Is there such a thing as the perfect game? | Hacker News Reader

Is there such a thing as the perfect game?

theguardian.com

14 pointsdrdee56 comments

Aside from a handful of pitched perfect games and Game 7 of the 1991 World Series?

IMO it's Go. It's simple enough that a child will spend more time learning how to hold a stone than the objective, but complex enough that concepts like playing a "divine move" can capture a Go player's fascination for an entire lifetime.

I take issue with this assertion, though:

"In the perfect game, you should still be in with a shot of winning even if you’re not as practised as your opponent. An element of uncertainty is required."

There are few games in life, very few pursuits in which you can remove skill or familiarity and be left with an entertaining test of anything. Maybe luck, which is why craps is the social game. But skill or familiarity? I don't think any sort of "game" exists like this that people take seriously.

Actually, for this train of thought, I maintain that betting on horses is as close to a perfect game as possible. All the research, backside knowledge and modeling might not pan out for you the same way it panned out for the thousands who bet Rich Strike in the 2022 Kentucky Derby because "the name sounded like it was going to pay."

> There are few games in life, very few pursuits in which you can remove skill or familiarity and be left with an entertaining test of anything. Maybe luck, which is why craps is the social game. But skill or familiarity? I don't think any sort of "game" exists like this that people take seriously.

The aim is to play a game where both skill and luck play roles, similar to real life. Poker is one such game.

For Von Neumann, the inspiration for game theory was poker, a game he played occasionally and not terribly well. Von Neumann realized that poker was not guided by probability theory alone, as an unfortunate player who would use only probability theory would find out. Von Neumann wanted to formalize the idea of "bluffing," a strategy that is meant to deceive the other players and hide information from them.

https://cs.stanford.edu/people/eroberts/courses/soco/project...

I learned about Von Neumann and poker from the book "Thinking in Bets" by Annie Duke.

https://www.amazon.com/gp/product/0735216371

Poker is a game where skilled players win more over time, but any player at the table can win a given hand, even if they are unskilled. There are optimal probabilistic strategies, but the cards determine the ultimate outcome.

All types of poker, or particular variations?

* https://en.wikipedia.org/wiki/List_of_poker_variants

All types of poker benefit skilled players over time, but can reward amateurs on any given hand because the outcome of the hand is ultimately determined by the cards (which are shuffled between every hand).

> Although chess and Go are beautiful games, their reliance on strategy means they suffer from a different sort of problem. Unless players are perfectly matched, the outcome can be rather predictable. In the perfect game, you should still be in with a shot of winning even if you’re not as practised as your opponent. An element of uncertainty is required. We decide to play because there is a chance either of us can beat the other.

Go doesn't really suffer from that problem since it offers a great handicapping system. The weaker player can take Black and start with placing n > 1 stones on the board as their first move.

And an even finer control is available, if for instance Black tends to win 2/3 of the time with 3 handicap, but 1/3 of the time with only 2 handicap. One can adjust the komi (points that White adds to their score at the end of the game to compensate for moving second) by half-integer amounts.

Thus you can get very close to each player winning about half the time.

Unfortunately, the skill gap in Go is astronomical.

At my best, I am a 7kyu player. This means that an absolute beginner (roughly 30kyu) has no chance vs me, even with 9 handicap. You need to be reasonably studied at 16kyu level before 9-handicap is fair vs my peak performance. IMO, 16kyu is achievable in a few months of practice and study, but its not something you'll get just casually.

Alas, 7kyu is over 9-stones of handicap away from the top-level 9-dan players. I probably have a shot of beating the #1 player in the world with 9-handicap, but I'm still favored to lose by a decent margin.

----------------

9-handicap is roughly equivalent to letting the opponent move 9 times before even moving once. Its astronomical, far more than a "Queens" advantage in Chess IMO. Yes, the Go handicap system scales extremely well, probably the best handicap system I've ever seen in my life.

But its not quite enough somehow. The depth and complexity to this game is huge.

> This means that an absolute beginner (roughly 30kyu) has no chance vs me, even with 9 handicap.

A beginner will find games on the smaller 9x9 board MUCH more enjoyable. They will likely have a decent chance vs you at 9 handicap, while reducing the very real risk of feeling lost and getting bored by lengthy play before seeing the result.

And you will probably find small board games more enjoyable as well when playing beginners. When they get a little stronger, I would first try the intermediate 13x13 to see if they like it better than 9x9 (there's a chance they prefer sticking to the small board for a while longer).

Beyond that, there is no reason to limit handicap to 9 stones, especially with free placement.

True, but I should note that 9x9 games balance with Komi better in my experience, rather than stones.

Some combination of stones (~4 stones on the 9x9 board), followed up with severe Komi (30+ stones of Komi) is probably better for both players rather than giving the beginner like... 15 stones... on the tiny board.

-----------

The 19x19 board is reasonably balanced up to 9 stone handicap, but I think at that point it makes sense to balance with Komi rather than stones to push the handicap further. Not that I've tried this myself, but severe amounts of stones make the game play very differently and are possibly bad for the beginner player's development.

The author lead me well, I had the same thought before he said it: Backgammon is a good balance between strategy and randomness, the perfect game.

I've been thinking about games as "shaped randomness" lately. Choose the right distributions, pick the right mechanics, stick some art on top to stimulate people's imagination and you have the ingredients of a good game.

Just rolling a dice is fun. What will it roll? There's a kernel of fun there, not much, but something. Add some art, like Snakes and Ladders, and you have a legit game that has lasted for centuries. Finally, add some actual decisions to the game, and now we have something really interesting.

I've also been viewing games through the lens of interaction and watching. I like to interact and make choices, but I also enjoy passivly watching the story unfold. The auto-battler genre shows people can enjoy games where 80% of the game is just watching the consequences of your choices unfold in a nice spectacle.

I find backgammon extremely boring despite being fun in theory. I'm not saying it's a bad game, but it doesn't do anything for me versus a decent RPG, even an old one.

So I think the answer, as Betteridge's law would predict, is no.

Good ol' Betteridge, destroying fun conversations everywhere.

(There's a place to choose words carefully. Talking about games doesn't have to be one of them.)

As for RPGs. They're just backgammon, with more pieces, a bigger board, and some art. But yeah, I probably like a good RPG better than backgammon as well. But no RPG has lasted 5,000 years like backgammon has, so which game is better?

Better to whom? I prefer RPGs. My criticism was aimed at the concept of "perfect game", that's why even a great game like Ocarina of Time isn't an unanimity.

RPGs aren't just backgammon, there has been innovations in game mechanics and the understanding of fun in the last several decades. For instance, idle games would be thought impossible a while back, yet they are probably more popular than backgammon among the youth. Idle games are not backgammon, in many of them there isn't even a way to lose.

Interesting perspective.

I’d add:

- “Hard problems” are often good game mechanics. Think traveling salesman or bin packing

- There are player archetypes. They each want different experiences. The collector vs the masterer.

- Emotion has to fit somewhere in the picture. It’s not just paint on top of mechanics. More likely the reverse is true. “Fun” isn’t an emotion.

- On distributions, too much randomness quickly becomes boring. Too little can be bland. It’s like a spice. How much is called for will vary by dish.

I fail to define what is "fun". I fail also to determine the limit of what is an emotion and what isn't. Emotion must have an effect on my brain, and maybe cause some behavior.

Joy and sadness are emotions, they do encourage specific and somehow different behaviors.

It seems like "fun" is local, and "joy" global. Also "fun" seems to be more immediate and have a specific (shorter) lifespan.

Fun seems to be linked to a situation. When the situation ends, fun ends.

You perfectly identify what is fun while the cause of joy can perfectly be unclear or complex.

But does that make "fun" something else, in nature, than joy ?

Backgammon lacks hidden state.

Hmmm. Is hidden state inherently good or does it serve the same purpose as randomness?

From an individual perspective, an opponent playing unknown cards from their hand isn't much different than a complex and well balanced RNG in some single player game.

After playing 10,000 games of Backgammon on my phone, I disagree.

The author is asking for something which isn’t possible. They are asking for a game which isn’t too random but allows novices and experienced players to have a fair match. That is, they’re asking for a game where the players’ moves are a function whose output is predictable but also not predictable. This is a contradiction.

There are functions which are highly unpredictable but we just call them “chaotic” or “random”. Anything less than that can be learned and thus a player with more experience will have an advantage.

Poker is an example that comes to mind. Most experts hate playing against amateurs because amateurs make weird, unpredictable moves and don't know how to value their hands very well. Skill benefits good players over time, but any player can win any given hand.

Valid yes, but with any sort of volume in hands the pro/shark is going to rob the amateur blind.

I could not disagree more, in my opinion the perfect degree of randomness is 0%. Now, of course that is unattainable: if I play a game of Go, for instance, there's the random chance of "did I manage to get a good rest the night before", the chance of "will I know the joseki my opponent chooses", "will I lose my concentration to unforeseeable circumstance",...

it's essential for a game to separate the in-game and external factors. otherwise there's no possibility of fairness

Why? A degree of randomness makes games more fun.

I think randomness is mostly used to hide the flaws in an imperfect game. In a sufficiently complex game, the natural cause-effects from play should result in new and interesting game states. The opponents inputs + the game rules should create something inherently exciting and intriguing. A poor set of rules can be made interesting by adding dice rolls. Creating an interesting game without randomness is much harder.

I would very much argue the opposite: having a decent game ruined by luck of the draw is the very opposite of fun. It's bad enough when a winning player loses a game by accidentally making a game-ending mistake, having a winning player lose without even making a mistake simply by having the dice/random number generator/whatever saying "no" is awful.

While I agree, I understand that some people can prioritize the feeling of mastery over all other aspects of gameplay. When there's no randomness, then the game is a pure test of skill. If you want to get into a single game as a hobby, then having low chance and a high skill ceiling is a feature that allows for continuous fulfillment of that need.

Even Go has randomness, no? After all, you need to decide who goes first, usually at random

Between the 5.5 Komi vs 6.5 Komi debate, its not very clear if Go has a 1st-player advantage or a 2nd-player advantage.

Superhuman strength AI-Go has given us more data on this subject (suggesting 6.5 Komi is balanced). Since both players are balanced as far as we can tell (first turn starts off with a 6.5 point disadvantage, and the half-point difference forces a winner. No draws), it doesn't really matter who goes first.

If in the future, a super-super AI performs deeper analysis and determines say... 7.5 Komi is the best, I'm sure that Go players all around would update the game rules as appropriate to the new understanding of what a balanced game is. (After all, 6.5 Komi itself is an experiment, albeit the most successful experiment, for balancing the game).

Komi already ranges from 5.5 to 7.5 in tournaments today, I think. The long-term trend has so far always been upwards; as recently as the 50s 4.5 was standard.

Yeah, Komi debate is still alive. But... with AlphaGo / KataGo available today to perform automated analysis of Komi, I'm thinking that the debate will finally end with 6.5.

Though maybe I'm just being optimistic with regards to the debate. IIRC, engines like LeelaZero / Katago has the following result: 5.5 favors 1st player, and 7.5 favors 2nd player. And 6.5 is just right.

---------

Komi has trended upwards because it originally was zero, which was obviously wrong.

Note: Chinese rules changes Komi calculations. 6.5 Komi Japanese is IIRC equivalent to 7.5 Komi Chinese.

You make a move in Go, it feels good, you mentally play out likely lines of play and they feel good too, but there is a grey cloud obscuring some lines of play. You can't quite see every possible outcome, not even close. Uncertainty lies ahead in that cloud beyond our mental abilities. What do we call that uncertainty? It's quite similar to randomness whatever it is.

> In the perfect game, you should still be in with a shot of winning even if you’re not as practised as your opponent.

The article just casually drops this statement and then runs with it. I feel it should be examined, though. There should definitely be range in which a not as practiced player can beat a more experienced one, but to have that decided by luck makes the victory unearned. I believe a perfect game should be one with no luck, so if I win my opponent cannot dismiss my victory as "getting lucky".

Losing to an opponent who got lucky just feels bad, and for me personally, winning because I got lucky feels empty.

Rocket league meets the authors criteria. The fact that it's a physics simulation means that the player who's better at predicting the bounces/other player has an inherent advantage, but if the other player puts a top corner shot at unsavable speeds it doesn't really matter how good your predictions are.

https://www.youtube.com/watch?v=W2kcGht3vGA

For me a perfect game is a game that unashamedly sticks to its vision, whatever that may be. The most recent ones for me are include vampire survivors, risk of rain 2, Baldur’s gate, and Streets of rage 4.

So many games try to cast a wide net and incorporate rpg elements, micro transactions, crafting, etc and it just feels like they are checking off boxes.

Same for me, although I would also say that the level of polish a game can attain helps a lot. Red Dead Redemption 2 and Baldurs Gate 3 come to mind as games that were meticulously worked on to the smallest detail.

I feel like earlier versions of Minecraft stuck to their vision, later ones haven't strayed too far however I think that different choices would have been truer. Risk of Rain 2 is a work of art, from my perspective. I haven't played the last two that you mention.

I'd say Go maximizes the simplicity/depth ratio.

I think games with randomness and a hidden state are superior.

So you're like author of the post? Quote:

> Unless players are perfectly matched, the outcome can be rather predictable. In the perfect game, you should still be in with a shot of winning even if you’re not as practiced as your opponent.

1. (Anecdotal) Back when I was learning Go, I had a game with 9k player with no handicap. I managed to be way above until the endgame where bad prioritizing made me lose the advantage and then some.

2. Ever heard of Starcraft: Brood War? It has a hidden state and sources of randomness, the whole ZvZ match-up is basically RPS... yet Jaedong had whooping 76% winrate [0] in his career, because his micro was on another level.

[0] https://liquipedia.net/starcraft/Jaedong/Overall_Statistics

Randomness is just the thrill of gambling. I understand why it's fun but it's not really a game if winning can be a matter of luck.

Tought question. But, if I had to choose one game, I would choose legendary DooM game. I still play it today (even w/ friends) using opensource zandronum engine. Amount of high quality WADs is astonishing. You can even generate pretty decent maps using Oblige. Amount of various MODs and TCs is also huge. Its still amazing game. Kudos to ID Software for making it :)

from what I recall, it only aims horizontally, right? still fun?

You only explicitly aim horizontally (at least in the original, unmodified version), but the Z of the characters does matter; it will simply auto-aim up and down for you.

But yeah, it's still pretty fun; if nothing else the modding scene for Doom is incredible.

In orginal DOS DooM yes. But why do we have those cool source ports for. ZDoom is very good one, with spawned SkullTag, with spawned Zandronum. ZDoom support jumping/crouching, freelook (limited and very usefull). DECORATE + ACS support for modding. Zandronum adds up client/server multiplayer with dedicated server support. It even have OpenGL renderer for those heavy mods.

I personally play on software with freelook and no autoaim at all. You can shoot just above monsters heads (and you often DO! when fighting strong monsters).

Worms Armageddon. Enough randomness to maintain replayability. Deep strategy elements for those players dedicated enough.

Gloomhaven is a cooperative game I've played a lot with my family where we repeatedly thought we were all going to win then we all thought we were definitely going to lose, but decided to stick it out, then won, with a few actions left. This pattern repeated over and over. That makes for a good game

Gloomhaven… confuses me. I bought a physical copy and found it an awful experience. We were all laughing at the tedium it demanded. We have been playing the digital version and enjoying it but it’s still tedious. I just can’t imagine getting through it on a physical copy.

The balance is remarkable though.

Azad from Player of Games. No, really, I'm very curious to see it modified for IRL usage.

As an addendum: would a global middle class person be able to afford playing this modified-for-Earth Azad?

I completely disagree. I like games that cater towards players who build skill vs some random chance of winning.

Then again I like sim racing, which is like real racing but at home :).

clearly Factorio

I vote for Spy Party as the perfect game

- sufficiently random

- entertaining theme

- 1v1 so no dependence on teammates

- extremely high skill curve with no obvious optimal play but fun at all levels

- short and without one side feeling like they cannot possibly win

- no particular reliance on twitchy reflexes

The only game that satisfies the requirements is Backgammon. A game that allows deep study and expert play, but any game no matter how bad your positions can be turned around with a string of good luck.

Minecraft. Next question

Reply on news.ycombinator.com