An Indie Dev and a AAA Dev discuss strategy through the lens of Magic: the Gathering. Hosts: Forrest Imel forrestimel.com/ Gavin Gray Valentine www.ggvart.com/ Join the Distraction Makers Discord: / discord
I think one of my favorite Magic heuristics is that of "make them have it", particularly regarding counterspells. You play around counters for so long, but at some point the upside of a risky play is so good that you essentially call their bluff. It really simplifies the infinite possibilities into the binary state of "they have it or they don't". And for me, once I've made that decision to "make them have it" there is almost no pain point if they do have the answer.
That's always been my approach to counterspells. Just make them use it, if I refuse to ever play my good cards, that's as good as having them get countered anyway.
I cast Etali this weekend fully expecting it to be countered, it was still my best play even with that possibility. They didn’t have it and I got 3 free spells off it.
This is partly true but a bit overgeneralized. Playing around counterspells has tempo implications that make it not that simple - its far from just being a matter of a 1-for-1 trade. Depending on the matchup, speed of decks, etc., you can force a ton of mana inefficiency and even sometimes completely blank counterspells by manipulating tempo. Virtual card advantage, tempo, and mana efficiency are all things you can gain. If I see 2x counterspell in your hand with thoughtseize, and then wait and play 2x 2-drops when I have 4 untapped mana and you have 3 untapped mana, I am coming out ahead in a variety of ways. Soft-counters like daze or spell pierce you can *literally* blank by manipulating tempo as well.
The whole "bolt the bird" heuristic is way more interesting to discuss if you have another possible card to play. If your only possible options are to bolt the bird or not it's very clear that you should do it, even if you knew the opponent doesn't have a 3 mana play (on turn 1). If you on the other hand have you own proactive play besides the bolt, the correct play could change if you knew your opponents hand.
The logistics of bird bolting goes a little deeper in that it heightens the power ceiling per turn for the bird owner. Being able to power out 3, 4 and 5 mana plays on subsequent turns pushes the bird player's maximum output higher than the bolt players and it's hard to claw back from an early lead against you when all other power level aspects are roughly equal. If you bolt a turn 1 bird, you're also trading a resource 1-for-1 instead of letting them gain momentum by stifling several turns' worth of mana advantage. Magic is cool, man.
Excellent discussion. Though what makes Birds SO big of a target is not just the extra resource, but that it provides ANY color. Meaning if you bolt the bird, you can choke your opponent out depending on what colors they need of spells in their hand. Hence the bird is a bigger target than an elf.
The biggest reason is that, any deck that runs a dork is likely to really want that extra mana to power out a huge haymaker, or to ramp out even more ramp. Taking out the bird has a high chance of curbing their velocity, and a fair chance of stranding cards in their hand if they screw. People often count the dork as a land (if they can cast it) and don't consider what happens if it gets bolted. That it taps for any color is a major plus, but "bolt the bird" applies to Noble Hierarch, Skirk Prospector, and any other 1 for 1 you can get, as well as better removal than Bolt. Think about it a bit more if its stronger removal like Dismember or Path of Exile with a higher cost, but a Fatal Push, Sheodred's Edict or a Portable Hole on an Arbor Elf turn 1 or 2 is probably worth it.
@@LibertyMonk Cradle Control in legacy runs a critical mass of dorks to create mana and targets for Natural Order. If you kill some of their dorks, it doesn't really matter because they have so many, and the mana multiplies with cards like Gaea's Cradle.
How much worse than the basic scenario does it need to be before bolting the bird is no longer a good play? Dismember the Ignoble Heirarch? Solitude the Llanowar Elves? Brainstorm into Miracle Terminus the Wild Cantor? Assassin's Trophy the Ragavan?
A big strategy i like to use in magic is what I like to call “oh look a scary thing” Where I’ll play a big threat card that has nothing to do with my strategy early on in the game mostly planeswalkers that distract my opponent and makes them waste their resources to get rid of it because most players get tunnel vision when they see a planeswalker
This is also why cards like Thoughtseize and Gitaxian Prone are so valuable in MTG. They give you hidden information in the form of what is in your opponent's hand. Knowing what is there is insanity. It's also why cards like Oracle of Muldaya and Courser of Kruphix hurt the player who played them as much as helps them. Knowing your opponent's draws is another set of hidden information that both players can benefit from.
Interestingly, just seeing your opponent's hand isn't good enough for 1 mana -- Glasses of Urza is unplayable for instance. It's the card advantage in addition to the information that makes those cards worth it.
@@nsnick199 This is true. But effects like Thoughtseize get worse the latter in game that it's cast, to the point of being completely unplayable in situations where both players are simply drawing and playing whatever is on top of their deck. The only one of "the look at target player's hand" cards that remain good the longer the game goes is Git Probe, since it's a redraw. But, knowing what is in your opponent's hand is still valuable knowledge to control decks especially.
This is why my favorite card is Zur's Weirding. Every card is revealed, and anyone can pay life to stop someone from drawing a card. The game grinds to a halt.
awesome channel to stumble upon, subbed. I'd love to see a breakdown of "whos the beatdown?" because i think of that article ALL the time, it's applicable in so many ways to so many games.
Also when it comes to all the unknowns on MTG and how that affects all the "bolt the bird" interactions, that's the power of cards like Thoughtseize, Kozilek, Duress and so on, cause even if they miss the full discard effect, you still get away from it knowing your opponent's hand and being able to make future decisions more effectively
If they miss, that's really bad, you spent significant resources on improving your information from "I have a good guess" to "perfect until next turn". Nobody plays Peek or Telepathy because as good as *knowing* is, you usually either are forced to "make them have it" or have the luxury of playing around it regardless of if they have it or not. Spending a single mana on looking at their hand is not worth it. Even Surgical Extraction and arguably Gitaxian Probe aren't worth running just for the look at their hand (and deck) for 0 mana.
@@LibertyMonk Right. Git probe is usually played in storm where upping spell count is also beneficial. The peek is a nice bonus to know if it's safe to go off.
@distractionmakers gitprobe is also the exception where the resources expended (2 life) are minimal, especially since it replaces itself. There's good reason it's banned and restricted in every format besides commander.
@@LibertyMonk I disagree on Git Probe, being free to cast in every way (you don't lose mana nor cards) it is bonkers, but you don't play it to have information, you play it to have less cards in your deck.
Erebody be talking about power creep. Whens the last time we talked about keyword creep. Any more keywords and commander decks are gonna have to come with dictionaries.
Oh man I agree. complexity creep is getting really bad. The issue becomes how do you make new unique things? You either go up in power or out in complexity.
Just had a lecture/discussion with my intro design students talking about strategic skill. Really wish the algorithm put this on my feed last week instead of today. As this is great. The way I phrase it to my students is that strategic skill hinges on identifying patterns and potential patterns through the game. Players with greater skill can take in the information they have and create a list of potential patterns that can emerge and make their decisions accordingly. Like a player critically thinking about bolting the bird/keeping that bolt for something else. Blue has to make this type of decision all the time with counter spells. To use a non-magic example, calling a play in football/Madden. You have all of the relevant data about the game state: score, ball position, down and distance, personnel on the field and as a play caller you have to mix that with the data you collected before the game started and the data you have been collecting throughout the game to set the next pattern in motion based on the interaction of your decision and the opposing coordinator's decisions.
Chess has imperfect information even with perfect information, simply by having the possibility space of future moves be incalculably huge past a certain point. Heuristics then develop around things like positions, piece values, to the point where top players will sacrifice pieces for positional gains even if they can't perfectly see the line to victory.
You have not heard top players use the word "calculate" enough then. Even strategic positions have top players like fabi say "calculate" every few minutes. Also, there is a Wikipedia page on "Complete_information" with the subpoint "Examples of games with incomplete but perfect information..."
My favorite Magic heuristic is “walk your opponent into a place where they can make mistakes” - the Great One. And of course, “Who’s the Beatdown”? - Flores
New here, Great discussion, tho I'm thinking even as time goes on the Heuristics can change or fall out of favor, the current incarnation of "bolt the bird", at least for the Modern format, is Turn 1 Ragavan as a "turn 1 removal check". A lot of it comes from power creep, but where OG bolt the bird is more abstracted since the innate pay-off for Birds of Paradise is partially hidden, or later delayed, Ragavan has become infamous both in how much it snowballs momentum for a 1 mana creature, but also that Modern's power-level is so Optimized that if you don't have an immediate answer in your hand, you "just lose". In a way, the heuristic only developed through experience and expert knowledge has become normalized, if not Mandatory for Viability, within the format, because the cost and consequences are so immediate and harsh, that there really isn't any room to fail or for sub-optimal decisions that still let you compete. I do wonder if it's worth checking and comparing between formats just to see, how the heuristic still distills an awareness of the game system but the "punishment" for knowing vs not is much steeper. Of course, this can be heavily contrasted by say EDH, the casual multiplayer format, which partially via social negotiation and standardization, it's heavily frowned upon to "bolt the bird" since pretty much everyone has an almost unsung agreement to acquire as much resources un-impeded to "do their cool thing" so to speak. Obviously this might be Apples to Oranges of Comp skill development vs variety in Casual formats, but I do think it's fascinating to acknowledge that part of a system still gets different treatments depending on what style of play is desirable.
Another reason to bolt specifically a t1 bop is that bop can make otherwise sketchy hands look better. Killing it might not be slowing them back down to your pace but instead completely locking them out for multiple turns
Timing is the biggest problem i notice with new players. The stack and preceding events before they happen are hard to learn for new players but they are a huge segment of the skill expression in the game too, I wouldn’t want to take that away from magic but it’s for sure complicated
For sure. I remember when magic online came out and being able to see how all of the timing actually worked was a revelation. Magic arena helps with that now.
Most of the decisions in magic that happen at instant speed amount to "I've already held it up, the mana is already spent. Using it now to trade 1 card for 1 card is fine, unless I'm specifically waiting to counteract the payoff. And if I don't do it now, they'll gain tempo." You've already made 2/3rds of the decision when you decided to hold up interaction or go "shields down". So, you're kind of right when you say Magic doesn't need the "on your opponents turn" actions to have these interesting decisions, but it also doesn't hurt as much as your saying, because you've already done most of the thinking ahead of time. Its also worth noting that a lot of the possibility space in Magic is *functionally* equivalent with other bits. Any creature with 3 toughness or less will die to lightning bolt, any creature with 3 power is probably a 6-turn clock (because of fetch lands), any board wipe will wipe the board unless you can counter it, etc. So, if you know what a deck of thar archetype generally looks like, (how many land, how many cantrips, removal, creatures, engines, etc) you can crunch odds, and see if you have the luxury of playing conservatively, or if you have to go all in and hope they won't draw it until its too late.
I think the advantage of waiting comes from you not having to present your line of play to your opponent first. You get to maintain information asymmetry and are able to make more informed decisions because of it. The most interesting interactions come when both players are aware of this dynamic. The stack could effectively be looked at as microturns, but with only instant cards being playable. So instead of handling all of that on the stack, what if the stack was just turns? As far as functionally equivalent possibility space. I agree. This is part of the complexity overload for new players. They haven’t learned to chunk the information into like cards yet. Colors help immensely with this and it’s why the color pie shouldn’t be violated. It’s also why formats are so important, they limit the possibility space to a manageable level. I think the idea of a commander is also a great way to telegraph what archetype a deck might be.
As far as "the mana is already spent" that is true in some decks and meta's but I would no go as far to say most interactions amount to that. Sure, a red deck wins that holds up a mountain is probably going to throw something regardless of what happens, but in a classical blue temp deck you are not just holding up mana for counters, but also draw instants and in modern times flash creatures. Thus, even though you are holding the mana up you have to decide weather or not hindering the opponent or progressing your game plan will put you further ahead. Likewise a slower control deck does not need to drop it's instant speed interaction early if it is not threatened even if it has mana up as it is more concerned with long term value than efficient mana use. I do agree with your other point about the possibility space is fairy finite thus one is capable and rewarded for using deck and meta knowllage to crunch odds to determine how risky you should play.
You’ve only done the thinking ahead of time if that’s the only card you’re able to play during there turn. If I have an untapped Island and pass holding up spell pierce and opt, I’m making 3/3rds of that decision in response to my opponent.
I've always been of the opinion that "bolting the bird" is usually the wrong decision. My heuristic is "remove the payoff not the setup". Given that removal is generally cheaper than big payoff cards but more expensive than setup cards and I can only run so much of it and still be developing my own gameplan it's better to leave the birds out. The side benefit is that opponents often see an untouched bird, think they're safe, and play right into my answer.
I don't think heuristics are entirely tied to imperfect information. Rather it is a way for players to approach a game that is too combinatorically complex for them to solve exactly, but that offers a reasonable way to chunk strategic ideas into a gameplan that usually works. Magic does this via unknowns, an unknown hand or an unknown deck, but a game like Chess does this by being too complicated to calculate out fully in the early-mid game, especially for players who aren't at master level. This is where the heuristics of attacking the center, developing pieces and gaining space come from. These aren't single decision heuristics, as in the case of bolting the bird, but they are heuristics: attempts to approximate perfect calculation from first principles via pattern recognition.
Good insight. I was probably too board in my statement. What I’m getting at is heuristics can be created in much more simple board states with unknown information. You can limit the number of possible decisions to be made by a player, but the influences on those decisions are as big as the number of cards playable in that moment. This is extremely advantageous to new players as they are most susceptible to analysis paralysis. You’re moving the interaction complexity to a space only knowledgeable players would consider.
@@distractionmakers I agree, and this is definitely a big boon to onboarding newer players. Unknown information helps people make decisions by limiting the scope of information they have to think about, and forcing them to come up with heuristics. Some of my favorite games don't do this, and as a result do have a much harder time getting new players to be able to play somewhat strategically. Specifically, the 18xx genre of railroad themed board games, and some of Splotter Spellen's games tend to be perfect information multiplayer games in which players essentially create a shared economic space together. These games tend to have very entangled game spaces, where it is very difficult to isolate the impact of a single action you take because it creates a cascade of shifting incentives. I adore these games but new players have to essentially learn by observation, because games with inexperienced players often go off the rails. These games are definitely more niche as a result, although they do reward handsomely any investment made into understanding them better.
@@Sidnv Yes, this was what I came to comment too, there's a decent number of board games with perfect information and they all still require heuristics. Even the bolt the bird scenario if your opponent played with their hand and deck face up, attempting to stare at it and compute the exact value of bolting the bird over the entire rest of the game is unrealistic except in the simplest of cases. You'd just have a far better heuristic with that extra information.
@distractionmakers 100% agree. I'm a competent chess player and returning M:TG player (from the 90's, now about 1 year playing mostly draft). Chess tests your ability to visualize, predict and calculate scenarios - variations, in chess terms. There are still heuristics for sure, and a big part of the game is developing the pattern recognition skills so you can do those tasks efficiently. But it's a common adage amongst chess players that you don't win games, you lose them - perfect play results in a draw. Magic is different - you're basing your pattern on guessing - all the time! There are boundaries that make those assumptions more manageable but all the time you're pitting resource use against potential outcome - and the timing of the responses is a lot more critical. You have a good spell and you opponent has two blue mana? You think twice. There's also the bluffing and double/triple/quadruple bluffing - 1 power creature attacking on a 2/2... combat trick? Or tricking me into letting it go in? and if there's indeed a combat trick, is it better to have them spend it now than later? and if I have it, should I let my trivial 1/1 die (maybe it already provided an ETB value) to make my opponent think I don't have that resource? That's why both games needs a play clock :)
Good insights here. I'm making a fundamentals of strategy series that aims to explore, identify and categorise strategical concepts by drawing examples from different games (videogames, tabletop, competitive sports). I have a passion for this type of deconstruction and analysis and have often found myself searching for content that just doesn't exist. So I thought I'd have a go at making it. Does anyone recommend any articles or other content that really helped them strategy-wise?
In xcom, if it tells me I have an 80% chance to hit, it means I have a 20% chance to miss but it always feels greater than that. I've missed do many 90% shots.
The reason you bolt the bird is because you are playing constructed, and if your opponent has the bird he most likely wants to skip 2 mana cards and is heavily focused on 3
Part of bolt the bird is likely you’re ruining their turn 2. Decks designed around “birds” might need the color fixing OR be designed around having 3 mana on T2. Either way, you’re likely hurting their turn 2 so badly you can gain an insurmountable advantage.
I think unknown info isn't necessary for this kind of heuristics and strategy, if the game has other things limiting your ability to calculate stuff. in a fighting game, you generally know every option your opponent has at any given moment. you still have to build strategy around it because it's in real time, so you can't pause and calculate the optimal thing to do in every interaction. put another way, the "unknown information" is what your opponent is about to do. you know what all the options are, but don't have the time to figure it all out in the moment.
Right! We’re mostly discussing this for turn based games, but any limitation put on the player in calculating the possibility space would work. Like Chess played with a turn clock.
I don't quite agree with your statement that you need unknown information to get heuristics. I mean, on a surface definition level, sure, a heuristic is for something you can't fully compute or memorize. But I disagree on the implication that you need hidden information or output randomness in a game to get players to use heuristics. For example, one of my favorite games is Terra Mystica which has randomized setup and then 0 random or hidden information midgame. The setup randomization along with multiple players ensures that you have never played that exact combination before and can't rely on a memorized script while the vast amount of turn options ensures it is noncomputable.
For sure. Random chess is an example that would also technically solve the issue. My point is more that if there is always some amount of unknown information you can be sure that the game won’t enter a solvable state at some point. If set up is random and pieces are removed eventually the game could become solvable again.
"if I have perfect information, I'll know" Ah. The problem is the possibility space is still too large. Consider chess? It's a perfect information deterministic game with no shuffling of the starting position. And yet even the best engines that can think as good as a master and millions of times faster are not totally sure what optimal play looks like. The equivalent of bolting the bird may look good at a depth of 20 but bad at a depth of 30.
Being able to evaluate how you’re going to win the game with the cards you have tells you if you should bolt the bird or go face. That’s the strategy of mtg.
The problem here is that this is a failure of understanding what "bolt the bird" actually means. It's not "always kill your opponent's mana dorks". It's specifically "kill your opponent's turn-one mana dork". Talking about it in any other context isn't pointing to a failure of the heuristic, because that's not what it's talking about.
