Generating Magic cards using deep, recursive neural networks | Hacker News Reader

Generating Magic cards using deep, recursive neural networks

mtgsalvation.com

269 pointsdunstad67 comments

Open articleView on HN

> Shring the Artist

> 2BB

> Legendary Creature - Cat

> Flying

> Whenever you cast a spell, you may return target creature card from your graveyard to your hand.

> 2/2

That's actually a fairly plausible card, in terms of mechanics. It's thematically appropriate for black, it's about the right cost, it's powerful but specialized (returns creature cards only, and only to hand rather than to play), and it's legendary to avoid multiplying its effect. This would work well in a deck with many creatures with "enters the battlefield" effects (common in black, as well as blue and red which are often used together with black), as well as many ways to sacrifice cards. Would even work as the general in an EDH/Commander deck.

The only thing slightly odd is the flavor: it's a flying cat. And with the right art (and perhaps a "black cat crossing your path" joke in the flavor text), that could work quite well.

> The only thing slightly odd is the flavor: it's a flying cat.

Flying cats are actually already in the game. Back in the Planar Chaos days I used this one in a really trollsy black deck:

http://gatherer.wizards.com/Pages/Card/Details.aspx?name=mir...

It's a vampire, seems slightly more legitimate.

There are others, mostly cats riding stuff that flies.

http://gatherer.wizards.com/Pages/Search/Default.aspx?action...

I'd agree except for the legendary part, there's really no mechanical reason for it to be legendary, the effect isn't particularly powerful.

Duplicate it a few times over via any number of duplication mechanisms (common enough in red or blue), and playing one spell would get you multiple creatures back. Since those creatures themselves are spells, it'd be too easy to set up a loop or combo. Making it legendary means you can't easily end up with "play one spell, get back N creatures".

For that matter, creatures are cards, and there are effects powered by discarding cards or cards going to the graveyard, so if you can play one spell to get 5 more cards in your hand, and then discard 5 cards to get an effect...

Not going to lie, I would totally play that in my junk reanimator EDH deck...

How long until someone draws up artwork for Shring the Artist?

How long before the neural network draws Shring the Artist?

I'm not so sure this card is on cost.

My neural network is telling me that it's actually too good at 4 mana.

Don't forget about potential combos like having a bunch of zero cost creatures in your graveyard + anything else that triggers when you play a spell or when a creature enters the battlefield. But even without that, this is a card advantage machine.

It's rather good at 4 mana, though that could be balanced in the metagame by making it rare or mythic.

What zero-cost creatures did you have in mind? Those aren't incredibly common. And in any case, a combo that requires 3-4 specific cards is not overpowered; one that only requires 2 specific cards often is.

you need two zero cost creatures and a way to sacrifice them for benefits, then just trade off. I can think of infinite mana with four cards.

Magic is not a game that considers all combos a bug, and there are any number of ways to get to infinite mana (or infinite many other things) with less than four cards.

Yep, Ashnod's Altar, that card, and any 1-2 cost artifact creatures and you're set.

FWIW, it's about the same cost as Enduring Renewal.

Fair enough, but this card doesn't hinge on its combo ability. Imagine it in BW weenie with things like Anafenza, Kin-Tree Spirit.

Even the synergy with Fume Spitter is pretty great. :)

You cannot balance a card [in constructed] by making it rarer. You just make it so fewer people can run the deck based around it in a given tournament.

It wouldn't be particularly broken in draft anyway - which is the only meta where rarity does impact balance to some extent.

2/2 flyer for 4 isn't dominating Standard, and it would just depend on the surrounding block for how it played in Block. Legendary makes Limited moot.

Magic is an interesting game just from an text analytics perspective. Using a source like mtgjson you get a reasonably sized corpus of text that looks natural enough for a human to comfortably grok but sufficiently templated to apply even naive methods and acquire awesome results.

I built a tool that uses n-grams to find cards similar to each other. The result is comparable to manual suggestions done by users but takes around a minute rather than months to cleanly update after a new set is released!

Is this tool open source anywhere?

Check out my reply to centizen

I am extremely interested to see anything you have public on this project - both from a technical and deckbuilder's perspective!

You can see a working example over at https://preorda.in/card/Kiora,%20the%20Crashing%20Wave/Born%...

I wrote it around a year ago when I was still getting a good feel for go so its a fairly gross ~500 line go package. However, it exports a moderately sane QueryableSimilarityData with the ability to Query that data for individual names. A query result is both the cards that were found to match and how confident it was. It caches after computing the entire dataset's results so it is fast to query after closed.

If you want to muck around with it, you'll need to use a relative import because its just easier for a single file package.

Everything you need apart from the dataset is in https://github.com/Everlag/goPricesBeta/blob/master/utilitie...

Just stick an AllCards-x.json from mtgjson.com in the same directory as binary the package was used in and it should expose the interface I described above to anything trying to use it.

As a note, its non-permissively licensed but message me on reddit under this username and I'll re-license that component. Also, feel free to message me if you want a hand or have any ideas.

Are you the creator of mythicspoiler? I know they wanted to do something like that. As of a few sets ago the 'Similar Cards' in their spoilers were chosen manually.

I am not but curiously enough I got the idea from a short response the creator of mythicspoiler had made on a reddit thread asking for an api to find similar cards or some such. He mentioned that he had played with n-grams but found it to be a dead end.

I ran with the idea and found a version weighted on matched n-gram length gets reasonable results.

If I'm interested in Neural Networks, and I've mainly been working in web development (I know Java, C, and Python). Where would be a good place for me to start so that I can eventually make something like this project?

Andrej Karpathy's blog is excellent as a first introduction. Start with http://karpathy.github.io/neuralnets/

He also has a fantastic article specifically about recurrent neural networks here: http://karpathy.github.io/2015/05/21/rnn-effectiveness/

Thanks! I'm really enjoying seeing the huge variety of domains people have applied char-rnn to. I haven't anticipated response at this scale, to both the original blog post and the code release. (I try to keep a somewhat up to date list of responses on the bottom of the blog - cooking recipes, music, obama speeches, magic cards...)

In terms of learning, I would also encourage people to try the materials from our CS231n class - this is a class I taught at Stanford last quarter with my adviser. It's technically about Convolutional Networks, but most of the class material is building up generic Neural Networks, backprop, and so on. You can also try our IPython Notebook assignments.

Course notes: http://cs231n.github.io/ Syllabus with slides too: http://cs231n.stanford.edu/syllabus.html

Another good pointer is Andrew Ng's Coursera class - that's a thorough introduction as well.

Awesome, thanks for this! I started reading the first link and I'm hooked, that's some extremely interesting stuff

Matlab has a neural network toolbox that you can use to train small recurrent networks.

There is a book that I can link if you are interested.

I am interested :)

This is from one of the people who helped write the toolbox in Matlab. There is a page limit on the Amazon publishing service, so that is why it's posted here[1]. Chapter 14 describes notation for the recurrent networks and explains their training.

[1]:http://hagan.okstate.edu/NNDesign.pdf

If you are comfortable in MATLAB this tutorial is great: https://theclevermachine.wordpress.com/2014/09/11/a-gentle-i...

I've found http://neuralnetworksanddeeplearning.com by Michael Neilson to be an excellent resource.

This is the one I've always recommended as well. It's a little longer, but very in-depth.

Awesome, will look into, Thanks!

Keras http://keras.io/ is a neural networks library for Python that is rather easy to get started with compared to the alternatives.

I also recommend the tutorials on http://deeplearning.net/tutorial/ (Python / Theano)

And Andrej Karpathy's blog is also a great resource for explanations of deep learning concepts in simple terms: https://karpathy.github.io/

Wow, this has to be my favorite HN post. I couldn't finish reading it at work due to tear filled laughter.

I think it is less funny if you don't know anything about MTG. Which is interesting - if you don't, or don't know it well, these cards seem fairly indistinguishable from real cards. The differences are hillarious.

Mointainspalk.

The author remarks that most of these are nonsensical. As a non-magic player who lives in a house that sometimes hosts magic games (and thus clicks links re:mtg), I found every line to be entirely plausible overheard banter.

> 3: add 2 to your mana pool

Another card nerfed by the removal of mana burn! (http://archive.wizards.com/magic/magazine/article.aspx?x=mtg... section 3, "Mana Pools and Mana Burn")

Combos with Power Artifact.

This thing is really cool. Also, he's using mtgjson.com, which I've used in the past and it's really awesome.

A real gem of gibberish interpretation by the thread: https://i.imgur.com/AWTQ0bM.jpg

This actually uses the same neural network discussed here a few weeks ago: https://news.ycombinator.com/item?id=9584325

These are recurrent neural networks, not recursive. That's a different model, also used for NLP, but not so readily applicable for generating text AFAIK.

Would be interesting to do this for other other games (that interest me more ;)).

Netrunner ICE generation would be interesting. I'd also like to run some algorithm over all Star Wars LCG cards and have it generate new arrangements of objective sets. [for those that don't know the game you don't build your deck by picking cards like in Magic but rather predefined bundles of 6 cards (objective sets)]

Reminds me: it feels to me like MTG card text (excluding flavor text) obeys a formal grammar of some sort. Does it?

If it does, the NN's could possibly be made a lot more powerful if you put a "parser" in front of its input end (so it was being fed ASTs instead of text) and a "code generator" on the output end (to convert confabulated ASTs back into text).

I'm not sure if it's officially a formal grammer, but there are definitely correct and incorrect phrasings for certain effects, and those are changed over time with official rewordings posted for old cards.

It's one of the ways that unofficial card spoilers are judged. If there are phrasings that sound "out-of-style" for the effect, its usually judged as a fake from the first pass.

My understanding (based on some old blog posts[1]) is that while there's no (published) formal grammar there's definitely a set of internal rules that WoTC uses for determining how things are phrased. What is considered "technically correct" has changed somewhat over the years. I believe Sixth Edition was the first attempt to standardize and make consistent card text. Generally though cards that do the same or very similar things are going to follow consistent word ordering.

For example take Shrewd Hatchling[2] and Coral Reef[3]. Both are fairly different cards, but notice the wording of the first part of their respective card text:

> Shrewd Hatchling enters the battlefield with four -1/-1 counters on it.

and

> Coral Reef enters the battlefield with four polyp counters on it.

Cards with that sort of ability are (almost) always going to have it phrased in a way that fits the pattern of "<card name> enters the battlefield with <n> <type> counters on it."

Lots of other sorts of common mechanics are the same. Additionally, there are the so called "Golden Rules"[4] which are important to take note of when parsing the text of a card.

While Googling for links for this reply I found Gleemin[5], a "rules engine that aims to fully and correctly implement the game's tournament rules, an AI player programmable with expert knowledge and an interpreter for the game's language (the rules text on the cards)"! Seems you're not the first person to think about this!

[1]: http://archive.wizards.com/Magic/Magazine/Article.aspx?x=mtg...

[2]: http://gatherer.wizards.com/Pages/Card/Details.aspx?multiver...

[3]: http://gatherer.wizards.com/Pages/Card/Details.aspx?multiver...

[4]: http://mtgsalvation.gamepedia.com/Magic_Golden_Rules

[5]: https://github.com/stassa/Gleemin

How about IDE that suggest you next line of code? Or next variable, or method to call?

Add static analysis and it can actually be fairly useful. E.g. val name = object.[maybe you wwant to use getFirstName]?

http://pliny.rice.edu/

http://www.darpa.mil/program/probabilistic-programming-for-a...

Believe me, it's been thought of (it's been in my head ever since my colleague Frank Seide introduced me to RNNs 4 years ago, I'm sure I wasn't the first!). It is one thing to generate code that looks like code, quite another to generate code that is useful. Character-by-character probably isn't good enough for this task.

Anyways, I talk about how to use probability (but not compute it!) in http://research.microsoft.com/pubs/179363/mcdirmid12.pdf

The old Jungloid technique for Java factory automation/type conversions (I have an A, I want a B) seems like a possible use for this kind of thing. http://dl.acm.org/citation.cfm?id=1065018

There has been a lot of work on code mining. It still isn't that useful, and no one has used RNNs (well, there is one paper I remember reading).

If a prediction engine can consistently predict your next line of code, wouldn't that be a sign that you're doing way too much copying and pasting instead of abstracting common patterns?

Thanks for the giggle fit.

The cards do get very "real" very fast. Love it

This is a very interesting idea. I'd like to see it played with 100% of the cards being generated through a system like this. 4 players, 4 automated decks of random (possibly nonsensical cards), that would be _so much fun_!

You probably would draft instead of just automate the deck.

I think the mechanic is hard to generate a deck that utilize these mechanic and build synergy and combo out of these decks.

Oh totally, draft would be fine, but even the idea of playing something randomly generated like this would be pretty fun.

A question I have about this, and I'm new to neural networks so this might be entirely wrong, but would it be possible to create another neural network that returns whether or not a card is overpowered or not and run that on all the cards generated by this neural network? Then you could get a set of fairly balanced cards even if a few might be somewhat nonsensical

Wonder what MaRo and the gang thinks about this. I am pretty surprised at the results so far from that post. They mostly don't make sense but I'm sure if you went through say 100 of them you're bound to find one or two legitimate ideas in there.

Unfortunately they might not be able to view this - as it might be viewed as "unsolicited design" - which they can't view due to company policy.

Stops people suing Wizards if they print a card identical or similar to a homebrew card.

I was very excited to learn about recursive neural networks and how they are different from recurrent neural networks. I began imagining self-similar fractal topologies and automatic convolution layer creation etc... Disappointed to see it was just a typo :-)

There are actually recursive neural networks. https://en.wikipedia.org/wiki/Recursive_neural_network

Magic + AI = mandatory upvote. Nice one!

I am enamoured that geeks can find amazement in computer generated content. I wonder if Shakespeare scholars felt the same with the recent neural net creating Shakespeare-like content. What is to come, when cgc starts to become even more amazing? A individualized, computer generated TV-shows that hit all your favorite subjects?

Probably offtopic, but the second or third poster in that thread had "Stormfront" in the sig.

Well.

It doesn't appear to be anything obviously associated with the Nazi group, though.

The full signature is:

    My Custom Sets
    
    Avant Block: Avant -- Stormfront

Which suggests that it's the name of the in-progress second set of a block of custom user-created cards the person is working on. Quick glance at the posts the signature links to gives me the impression it was chosen simply as a generic "storm is coming" fantasy-setting name.

And for people who know a bit about Magic: seems a lot like "Scars block without the Phyrexians", even down to the user posting a variation on Koth.

Reply on news.ycombinator.com