[alasdair_]

Magic may be the best game ever devised.

Unlike chess, it’s stochastic and unlike backgammon, it has hidden information and bluffing. But it’s more than that - the cards change the rules of the game itself.

On top of those layers, the fact that each player builds their own deck makes the game asymmetric and ultimately the meta-game of building the deck to beat the Keynsian beauty contest of optimal deck selection becomes the most important part.

It is this meta game that makes me think it will be a long long time before we have a machine learning model that can play the full game (and the meta-game) consistently better than an expert human player. At least while new cards are being added to the game.

The game is turing complete. It can have infinite loops and crazy, ridiculous complexity and is gloriously fun to play. I’d love to hear about any AI projects that have taken a serious stab at playing a complete game.

[Smithalicious]

Another great innovation it has over chess is making people pay hundreds if not thousands of dollars per year to the creator. I'm only being half sarcastic here; monetization seems like a difficult aspect of game design for an "analog" game like this.

[alasdair_]

>Another great innovation it has over chess is making people pay hundreds if not thousands of dollars per year to the creator. I'm only being half sarcastic here; monetization seems like a difficult aspect of game design for an "analog" game like this.

In the original rules of the game, players had to ante up a random card from their deck every game. There were other cards that added to this ante during the game itself, much like doubling in backgammon. The point was to make the person who went out and spent a lot of money on the deck likely to risk more "monetarily" than the player who only spent a few dollars.

I played this way for many years and I still think it's the best way to play the game, with the trading of cards to get back the ante card a key mechanism to improve one's deck.

[Smithalicious]

Some parts of this are appealing but overall it seems like a bad design; new players and players with worse decks will have to give up cards far more often. Their cards will be cheaper on average, yes, but having to frequently replace a random card sounds like it would be very tedious. This also icentivizes people to pick on weaker players.

[alanfalcon]

There’s just enough luck involved that the risk-reward ratio was probably mostly right in those early days, especially as the game wasn’t designed with the expectation players would ever have perfect information about scarcity and relative power level of individual cards.

Even so, it wasn’t fun to play with ante, so most people didn’t do it; Magic smartly went with the flow rather than fight against it, and that’s a huge reason why the game persists successfully to this day.

(Details gleaned primarily from episodes of Mark Rosewater’s “Drive to Work” podcast, at least the best episodes of which are mandatory listening imo for any designer of any stripe, especially one who is also at all a fan of MtG. Yes, Garfield is the game’s creator, but Maro is the game’s central nervous system and has been for more than 15 years!)

[gbrown]

I think it's fun, but I HATE that it has such a strong pay to win component.

If a video game let players spend thousands of dollars on high power abilities to completely dominate the lower paid tier players, they'd be ridiculed in many circles.

There are certainly ways to play around this, but it's definitely a thing.

[alasdair_]

>I think it's fun, but I HATE that it has such a strong pay to win component.

At most tournaments, all the players have essentially equal access to any cards they want to play with. Some formats are more expensive than others to enter of course, but that's about the same as any other sport - I don't race thoroughbred horses or drive Formula One motorcars either.

In terms of non-tournament (or tournament practice) there is nothing stopping people from taking a sharpie and writing "Black Lotus" on a one-cent card and playing as if it's the most expensive card in the game. People do this all the time.

[antoinealb]

There are a lot of different formats. In Limited for example, you buy the boosters you are going to play with when entering the event (you don't bring a constructed deck). This makes it very level and pretty cheap (pay to enter instead of pay to win). It is also quite competitive and is a format often played in championships.

[whereismylogin]

And sets these days are designed with limited in mind. Limited has gone from this weird thing you do at a PTQ side tourney after you flunk out round 4, to a real, super fun format!

[thaumasiotes]

> And sets these days are designed with limited in mind.

For reference, sets have been designed with limited in mind since Mirage in 1996.

[Aeolun]

I thought it was interesting to see that at the start they explicitly tried to make the common cards just as powerful as the rare ones.

That’s definitely changed for the worse, as well as making individual cards more powerful.

I used to make fun of YuGiOh for being full of ‘this card wins you the game’ cards, but magic has very much transformed into exactly the same thing.

[thaumasiotes]

Mark Rosewater (the head designer) commented on this very issue in the most recent State of Design article:

> Two years ago in my "State of Design" articles, I said we'd let complexity get a little too high. Last year, I said we overcompensated and ended up with complexity a little too low. I'm happy to say that we've found the middle ground and have been producing sets that seem to be hitting the sweet spot.

> The key to this success seems to be us towing the line of complexity at common, but upping the amount of complexity we allow at uncommon. This allows us to take in-theme things for the set that would normally be rare and pull them down to uncommon to allow us to raise the as-fan of the theme. A good example of this would be the planeswalkers in War of the Spark. The uncommon planeswalkers, in a vacuum, would probably be rares in a normal set, but by allowing ourselves to lower their rarity, we were able to infuse War of the Spark (especially Limited) with the planeswalker theme.

> Another component that allowed us to pull this off is a willingness to be more aggressive with the power level of commons, especially answers, to help make them more relevant without having to up their complexity.

( https://magic.wizards.com/en/articles/archive/making-magic/s... )

The power level of commons was long one of their primary defenses to the question "why do you print bad rares?". Combining the introduction of Mythic rarity with the axing of staple commons did terrible things to the balance of power across rarities. As alluded to in passing here, they seemed to believe that powerful commons made the game more confusing to play and/or less fun, hurting their potential market.

You can track the issue on Doom Blade ( https://gatherer.wizards.com/Pages/Card/Details.aspx?multive... ). In M12 (2011) it's a common. In M13, as part of their commitment to continuously rotate which cards fill which roles, Doom Blade is replaced by Murder ( https://gatherer.wizards.com/Pages/Card/Details.aspx?multive... ), which is also common. In M14 (2013), Doom Blade is back! But it's an uncommon, where it stays for the next several years while they publish articles about how their new philosophy of design means you shouldn't have broadly useful removal at common for under 5 mana. But between M19 and M20, Murder shifts from uncommon to common. Murder is still much weaker than Doom Blade. But the philosophy of "no common removal unless it's either intensely situational or too expensive to play" has disappeared.

They're trying to balance money extraction, player demand, and the broader health of the game ("after playing 20 games, do I still like this?"), and feedback on those three issues has very different patterns of immediacy and accuracy. (And, of course, money extraction and player demand are in direct conflict with each other.)

[tw04]

That's entirely by choice? You can play in all sorts of tournaments. If money is an issue, play a sealed deck tournament. I get that MAY run you a hundred bucks, but if even THAT is too much, play with friends and set rules about what cards are allowed.

The beauty of magic is that people with even a modicum of spending money can find a competitive game.

[whereismylogin]

I built my stepson a deck for about 50 bucks, and he won more than one reasonably-sized (>15 players) tournaments, as a 13 year-old.

Yes, if you currently want to be relevant you probably need Oko and friends, but if you want to play cheap, you can.

And if you want to play commander for cheap, check out The Commander's Quarters on YouTube, all deck techs are 25 or 50.

[rantwasp]

there are many formats. in some the field is as level as possible, in some it takes years to understand all the mechanics and interactions.

the cost to enjoy it is minimal if you’re not playing competitively and even if you are paying attention to the meta matters more than investing thousands of dollars in the game

[city41]

I really have no doubt at all that Magic is the greatest game ever devised. However, I do think it was a better game in the early days before strategy and meta was disseminated so broadly across the internet. Each local community had their own deck styles and approaches, and a big part of the game was in discovering new combos and decks on your own. I feel that part is a bit lost these days.

[arvinsim]

I don't think information availability affected the game so much as the economic barrier. I mean, if people have infinite access to cards, there would be a lot of experimenting going around.

[city41]

I think both are factors. In the 90s I'd estimate my Magic circle was about 100 people or so. Sure that's a lot of room for experimentation, but nothing like the millions of players today.

[alanfalcon]

The tools exist for people to experiment as if they had infinite copies of every card ever printed, and the people who most enjoy that sort of thing definitely make use of those tools!

[arvinsim]

That’s right but they wouldn’t be able to test on real tournaments though.

[Steko]

> Magic may be the best game ever devised ... [basic strategy plus] ... stochastic ... hidden information ... [self-extending ruleset] ... asymmetric ... meta-game

Magic was incredibly innovative but I think just having this list of characteristics was not so revolutionary in the early 90's. A number of wargames and RPGs that predate Magic have most of them and a few have all of them.

[meroes]

Can I take this opportunity to ask a stupid question?

Is ML anything more than pattern recognition? If I could tally nearly every game state vs game state -> win %, then run a simple if [state], then {} program, then what do I miss out on vs an ML approach? Is the magic just in how we feed good data sets to an ML algorithm so it can efficiently mimic the above much quicker?

[rsrsrs86]

This is a very clever question.

If your problem's domain is not too big then you are basically right, but in practice things are more complicated, larger, noisier, etc. I am conflating many things here of course.