion, and I want a chance to read the cards so I know what the experienced players are basic their decisions on. Having started with two paragraphs of complaints linking to two pages of complaints, why would I be interested in playing again? < --more-->

Tokaido is thematically lovely. It is an equal and opposite of Blood Rage, where a peaceful theme plays into the mechanics. The goal of Tokaido is to walk along the coast. Whoever has the most fun wins. That is literally what you are scoring; whoever has the best time meeting people, visiting seeing the sights, and buying souvenirs wins the game. The visuals are peaceful and elegant. Gameplay is just moving along a line and picking up cards. It is a nice idea.

At a level beyond Candyland, there is cutthroat tactical play. Where you place your pawn matters, and people can aggressively block each other to thwart each others’ goals. Once you know what is hidden in all those decks of cards, you can rationally gamble on the outcome of picking squares with random results. There is gamer play here as well as casual play.

Make sure to have excellent lighting. The other great problem we had was just seeing what was going on. The icons are very small, and several of them are similar. In low lighting, there is not a lot of difference between gray and light blue, especially with glasses like mine. “Is that a hot spring or a view of Mount Fuji?” Tokaido is also one of those games that uses icons in place of words, most of which are good, but some of which only make sense if you already know what they mean. (The worst of that was probably the Crossroads expansion, which I will again curse. One of my friends was enthusiastic because one part helped solve a cash flow problem he had with his shopping-centric strategy; more than doubling the complexity of the game to add exactly six yen is a really bad trade-off.)

I like the idea of the game, and several people were enthusiastic. Under the right circumstances, it looks like it could be a good game.

: Zubon

[TT] Icons and Text

Many Eurogames seem to be minimizing text in favor of icons. This makes sense given an international market; if you are selling your Eurogames in Europe, you have a dozen languages to contend with, and if you are aiming for the lucrative markets in North America or east Asia, having lots of German text will not help your sales. Sadly, many of these icons are horrible, maybe meaningful if you already know what they mean.

For example, Kingdom Builder explains what location tiles do using an arrow for “move” and a somewhat curved arrow for “place.” You will be forgiven for remembering that backwards, and good luck seeing the curve from across the table. And then some do not follow that format, and expansions add more symbols. The icons are an okay reminder if you already know what the location tiles do, but if you do not, they will not help you unless that tile follows the standard format and is close enough to be clearly visible.

For example, Hyperborea uses truly awful symbols to explain what technologies do. As with Kingdom Builder, some follow a basic format that is easy to understand. Some have tiny variations. Some have unique symbols that do not appear elsewhere and may not be explained anywhere. You have not overcome the language barrier if players are having a five-minute debate using inferential logic to figure out what a symbol is supposed to mean. Given the number of similar but slightly differing technologies, these do not even serve as much of a reminder.

very small icons from the Tokaido board game For example, Tokaido has three space types that are mixtures of gray and light blue, some in close proximity and sometimes rotated on the board. The accompanying symbols for each space are measured in millimeters.

My favorite linguistics blog has an entire category for what it calls “nerdview.” This is described as “writing in technical terms from the perspective of the technician or engineer rather than from a standpoint that would seem useful to the customer or reader” or “a linguistically misleading communication in which the failure is not of grammar or meaning but of failing to keep in mind the viewpoint of the reader rather than the specialist (possibly nerdy) view of the writer.” More simply, the symbols make perfect sense if you are the game developer, but they can be incomprehensible for someone trying to learn the game. We have previously discussed MMO icons that are about 10 pixels across and hide multiple sentences of information. They can actually make perfect sense to a veteran player who can tell at a glance the fine distinction between a dragon and a fire-breathing drake. Players newer or less visual may be stymied.

I appreciate the attempt to internationalize the game with a standardized appearance. I sometimes question the effectiveness of the solution.

: Zubon

Tesla vs. Edison: War of Currents

Artana had a big booth dedicated to Tesla vs. Edison at Gen Con, and the game certainly merits trying. I have just the one playthrough under my belt, but everyone enjoyed it. The game supports multiple parallel and conflicting strategies and mechanics on the path to victory.

The theme is dear to nerd hearts: the competing current standards and the electrification of America. Tesla is a geek hero, so players are naturally drawn to him, with Edison cast as the villain. The game remains neutral on that question, although Edison’s special ability is to ignore patents. The potential player characters also include three lesser names in electricity.

You are running an electric company. Your company founder gives you certain stats and abilities. Supporting luminaries are recruited via bidding. Each round, you and your luminaries can research technology, build a project, engage in propaganda, or play the stock market. You are competing not only for science but for hearts and minds, as the value of your stock is affected by whether the public thinks AC or DC is the current standard of the future. Yes, people have strong opinions about that. At the end of the game, whoever is holding the most valuable stock portfolio wins. (You start with four shares of your own stock, so unless you are pursuing a very aggressive stock market game, highest stock value should tend to win.)

The strategy comes from which actions to pursue and in what order. Continue reading Tesla vs. Edison: War of Currents

Honest Deception

I finally got a copy of Coup at Gen Con, not a new game but one I enjoy. Coup is a game of bluffing, deduction, and assassination. You have two hidden cards that give you abilities, but feel free to lie about what two they are and use whatever abilities you like. If you are called on the bluff, you lose a card; if you are challenged but were telling the truth, your challenger loses a card.

Playing a bit at Gen Con, I found that I am better at telling the truth unbelievably than lying believably. This is a drawback for Mafia/Werewolf, but it works brilliantly in Coup. If I never bluff, I have no tells that indicate I am bluffing. But he could not really be telling the truth every time in a bluffing game, could he? Especially when you get a really unlikely draw, like two of the same card, or when you use the Ambassador to pick an unlikely pair of cards rather than a powerful pair.

And if I can train everyone to believe that I never bluff, I have free reign to start bluffing…

: Zubon

Friendly Kickstarters

Two Kickstarters to point out from friends of the greater rat-slaying community:

Project Gorgon is back on Kickstarter and has already met its goal. The $35 and $50 tiers look like your best bang for your buck. We have occasionally blogged about Gorgon as a boutique MMO, with lots of interesting ideas.

Cultists of Cthulhu is a game in the category of Arkham horror, cooperative with one or more cultists waiting to betray the team. If you are like me and like to judge based on the rules, they are online. It advertises meaningful decisions in bold text, so the game is in the spirit of what we love here, although I cannot speak to the execution. The add-ons look rather pricey.

: Zubon