Thinking back on all the games I've lost, this analysis is depressing confirmation that I might just not be that good at this game lol. Super interesting read though!
Same. I've played the game for 40 hours and made 40 attempts, and 'won' only 4 times. 2 of them were daily challenges (and don't really count, they include very rule-breaking things), 1 was Standard for Ironclad, and another was Standard for Silent. I haven't even attempted Ascension once yet. It took about 25 tries on Standard before I won my first time, too.
I've even watched a few people's playthroughs and going over which cards are better than others and combo well together, and my win rate is still terrible. Just lost again this morning in Act 2 with another Defect run.
Meanwhile this guy is saying the best players are winning 99% of the time on Standard, and nearly the same on Ascension level 15? Wow.
By the way, for anyone thinking this game isn't really for you, you might want to give it a chance anyway (watch a video at least). I don't usually care for Rogue-likes, and I've been pretty tired of Deckbuilders, but I got into this game hard, pretty much immediately, after putting off trying it for forever and finally deciding to give it a chance.
One thing that really makes it awesome for me is that you know basically exactly what the enemies intend to do and how much damage they're going to do to you, so you can puzzle out how to prepare for it, should you kill this person who's going to do 18 damage to you, or reduce their strength, or build up your block to absorb it, or take the hit this round so you can go after something else, etc. If you've ever played Into The Breach, it feels a lot like that, but with RPG battles instead of a Tactical map.
Similar for me - there was a real sense of accomplishment on my first win in Slay the Spire on the easiest difficulty setting. To think that players have >90% win rates on that setting, much less on Ascension 20 is remarkable.
I've learned a lot from Jorbs - a twitch streamer. It's surprising how much statistics and planning can go into any given fight. It's possible to puzzle out how much of the fight will go ahead of time, with knowledge of one's deck and the enemies moves.
I'm aware it's contextual and there are different strategies for winning, and there's a reason for each card being in the game. But I wasn't aware of how some cards combo together, or some situations/battles where they might be stronger than others, etc.
And there are plenty of discussions online for pretty much every card on how strong or weak people think the cards are overall (like is it worth picking it up if it's offered to you late in a run because the payoff might not be worth it, or if you see you're getting a specific boss that act, etc), and most of the cards have been buffed or nerfed multiple times since the game has come out, so clearly the developers have been convinced that some strong may end up stronger than others on average too (and are trying to keep things pretty even as much as possible, I assume).
One thing you may want to try is sticking to a certain class for a while. It'll give you a better feel for how all of their cards work together, and you won't "forget" about certain aspects of the class. I found that I hit a wall around Ascension 8-10 before I started trying this.
Yeah, I agree, good advice. When I decided to win my first standard game, I at first floated around from character to character, but then decided to stick with Ironclad until I won. Took like 20 tries, but I finally made it work. Silent took less tries, currently focusing on Defect.
A really important point with card choice: even ignoring your current deck (and you shouldn't ignore your current deck), the best cards are Act dependent. At the start you want to pick flat high-damage cards that are immediately useful. In act 2 you need cards that scale with the number of enemies, because mobs are common. In the final act it's all about card synergy and things that enable you to manage your deck.
With that said, as the Defect, Defragment and Capacitor are nearly always a no-brainer. Capacitor not so much in Act1, because nothing lasts long enough for it to pay off. And if you've got anything that adds orb slots, Consume becomes really powerful (even without, you can do 10 damage a turn from two orb slots instead of 9 from three, which isn't nothing).
Loop builds are very strong as well, but trickier to get right because you need to pay close attention to your front orb. Finally focus+frost can easily give you so much block most enemies can't cut through. (Watch out for Reptomancer, though, she doesn't respect anyone without DPS.)
I had 1 Loop+, 2 Defragments, and 1 Capacitor that last game I lost this morning :) I got 'All for One' as my random rare card to start with, so I was doing my best to get a bunch of 0 cost cards, and I also ended up with 2 relics that let me pick cards to shuffle in every battle (Nilry's Codex and the Toolbox), along with a Jack of All Trades card, so I got to pick cards to help whatever my situation was each battle and turn.
But I was a little too weak on my defense that go (had some frost generation, but not enough), and I made a dumb mistake attacking the Masked Bandits instead of losing all my gold, who did 50 damage to me before I killed them, bringing me down to 18 health, then I couldn't get to a campfire until I fought 3 more regular enemies, and the next enemy brought me down 6 more, then the second enemy after that got me down the other 12. Oops.
I finally beat defect using, among other things, darkness plus recursion. It never occurred to me that recursion would retain the power generated by the dark orb you just evoked. All for One to pull upgraded 0 cost recursions to hand.
Nice. Darkness+Recursion is quite nice in the base game, but I've never found it very effective in Ascension. (I'm sure someone with more experience could analyze why, exactly.)
Good luck with Watcher. The biggest advice here is: play it safe until you have multiple cards to get out of wrath. Watcher is super-powerful and _very_ easy to mess up.
Ouch, that's a nice deck, but I suspect I wouldn't have bothered with the 0 cost cards (Silent's 0 cost cards are amazing, Defect's are meh). Glacier I _highly_ recommend. Autoshield and Stack are great too. (These two are brilliant for surviving early turns before you've got enough frost up.) Use your upgrades on Defragment, then Loop. Upgrading Zap and/or Dualcast from your starter deck is great because they become zero-cost. Creative AI, especially upgraded is amazing. Oh, and don't sleep on Biased Cognition. You can only use it right at the end of a battle, but it can end a battle very quickly. (Unless you've got a source of artifact,
I'm a bit tapped out at A8, but it's amazing how the base game goes from nailbitingly impossible to seeming pretty straightforward. (The mechanics contribute to this: the easier a battle is, the more upgrades you do and the more likely you are to grab an elite relic.)
Note that this is a general tendency and not a rule. Act 1 benefits from an AOE card for the sentry elite and the slime boss, damage scaling will help against most bosses at any time, and so on through act 2.
For the ironclad, you might take a cleave, or a whirlwind, or a spot weakness, or an inflame in act 1. The first two give you AOE, and the last two will help scale damage if you find a limit later on.
I have no idea if this will help you in particular (depends on how you play right now), but something counter-intuitive about deck-building in general that helped a friend of mine up his win-rate in StS a lot is that you need to consider the cost of taking a card.
StS lets you not take a card when you are offered one, and often this is the best play, particularly as you go later into the game and have a deck that is stacked with powerful cards.
Fundamentally having more cards is bad, because it increases variance in what you can draw. Obviously taking a good card can be worth that cost, but often players fall into the trap of assuming the best card of the options given is worth taking, when really the question should be "does this card beat the average card in my deck, or does it dilute the power of the deck", because you are reducing the chance to draw one of the other cards in your deck.
Obviously there are edge cases: sometimes something may reduce the power of your deck right now, but introduces the chance for later synergy, which may be worth the cost—especially early on, but in general it is worth considering not taking the card.
The ideal deck is always the minimum number of cards that allow you to pull off your win state, everything else is just stopping you getting there (this is also why cards that allow you to draw cheaply are good, and one of the reasons why removing cards from your deck is powerful).
An important idea is that you're not only drafting to make a generally good deck, you're drafting to defeat the immediate challenges that are coming up.
If you're in Act 2 and headed for an elite, you have to think, "can my deck beat book of stabbing? gremlin leader? taskmaster?".
If right now your deck sucks against the book, and you're offered a disarm or caltrops, it might be worth picking it even if it's doesn't generally make your deck stronger. Cause otherwise you might get stabbed to death.
On the flip side, Demon Form is a very strong card, but it's practically a curse for most of Act 1, because it's too costly and slow. So sometimes you might skip even a strong card if it doesn't help you right now.
Yeah, I definitely don't mean to imply that the comparison to your current cards is a hard and fast rule, but more that it's a good thing to start thinking about if you never skip cards.
That concept is what has helped me get over the A15-20 hump. It's more important to draft cards that shore up immediate weaknesses than it is to draft cards that might pair well with other effects in your deck. Draft for the short-term, not the long term, because stronger short term victories means you can pathfind more aggressively and acquire more gold/rewards.
I'd say that's probably the most important "ah-ha" moment where my winrate really started to significantly improve. That and knowing when to min/max # of combats while pathfinding.
What you're saying is true, but I think people with card-game experience tend to over-skip the card rewards in the name of consistency. Low difficulty levels reward this, because the bosses are the only real challenge and holding out for cards good against bosses helps your win-rate.
At high difficulty levels you need to pick a card (and possibly buy cards at shops) through most of Acts 1 and 2 because you won't be able to reach the boss without a bunch of damage commons.
Yeah, my advice was squarely aimed at those who are struggling with the base game difficulty. Your advice is sound for those who get better for sure, the game will definitely punish you for being greedy and trying to keep your deck too clean as you ramp up the difficulty.
I do skip taking cards, but I probably don't skip it enough. It's difficult to get over the psychology of 'I beat the battle, I get to pick my reward.' It's my reward, I should pick one, and anyway I get a choice of 3, so one of those 3 ought to be good for me, right?
No, not necessarily. But my brain wants to pick one anyway.
It takes a while to get a feel for the strategy and a lot of counterintuitive or seemingly impossible things become straightforward. Considering there's a lengthy stretch just unlocking cards, it doesn't sound like you're in some particularly unusual place, 40 hours in.
>Fundamentally having more cards is bad, because it increases variance in what you can draw.
StS has a number of enemies that shuffle bad cards into your deck, so this dillution works both ways. On higher difficulties more of these cards get shuffled into your deck. Trying a "thin" strategy with no countermeasures in such a fight (high damage output to end the fight fast?) can get you railroaded. You also start out with 10 starters that you want to dillute because virtually any card is better than them. Cards with "draw" and "exhaust" written on them also let you workaround the problem of drawing starter or negative effect cards. Picking up one or two versatile 2-cost cards early (attack or block card) can give you a good way to use your energy even if you draw mostly starters or negative effect cards while avoiding excessive deck dillution later on.
StS does offer you worse cards at the beginning of the game, and limited ways to remove or upgrade these cards, so picking up too many cards can burden you later on.
Really to be good at StS deckbuilding mostly requires you to know which enemies will come up when and building your deck sufficiently to survive them without shooting yourself in the foot long term too much. Jorbs is usually focused on building his deck to beat the hardest fight in the game while absorbing the minimum deadweight to survive until that point. Watching a stream can give you a rough idea of what a good "pace" for adding new cards into your deck is.
Jorbs's way of thinking about the game is very well considered and I'm sure useful for new players but his skill at taking turns makes it hard to copy him. Try it - take a seed from one of his runs, copy his decisions, and play the fights out yourself. You'll get flattened.
Jorbs can reliably reach the heart with a much more challenging deck than you or I can. He's also playing a game mode that imposes challenges that don't exist for new players. Just shoving a lot of good cards in your deck might be pretty good advice for new players up until about ascension ten.
> By the way, for anyone thinking this game isn't really for you, you might want to give it a chance anyway (watch a video at least). I don't usually care for Rogue-likes
...Slay the Spire has nothing in common with roguelikes?
It’s procedurally generated, dungeon crawler, with collectible items and permadeath and resource management. I don’t like arguing over definitions, so call it whatever you want, but those are a number of components that feed into the appeal of roguelikes!
You're better than me at the game. I have close to as many hours and attempts and I've only ascended once. Although probably should be at ~3 ascensions and made a couple blatant tactical mistakes fighting bosses with The Silent.
I was about where you were 40 hours in. Then I bought this game on switch so I can take it everywhere with me and have since dumped about 500 hours into the game. I only recently got all characters to A20. Still working on A20 heart kill and a few other achievements.
It's always intrigued me seeing other people pick up the game and master it much more quickly than I did. There's some people that get their A20 heart kill just 100 hours into the game.
I really enjoy the feeling of slowly but steadily gaining knowledge and intuition about the game - in the 500 hours of play there's been no shortage ah-ha moments throughout. Truly one of the best $/entertainment value games I've ever purchased.
A1 is usually considered easier than A0 because you get more elites but they aren't very punishing yet.
You're possibly bringing baggage from other deck builders, mentioning combos and better cards is an example of this.
Generally you want to judge cards for what they do for you right now (in the next few fights) to maximise rewards and build long term value. Try not to build into "just in case" combos or synergies or you bleed HP before it shows up. The exception is if you are confident your deck can handle the hallway fights then you might take picks for the boss, and if you can handle that then you can start building for the future.
Apart from that, general pickup rules are: Act 1 is front-loaded damage, Act 2 is AOE damage with some damage scaling for the boss, Act 3 is scaling offense and defense (without damaging your other factors too much). You usually want to be plugging holes rather than building on your strengths but there are always exceptions.
One example: A single dark orb can be enough to kill Champ on Act 2 as defect. Any more is excessive.
Likewise. I regularly die in act 1, and after finally achieving ascension 10 with all characters I’ve spent weeks trying to kill the heart at ascension 0 with ironclad with little success.
You can kill the heart? I thought that was just kind of like a fun thing you can see at the end of a win where it shows how much damage everyone has done with it, kind of like a communal scoreboard. There's more to it than that? Note: I've only played standard runs and daily challenges so far.
In ascension > 0, there are three keys you can collect through the run by choosing them in lieu of other rewards. If you get them all, there's an abbreviated Act IV to play: a hearth, then a shop, then a special elite, then the Heart. The Heart is much, much more difficult than the Act III bosses.
There’s an achievement (per character), as well as it being the « true end » of the game (and it increases the challenge as you need to beat a buffed elite, sacrifice a chest and sacrifice a hearth).
You don't need Ascension, just to have unlocked Ascension. With that said, I swear it's easier to beat the heart on A1 than A0 because of all the elite rewards.
The heart itself, well it's definitely challenging, but it only works for very specific builds. I went to some effort to beat it once on each character and basically have no intention of doing so again.
It's been a while since I've played, but if my memory is correct you have to collect 3 keys, each obtained by sacrificing an opportunity to improve the run (Pick it instead of a relic/rest/something else)
Blue you get from a chest. Pretty easy in the Act 3 since it's likely the alternative relic is trash by that point. Green is the buffed elite. Again pretty easy in Act 3 because if you can't stomp it you're definitely not beating the heart anyway. Red is forgo a rest stop, which can be done anytime but I usually do it before the boss, especially if the boss is Hexaghost, because why both resting just to buff the boss?
> Technically the green key only calls for an extra challenge in facing a buffed elite instead of sacrificing, at least in the latest version.
I don't know if the latest version changed this, but the green key was definitely also a sacrifice -- if you look at the map, paths that allow you to get the green key have one less fire.
The heart was added in an update maybe 2 years ago. If you had the game longer than that, it wasn't unlocked automatically. Call that a bug or bad user experience.
Suggest you google how to unlock the heart and replay whatever needs to be done. It's really not intuitive if you already got all the characters and cards.
I can recommend watching some twitch streamer, like jorbs or baalorlord, a couple of times. They tend to do it during either ascension 20 or under some other restriction, but the lessons they teach will help a lot. In some cases they will expose combinations of cards that you might not have understood how well they work together, and in other cases they will show and explain cards that is good in the endgame but not very good before that. Very, very helpful.
After playing it for over a year, I had to watch a strategy video last month to realize that you might want to keep a small deck and not always pick a card every time you can. Didn't even occur to me.
I'm terrible at strategizing. To this day I can't figure out anyone other than the Ironclad with that power that grows his strength every turn.
I played a lot of M:tG at various times in my life, and games like Magic teach you a lot about these concepts. I would have never thought about it has I not had that experience! No shame in learning about the meta game!
Deck-building board games are probably better teachers there. Deck-thinning is a useful strategy in mtg but it’s absolutely crucial to increase the power level & consistency of a deck in games like ascension or clank.
It's also interesting that there are deck-building games which remove deck-thinning as as strategy e.g. in mystic vale your deck has a fixed number of cards, you can only fill advancement slots in your base deck[0], and you can't remove advancements. You can also acquire "permanents" (the vale cards).
[0] every card has 0~1 slots filled from the base deck, and position matter, you can't put two advancements in slot 2 of a card
Yeah, if nothing else, those games pretty much hammer into your head that all your starting cards are garbage and you should get rid of them, at the very least, at the soonest opportunity.
There's a couple of reasons you might want to hang on to them in Slay the Spire, depending on what events come up, but getting rid of them is pretty important in this game as well.
AIUI, the creators of Slay the Spire are very active in the Netrunner community, and play-tested it thoroughly with top Netrunner players. I haven't played it before but I think it plays closer to MtG than to deck builder board games.
It's a deckbuilding boardgame turned into a video game with a campaign. It is far, far closer to something like Star Realms, Thunderstone, Ascension, or Dominion (All deck building games that re-implemented Dominion's deck building ideas in different ways) than it is to MTG or Netrunner.
Virtually every idea in Slay the Spire other than the rogue-like addition previously existed in these deck building games.
Maybe I don't play enough deckbuilders, but it seems like there was a decent amount of new concepts in here that didn't directly come from other games. Abilities like Innate, Retain, Ethereal, cards getting upgraded for all future encounters every time they're played, costs of the card being reduced based on something specific happening that encounter, the stances of the Watcher character in general, and all the cards surrounding that idea, The orb idea for the Defect in general and all cards revolving around that idea, cards giving you bonuses if it kills a character, sometimes permanently, etc.
Even something as basic as all the cards costing energy to spend and you have only a certain amount of energy every turn (that can be increased by other methods), forcing you to prioritize what to play, as opposed to just playing everything you draw and only the order matters, like in most deckbuilders, changes the feel quite a bit.
They were able to take advantage of the video game concept (allowing cards to morph within an encounter and over the course of multiple encounters) and put it to good effect. Technically you could do that with a board game with stickers or something, but it's a lot more effective here, and only permanent for a run as opposed to forever.
That being said, they did take a ton of ideas from existing games. It just feels a lot more inventive than most deckbuilding games I've played, where most of them feel like some combination of Ascension and Dominion, with little to differentiate it.
The Retain keyword is similar to the effects from Treasury or Mandarin in Dominion, where you put some card(s) back on top of your deck during clean up if you meet the condition. Treasury especially is also a cantrip, so it doesn't use up space in your hand on the next turn.
Innate of course doesn't have a precedent in Dominion, since you don't keep your deck around between games. Ethereal could - there's no technical reason you couldn't print a Dominion card with text like, "During clean up, trash this card if you would discard it from your hand". I don't think there are any such cards, and I suspect they'd be generally worse in Dominion than Ethereal is in StS. Actually, having a deck of cards using in junking attacks with that text separate from ruins and curses might be interesting - curses that trash themselves, so you could hand them out a lot more aggressively without destabilizing the entire game.
Dominion's traveler cards were Donald X's attempt to make cards that change during the game, though of course Slay the Spire can push that idea much further.
Artifacts and States are a little bit like the Watcher's stances, but less deck-defining. You generally don't build a deck around having a particular artifact or state, but you can use them to get a little extra juice.
I agree, limiting your plays each turn (either with mana or dominion's limited actions) does a lot to change the texture of a game. The Ascension family of deckbuilders where you can just slam your whole hand down every turn tend to leave me underwhelmed.
I don't mean to rag on StS for being unoriginal, quite the opposite, it's great.
My point was that it was an evolution of deckbuilding games in the vein of Dominion rather than CCGs in the vein of MTG.
I was a little hyperbolic in saying that virtually all of its ideas come from existing physical games, but the core gameplay itself does not stray far from its inspirations.
Most of the gameplay elements you listed are present in mechanically identical forms in pre-existing games where possible (obviously the game being digital allows for mechanics that physical cards do not). Innate, Retain, Ethereal, upgraded cards (ex Pirate Ship), cost reduction and stances all exist just between Dominion and Thunderstone.
Not to be too pedantic but, Dominion also has an action economy, its just all actions (until later expansions) cost 1 resource. Another deckbuilding game, Mage Knight, also has an advanced resource-to-use-cards system and now that I reflect on it is probably the closest game to StS.
I'm not trying to argue that StS is un-inventive, they did a very good job of bringing rogue-like elements and persistent upgrades into the deckbuilding genre and a VERY, VERY good job at adding an addictive "one more run" feeling to the genre.
Fair enough. And it's been awhile since I played Dominion, I forgot about the ABC aspect of a Dominion turn and how that's kind of similar to energy points.
Now you're really making me want to check out Mage Knight. I was curious about it before, but maybe I should finally pick up a copy.
I hope you like playing games solo or have a group of friends just as excited about Mage Knight as you because it is a lot of rules with instructions that only make it harder to learn.
It's really not that much more complex than StS but you have to keep track or rules and perform game tasks (like refilling shops and playing out enemy encounters) that the video game would perform for you.
> It's really not that much more complex than StS but you have to keep track or rules and perform game tasks (like refilling shops and playing out enemy encounters) that the video game would perform for you.
That's really where digital games shine, tracking lots of small numbers, states, and actions which would be hell to do by hand. That's e.g. why hearthstone can have persistent damage while that'd be super annoying in MTG: you'd pretty much need a -0/-x die per creature which would get very old very fast.
They left when StS took off, and Netrunner has since been discontinued (although there's still an active fan-run Tournament scene with fan-published cards).
I am not very good at card/deck building games, but I assumed they wanted you to cull your deck when I noticed they make you use gold to do so. After realizing that, I started trying to keep a lean and focused deck. I only play casually, and I still haven't won, so my strategy still might not be good!
It can be used on curses but usually you want to get rid of either the base deck or cards which don’t gel.
If you add powerful cards the power of the deck increases, but if you remove weak cards it also increases, and the consistency increases further: if you have 15 cards 10 of which are starter every hand will average 4 starter and 1 new card, but if you have 10 cards 5 of which are starter it’ll be split 3/2 ou 2/3 on average.
The basic cards are pretty bad. You can't play your full hand most of the time anyway, so cutting them makes it more likely that you'll draw the better cards.
This is also why Snecko Eye (the relic that gives +2 draw every turn and randomizes card cost) is arguably the best boss relic. Even if you only get to play one 3 cost card, it's more likely to be the best card in your deck.
> This is also why Snecko Eye (the relic that gives +2 draw every turn and randomizes card cost) is arguably the best boss relic. Even if you only get to play one 3 cost card, it's more likely to be the best card in your deck.
I can't agree with this. Snecko Eye is very strong, but playing a 3-cost card will cripple you. It's strong because you draw a lot of cheap cards -- the lesson of Snecko Eye is that playing several cards is better than playing one card, even if the cards you got to play were chosen randomly.
A lean deck can be very good by the end of the game, but the problem is getting to that point. On the harder difficulties it's often better to focus on choosing strong cards that help you now rather than worrying too much much about thinness.
So it's a good strategy still, it's just that most of the time you can improve your deck faster by adding good cards than by removing bad ones.
Dropping your starter attack and defend cards after getting better alternatives is a good tip for new players, but the number one factor in winning in my experience is selecting the right card combos and synergies. Identify possible combos early and pick future cards to build on those strategies. If a card doesn't fit in your deck skip it. Sometimes you'll pick a combo card early and not come across any synergistic cards and lose or have to pivot strategies, it just happens. Having a smaller deck with only the cards you want will usually help though.
Aiming for synergies can be a trap, though. Sometimes the correct pick is a bad card that has anti-synergy with your current deck, because it solves an immediate problem. For example, at the start of a run you generally should pick any attack better than a strike (almost all of them) even over super powerful late-game scaling cards and try to kill the Elites to get relics and build power.
After the first few runs through the Spire, you'll generally know what you will be facing, and what you need to do is pick cards that cover your bases against those potential upcoming fights.
Indeed, best advice I can give for people having trouble is: do anything you need to do to survive Act 1 and kill 2-3 elites. After that you've got a lot more flexibility.
This depends on what you mean by success. If your aim is to end up steamrolling hard enemies with a broken deck and getting a large high score than synergies are worth going for. If your goal is to eek by with a win as often as possible there is too much variance in betting the farm on being offered certain cards in the future.
I would argue aiming for synergies is fundamentally wrong headed unless you're trying to moonshot with a lucky deck. There's way too much varience in expecting to get offered certain cards. That being said some synergies like silent poison probably worth gunning for... But that's mostly because poison cards are disperportionately strong.
I haven't played slay the spire, but I have been watching happy hobs no hit, no level-up dark souls trilogy runs and it might be interesting to see people trying to win "unwinnable" runs.
If u like hob, dark souls, rpgs... You will love slay the spire. It's so fun man and was one of my first roguelikes. If you get hooked on a roguelike you are in for hundreds of hours. EASY!
I've been thinking about that a lot, about these games that uses procedural generation. I've been playing a game on Xbox where "random" events occurs and you have to act accordingly. I've noticed the game wasn't fair or even winnable. That random event striking me that moment of the game, there's no way I could be prepared to deal with the situation (frostpunk, I think is the name of the game). When I released Qubes, my game, I made sure every level was winnable. So when people write me emails telling the level doesn't have a solution, I know for a fact that it has, the player just couldn't find yet. That's a nice article and I appreciate you sharing.
Really good story. I'll have to remember that one. I think I've kind of come to the same conclusion when designing board games but after a lot of trial and error :).
I met Ignacy once several years ago, at Gen Con. He's a really nice guy.
the trouble with that solution is that players are aware of the direction of what they'll get and can make action decisions appropriately. if I know one deck is the bad one, I'll avoid actions that result in draws to that deck, unless impossible to avoid.
I think Ignacy wanted there to be ambiguity there.
The game as originally described already forced exactly 5 draws, so that's not an issue. If you're forcing draws, you can force some good draws and some bad draws.
You can solve the "players know what's coming" problem easily, and some games do ("before you begin, take three disasters and two windfalls and shuffle them together"), but that adds a large logistics overhead that may not be worth it.
Jorbs (a popular STS streamer) has said something like:
If random shuffling of your deck can cause you to take 40 damage from Lagavaulin, and the game is still 100% winnable, then that 40 damage doesn’t matter at all. At which point the game is just not interesting, because most of the time you don’t take the 40 damage and the rest of the game is way too easy and low risk. With procedural generation, in order for most games to be challenging and interesting, you have to have games in which the game just stomps you.
This is an odd thing that made the PSP Ghost and Goblins not fun at the hardest level. They made enemy positions a bit too random, such that you couldn't build a good reflex moving through the game and the environments.
Which is to add that there is definitely a balance, and if you are going to have designed levels, I think it is more fun to have designed experiences. But I can see the case for both, and agree that I expect some games to stomp me by design to be part of the fun.
I'm assuming there is a guiding principal here, but I don't know it.
"100% winnable" just means that you have a chance of winning if you do the right things at the right times in relation to what's happening in the game. Nobody wants to play a game that they don't even have a chance of winning even if they play optimally. Replay incentive comes from knowing that if you improve yourself you'll do better, not from flipping coins and hoping for a favorable outcome.
> With procedural generation, in order for most games to be challenging and interesting, you have to have games in which the game just stomps you.
Only if you have a chance of winning or you're given the chance to detect the situation and run away from the fight until you do have a chance of winning. Otherwise it's just annoying.
> Nobody wants to play a game that they don't even have a chance of winning even if they play optimally.
I play such games regularly. I'm just choosing place to die, then I try to stay alive as long as possible. Sometimes I even win, just because of pure luck. Then I win more often.
No, I do not mean the quoted section. They claim to play games where there is no chance of winning, then go on to claim they eventually win them often.
> "100% winnable" just means that you have a chance of winning if you do the right things
This is not the definition of winnable that the OP (and having listened to him discuss this on-stream, I'm pretty sure Jorbs as well) is using.
From the article,
> “Can a perfect player win every game of Slay the Spire?” (alternatively, “Is there a strategy that achieves a 100% win rate in Slay the Spire?”)
"100% winnable" means that you are _guaranteed_ to win if you do the right things, not that you have a _chance_ to win under optimal play.
What you're saying is the same thing as saying "a coin toss is 100% winnable, because you have a chance of winning if you do the right things". I don't think that's a useful way of defining "100% winnable" in the context of this discussion.
(There is a loose conversational sense in which your interpretation could be valid; 100%=absolutely, "sure, 100% it's winnable" but these discussions are all using the technical narrow interpretation.)
I think you're misunderstanding and getting caught up on my use of the word chance. "Chance" here means that winning is not foreclosed. If you prefer, let's change it to "opportunity". A winnable playthrough is one that you are given the opportunity to win, one that you will win if your skill is sufficiently high. Not one that anyone will always win, but one that CAN be won by skill. That's the "chance" or "opportunity" part. A non-winnable playthrough is one where you cannot win. A chess game where your opponent starts with all their pieces and you only start with a bare king is non-winnable. You literally cannot win. Nobody wants to play that game unless maybe you invent a different objective. For this context, calling a game "100% winnable" is appropriate if every playthrough is winnable, i.e. for every playthrough winning is not foreclosed.
> What you're saying is the same thing as saying "a coin toss is 100% winnable, because you have a chance of winning if you do the right things
This is a misunderstanding and completely the opposite of what I'm saying because there's no such thing as "the right things" for coin tosses other than "toss coin" unless you're cheating, because the coin toss is meant to approximate a random selection. Only being able to win by chance is literally the opposite of being 100% winnable. It means you'll lose many tosses without doing anything wrong unless you cheat. That's why tossing a coin is not a fun game.
> "100% winnable" means that you are _guaranteed_ to win if you do the right things, not that you have a _chance_ to win under optimal play.
That's exactly what he was arguing too. Having a chance to win means that if you do exactly the right things, you can win. It doesn't means that you may win if you do exactly the right thing... that's just absurd saying may about something fixed.
You would enjoy a game where it is litteraly impossible to win? Whatever the moves you decide to do? You just lost once you got a card but still have to keep going... even though it's already decided that whatever move you choose to do, you will lose? I know I wouldn't enjoy it. There has to be a way, a specific choice of move, the rights one, that will make me win, otherwise, I'm losing my time.
> There has to be a way, a specific choice of move, the rights one, that will make me win, otherwise, I'm losing my time.
This is the opposite opinion from what the first comment in this thread was saying though right? They actually do want there to be some seeds that are just impossible.
I think so, and I think they ("Jorbs", not the commenter) said a dumb thing. They're saying that winnable is bad when it's actually good. Being able to win or at least not die[0] is a pretty big precondition for being fair, and people generally don't enjoy unfair games.
[0] - This means at atomic components of the game, not necessarily the whole game. There are infinite games that have no ultimate win condition where the objective is to just go as far as possible, but if the progression is fundamentally unfair (i.e. if components are randomly unwinnable) then it will be also fundamentally unsatisfying.
i think non-ascension (lowest difficulty) spire is pretty close to guaranteeing that you're able to win every run.
many people who play a lot of the game prefer to play at higher difficulties. for me the main reason is that at low difficulty a significant % of my time is spent going through the motions in a run that's already won, where almost any choice will still win and so my relevance as a player disappears, and my enjoyment of the game increases as the % of the time that my decisions appear relevant increases.
i've played ~5,000 hours of the game and thanks to high difficulties i'd guess at least 4,000 of those hours have been engaging gameplay, whereas if the game had never released higher difficulties maybe only 1,500/5,000 would have been interesting.
it sucks that some runs just end up being losses for pretty much any reasonable player, but at least even in losses there is usually engaging decision-making up until the end, since you can almost always find some out to play toward which will let you win with a bit of luck, and you can't tell ahead of time that you won't succeed at hitting that out. easy wins where you get enough stuff to beat the endboss in 10 minutes but still have 30+ minutes of gameplay left are less desirable in my experience.
it's a nice design goal for a game to just always be engaging, but very few games like this come close to achieving that goal. i'd say ascension 20 spire is closer than anything else i've ever played.
Hi! If you're the same jorbs, I'm sorry to cast aspersions on you based on a secondhand account. I agree with everything you just said.
I used to be a competitive gamer myself, and going through the motions where a win is guaranteed no matter what you do is definitely the other face of fundamental badness in game design. I fully agree that the most important thing in game design is for players to feel like what they do matters, and that means winning when they earn it, not winning when they don't earn it, and not making people slog through the motions when the outcome is decided. It's part of why we don't like playing against opponents who are too far below us in skill level. Trashing an enemy without effort just isn't fun for long.
Slay the Spire implements the gambler's fallacy in a number of places to even things out a bit. For example, the longer you go without finding a rare card or potion reward, the more likely you will find one, you never fight the same elite twice in a row, question mark rooms odds are adjusted to prevent seeing the same type of room too many times, a lot of enemies have semi-random attack patterns so that they don't just steamroll you with big attacks.
what are you calling semi-random? A lot of enemies have semi-hard coded patterns with certain parts of it having multiple options (I think this last part is as much to make the runic dome a bigger penalty as anything)
Well, for example, Gremlin Leader and Reptomancer don't have hardcoded patterns but they will not do the same attack twice in a row, that's what I mean by "semi-random". A lot of enemies are like that, they cannot use the same move more than x times in a row.
Some enemies have fully hardcoded patterns, or hardcoded patterns with multiple choices like you say, like the Heart. That's maybe less "semi-random" and more "hardcoded with a bit of randomness".
(I'm not sure which is more common between the two actually, I don't pay that close attention to attack patterns of normal enemies.)
Newer Tetris games do, by using a “bag” randomizer. The original is notorious for withholding I-pieces for long periods of time, so much so that videos of “classic Tetris” competitive play show a “drought” statistic on the screen after long enough.
A roguelike (berlin interpretation) can be made always winnable, without this issue, because their fundamental goal differs from games like Slay the Spire -- a roguelike's victory condition is to steal the object in the depths, and escape, and enemy battles are just an aspect of that task; its not key to the gameplay. Which is why you can have no-kill runs, and non-combat-optimal builds and be successful. It's also why you can have a monster that spawned way too early and can one-shot you while rewarding nothing on death, and you can decide not to fight it.
The 40 damage does matter, but you as a player decide as part of your risk management strategy whether its worth potentially being crippled, or killed, by such a monster. It also means a roguelike can introduce unbeatable creatures (at least at your current strength), or overly high risk-reward scenarios, and it doesn't mean the game is actually over.
Slay the Spire on the other hand is a battle simulator -- a series of fights until you win. You don't get an option in the matter, because there is no grander goal. Which means the same monster that is a choice in a roguelike, is simply game over in StS.
Jorbs is right about StS; he is not right about procedural generation. This is a direct result of the game's mechanics.
Keep in mind Jorbs' bread and butter are particular types of games, in particular deck based roguelites and strategy games in the vein of the new X-Com series. Also before that he came from a poker background so a lot of his analysis of games comes through those lenses.
I’m surprised to hear the Berlin interpretation brought up here. I thought for the most part that people had moved on.
Anyway, a lot of classic (Berlin) roguelikes put you in situations where it is impossible to escape combat. If the enemy you’re facing is faster than you and you have no tools of escape, combat is probably unavoidable.
So there are plenty of situations in a classic roguelike which are simply game over because you saw a monster (likely out of depth) you had no chance of beating. It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
> I’m surprised to hear the Berlin interpretation brought up here. I thought for the most part that people had moved on.
Its primary value for me is to differentiate between roguelikes and roguelites, where the latter term seems to have been given too much weight and controversy (roguelike communities trying to hold onto their name too strongly, as to become unseemly); referring to berlin interpretation seems like a less sensitive workaround. Roguelite is a dumb name anyways. The differentiation is still very useful though, because the roguelike genre is a very interesting one to be able to reference. (and so thoroughly distinct from roguelites, that the short distance in naming really isn't very appropriate -- similar to NoSQL vs SQL DBs)
> So there are plenty of situations in a classic roguelike which are simply game over because you saw a monster (likely out of depth) you had no chance of beating. It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
This is true; the general design trends towards combat encounters (roguelikes really just being the DnD combat & dungeoneering mechanics ripped out and pushed on computers, dropping most of the other elements), but that's more out of design choice than out of any necessity to keep the game "interesting".
That is, you can trivially conceive of interesting situations and mechanisms that are not directly combat-related, and you can also find examples of them implemented (like the orb-chasing of DCSS orb spiders, which force an awkward dance -- the challenge is from limiting positioning, and the damage is what gives it weight, but it is in no way unavoidable).
Rougelikes aren't interesting because there's always a chance you might end up in an unavoidable death -- they're interesting because there's always a chance you'll push yourself into an unavoidable death.
> It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
The core point of my argument is that you could force a minimal set of escape tools to the player, and the game would still be interesting. The games aren't actually dependent on having unavoidable threats, and being able to win each and every encounter, because the goals are different from StS. The threat can be there, without the unavoidable loss.
That the procedural generation may produce unwinnable games is not a requirement of procgen games -- it is a just an artifact of the implementation. But for StS and its kin, I think Jorbs is right -- potentially unwinnable games may be a requirement of that style of game.
a problem with this sort of analysis is that it describes what the game is like when you play it with perfect knowledge of the seed you're playing and as many resets as you want. that isn't how almost anybody plays slay the spire, so it doesn't have practical relevance to what playing the game is like.
it's cool as a theoretical exercise, but the answer ultimately doesn't matter for almost any human player of the game, and isn't a relevant thing for considering how the game should be designed and balanced. if we finally brute force every option from a difficult seed two years from now and find out that you actually can't beat it in any way that's not going to change how good or enjoyable the game is for people.
the more problematic thing is all the seeds where a reasonable player dies because "good" play dies. any seed you can eventually beat by retrying and taking bad cards/playing cards in bad orders until one of the reshuffles or card rewards lines up perfectly to get you past your hump is functionally problematic in all the ways that a truly unbeatable seed would be theoretically problematic. playing spire at a high level requires accepting that you run into those sorts of seeds sometimes.
> 1. “Can a perfect player win every game of Slay the Spire?” (alternatively, “Is there a strategy that achieves a 100% win rate in Slay the Spire?”)
> 2. “For every seed of Slay the Spire, is there a sequence of decisions that results in a win?”
Given that there are only 2^64 seeds, the answer to these questions may actually be the same. The optimal strategy for a perfect player would likely involve determining the state of the RNG (which only requires witnessing about 64 bits of RNG output) at which point you know exactly what will happen for the rest of the game.
I'm not very familiar with Slay the Spire so maybe I'm misunderstanding though.
It could be, but considering the developers made the conscious decision to use a static seed and show that seed to the player they are probably okay with that.
Yes but lots of games like this use static seeds that are shown to the player so that you can replay and share games. In particular so that viewers can play along with streamers.
I'd think that assuming knowledge of the seeds isn't an interesting question from the perspective of the player. To me the much more interesting question, and what I presume 1 is asking, would be "Assuming true randomness, is there are strategy that achieves 100% win rate in slay the spire." (True randomness meaning ignore pseudorandomness and assume that if the game makes something happens 5% of the time, it could do so from any valid starting state.)
Really what the question is asking is is there a strategy that for any theoretically valid game of slay the spire would, if followed, result in a win.
Given that the article did prove that there are theoretically unwinnable seeds with A20 silent, the answer is definitely not that am optimal player could win 100% of the time on all rule sets. Does change the question to "For a given StS rule set, what is the winrate the maximally efficient strategy would achieve."
It seems like you didn't read the actual article all the way through. They are specifically explaining a series of theoretically possible events that would make it so that it is literally impossible to win even if you have perfect knowledge of what the seed is going to give you. And then they look for a seed where this series of events happens, making it impossible to win.
The thing about "Sploosh Kaboom" is that the end state of the board, essentially a hash, is all you need to know to figure out the seed. That's thanks to the precomputed table embedded in the helper program.
Unlike reversing a hash, however, you actually want to crack the code using as little information as possible! Nobody wants to solve for the seed after they've already revealed the whole board, after all. The truly novel part of this particular tool is in how it leverages calculus to narrow down the seed pool by exploiting the RNG that produced the board seed. Each entry in the precomputed table is also time-indexed, since the RNG step count can be loosely approximated based upon the game's runtime. As the tool eliminates seeds from the pool, it's also gaining confidence in the exact cycle count the RNG was at when it produced the seed value for the board, which, in turn, tightens up the distribution curve of the remaining seeds.
Mario Maker 2 deliberately mimics early Mario in not having actual randomness. Anything that seems "random" is a result of fine player behaviour e.g. maybe you tapped that button for a frame longer or you stepped one pixel closer, the "random" result is different.
Nintendo doesn't want people to upload courses that are impossible, so it requires you to prove any course you want to upload is possible by beating it yourself first. Of course it can't check the course is fun this way, resulting in courses that are "garbage" in the parlance of popular streamers. It also can't prevent the course design depending upon knowledge the designer has but which is concealed from players, such as so-called "pick a pipe" where there are clearly apparently plausible routes forward, all but one of which leads to death, and the use of "dev exits"† where the creator hid some simple way to beat the level while an innocent player will try to do what may actually be impossible.
If there was true randomness then even a course intended just to be quite difficult might randomly sometimes be impossible due to unpredictable behaviour.
†Unlike "pick a pipe" a "dev exit" could be a necessary part of some very sophisticated courses where there intentionally is no ordinary way to win from specific situations without dying, real players are expected to die and continue playing successfully but the upload test does not allow this.
I'm curious if performing this kind of analysis, or even knowing its conclusions, makes a game less fun.
I enjoy games for several different reasons:
- The mental challenge of trying to win, and develop winning strategies in general. Knowing a perfect strategy, or knowing that some games are genuinely unwinnable, would diminish this. E.g. "Plague, Inc."
- Immersion in a compelling, fictitious, narrative, epic story. This pleasure is probably unaffected. E.g. the "Civilization" franchise or "Endless Space 2".
- The excitement of uncertainty. This could actually be increased by knowing that some random seeds make a game unwinnable, especially if one can't know that at the beginning of the game. E.g. "FTL".
It depends on the actual numbers for a given game. If it's possible to prove that 0.001% of games are unwinnable, I don't think that's going to diminish anyone's enjoyment of a game.
At 1% unwinnability, I think that almost nobody would notice the existence of unwinnable games merely by playing, but having the knowledge of the exact percentage of unwinnable games might start to diminish enjoyment.
Solitaire, where 20% of games are provably unwinnable from the start, absolutely suffers from this effect, because even when a player loses a winnable game they will be tempted to blame the game rather than themselves, fostering resentment.
If a game was provably unwinnable 80% of the time, then that would be either a damning indictment of its design or an indication that the value of that game is to be found in some aspect other than winning.
I suspect it also depends on how blatantly unwinnable the game is. Imagine you started up a game of Civilization, where you started on a single hex surrounded by impenetrable mountains. That may have a very low odds of happening, but if it does, you'd probably be fairly upset at the game.
An important consideration in that case though is that you'll have unambiguously discovered the "unwinnable" condition within the first two seconds of the game; with no emotional investment in that seed, it'll result more in a curious shrug rather than resentment.
(I say "unwinnable" because it is possible to win that game; in Civ 6 you would just need to build a military district on the far side of the mountain and then you could begin spawning units there, at which point you're now in actually in a very secure defensive position!)
Honestly, given how masochistically some gamers chase max-difficulty wins, I wouldn't be surprised if a game in which only 1% of seeds were winnable would be appealing to some.
But it would definitely diminish many gamers' enjoyment, as you say.
Dark Souls is an example where some players just aren't interested in the difficulty (even though the game is winnable), and some players relish it. Perhaps "win % for the median player" and other populations are more practically useful lenses for the discussion of general enjoyment.
I don't think many people would enjoy that at all.
The appeal of games like Dark Souls (and Slay the Spire!) at high difficulties arises from the knowledge that you _will_ win if you are clever and observant enough.
It's the learning process that keeps players coming back to these games, not the Game Over screen.
I definitely think it can make the games less fun. A few years ago, I wanted to get into modding Skyrim so I watched the tutorial videos from Bethesda. Watching the designers make maps showed that much less care than I thought actually goes into designing them. For example, if two pre-made dungeon tiles didn't fit together, they'd just put some rocks over the seams, and they had a lot of dungeon rooms already created and they'd just paste together in a different order to make a new dungeon. It shattered the illusion that the environment was handcrafted, and I couldn't stop seeing the shortcuts designers took once I knew how they were made. I stopped playing shortly after, and will wait until ES:VI is heavily discounted before I buy it.
Well, for that phenomena the opposite could happen just as well. Exploring and picking apart a game to find out that it actually is handcrafted and made with care and precision with some unique techniques used to mask the possible issues would likely result in a higher appreciation for the technical aspects of the game. Now, would that make a person love a game they'd otherwise dislike? (For plot, mechanics or other reasons) No, they'd still dislike it but they'd certainly admire the craftsmanship.
For the case presented in the article, I'd actually feel much less compelled to play StS if I knew there were genuinely unwinnable seeds. It takes away the player's agency to a higher degree than it should and having a run where nothing you do will get you a win means you can't learn from your mistakes or enjoy a win - two of the main purposes. As for having fun with a run, well, being on a doomed ride isn't that fun either, at least compared to a regular StS run.
> For example, if two pre-made dungeon tiles didn't fit together, they'd just put some rocks over the seams, and they had a lot of dungeon rooms already created and they'd just paste together in a different order to make a new dungeon.
Interestingly, this is also common in typical U.S. house construction. Moulding, caulk, and paint cover a multitude of sins.
I think another factor not mentioned by my sibling comments is what the player wants out of the game.
The things I enjoy in a game very much align with yours, but I have several close friends who find solving the optimisation problem a game represents to be the most rewarding part of the gameplay experience.
'Min-maxing' stats and deeply understanding and exploiting game logic are all very enjoyable parts of the game to them. Immersion and storytelling are often secondary to those puzzles.
To players with those interests and goals, uncertainty is a puzzle to solve and the ultimate goal of the mental challenge may very well be to find the perfect strategy.
Some people enjoy breaking a game wide open, whether through degenerate strategies, hacking, careful character design, or something else. That's where they get their fun.
Once they do that, they might be done with the game, but it was probably the best way to maximise their fun (most of us eventually move on from a game for various reasons anyway).
I'm guessing your win rate estimations are super low. For a game with this much depth, the gap between the best players and optimal play is absolutely massive.
For intuition around this, look at how much chess has evolved over the past 400 years. Look at the history for speedruns or the ABC challenge for super mario 64, or look at speedrun histories for other games like golden eye or super mario odessy.
Just based on the fact that this is a relatively obscure game with a lot of complex mechanics, I would estimate that with optimal play, the win rate should be over 99% for all characters. This probably requires the development of a large number of new techniques and play styles, and probably has many surprising optimizations that would seem ground-breaking today.
this game has a bunch of randomness which stops us from knowing where we're going to end up if we make decisions. it'd be like if super mario 64 had a bunch of jumps in it where there was a 1/10 chance you died on screen transition and you couldn't tell where you were meant to jump ahead of time. if you replay or check the seed ahead of time you can find out and stop dying to those things, but in practical play there's no obscure mechanic that stops you from dying because you only had the ability to prepare properly for two elite fights but ended up fighting the third, or because you were presented two options for synergistic scaling for the endgame fights but didn't get offered any more of the one you chose.
they can all get above 50% and watcher can definitely get way higher than that, but 99% is unthinkable because not enough of the ways you can die are solvable with the information available in a run.
love slay, this is really interesting and fun work i would guess. Game is hard but after you play a lot, it becomes less hard because you know a lot. A20 still fucks ya hard, not sure i really enjoy it that much. Climbed to A20 with all 4 chars. Recommend the game!
141 comments
[ 4.3 ms ] story [ 337 ms ] threadI've even watched a few people's playthroughs and going over which cards are better than others and combo well together, and my win rate is still terrible. Just lost again this morning in Act 2 with another Defect run.
Meanwhile this guy is saying the best players are winning 99% of the time on Standard, and nearly the same on Ascension level 15? Wow.
By the way, for anyone thinking this game isn't really for you, you might want to give it a chance anyway (watch a video at least). I don't usually care for Rogue-likes, and I've been pretty tired of Deckbuilders, but I got into this game hard, pretty much immediately, after putting off trying it for forever and finally deciding to give it a chance.
One thing that really makes it awesome for me is that you know basically exactly what the enemies intend to do and how much damage they're going to do to you, so you can puzzle out how to prepare for it, should you kill this person who's going to do 18 damage to you, or reduce their strength, or build up your block to absorb it, or take the hit this round so you can go after something else, etc. If you've ever played Into The Breach, it feels a lot like that, but with RPG battles instead of a Tactical map.
I've learned a lot from Jorbs - a twitch streamer. It's surprising how much statistics and planning can go into any given fight. It's possible to puzzle out how much of the fight will go ahead of time, with knowledge of one's deck and the enemies moves.
I mean, that is perhaps part of your problem. There aren't any cards I've not picked at least once and made good use of on ironclad even at A20.
The game really is more about building a deck based upon what comes up and constantly reassessing what is actually working in your current deck.
Choosing "good" cards based on their overall strength out of context is essentially a brute force approach to winning.
And there are plenty of discussions online for pretty much every card on how strong or weak people think the cards are overall (like is it worth picking it up if it's offered to you late in a run because the payoff might not be worth it, or if you see you're getting a specific boss that act, etc), and most of the cards have been buffed or nerfed multiple times since the game has come out, so clearly the developers have been convinced that some strong may end up stronger than others on average too (and are trying to keep things pretty even as much as possible, I assume).
With that said, as the Defect, Defragment and Capacitor are nearly always a no-brainer. Capacitor not so much in Act1, because nothing lasts long enough for it to pay off. And if you've got anything that adds orb slots, Consume becomes really powerful (even without, you can do 10 damage a turn from two orb slots instead of 9 from three, which isn't nothing).
Loop builds are very strong as well, but trickier to get right because you need to pay close attention to your front orb. Finally focus+frost can easily give you so much block most enemies can't cut through. (Watch out for Reptomancer, though, she doesn't respect anyone without DPS.)
But I was a little too weak on my defense that go (had some frost generation, but not enough), and I made a dumb mistake attacking the Masked Bandits instead of losing all my gold, who did 50 damage to me before I killed them, bringing me down to 18 health, then I couldn't get to a campfire until I fought 3 more regular enemies, and the next enemy brought me down 6 more, then the second enemy after that got me down the other 12. Oops.
Good luck with Watcher. The biggest advice here is: play it safe until you have multiple cards to get out of wrath. Watcher is super-powerful and _very_ easy to mess up.
I'm a bit tapped out at A8, but it's amazing how the base game goes from nailbitingly impossible to seeming pretty straightforward. (The mechanics contribute to this: the easier a battle is, the more upgrades you do and the more likely you are to grab an elite relic.)
For the ironclad, you might take a cleave, or a whirlwind, or a spot weakness, or an inflame in act 1. The first two give you AOE, and the last two will help scale damage if you find a limit later on.
StS lets you not take a card when you are offered one, and often this is the best play, particularly as you go later into the game and have a deck that is stacked with powerful cards.
Fundamentally having more cards is bad, because it increases variance in what you can draw. Obviously taking a good card can be worth that cost, but often players fall into the trap of assuming the best card of the options given is worth taking, when really the question should be "does this card beat the average card in my deck, or does it dilute the power of the deck", because you are reducing the chance to draw one of the other cards in your deck.
Obviously there are edge cases: sometimes something may reduce the power of your deck right now, but introduces the chance for later synergy, which may be worth the cost—especially early on, but in general it is worth considering not taking the card.
The ideal deck is always the minimum number of cards that allow you to pull off your win state, everything else is just stopping you getting there (this is also why cards that allow you to draw cheaply are good, and one of the reasons why removing cards from your deck is powerful).
If you're in Act 2 and headed for an elite, you have to think, "can my deck beat book of stabbing? gremlin leader? taskmaster?".
If right now your deck sucks against the book, and you're offered a disarm or caltrops, it might be worth picking it even if it's doesn't generally make your deck stronger. Cause otherwise you might get stabbed to death.
On the flip side, Demon Form is a very strong card, but it's practically a curse for most of Act 1, because it's too costly and slow. So sometimes you might skip even a strong card if it doesn't help you right now.
I'd say that's probably the most important "ah-ha" moment where my winrate really started to significantly improve. That and knowing when to min/max # of combats while pathfinding.
At high difficulty levels you need to pick a card (and possibly buy cards at shops) through most of Acts 1 and 2 because you won't be able to reach the boss without a bunch of damage commons.
No, not necessarily. But my brain wants to pick one anyway.
StS has a number of enemies that shuffle bad cards into your deck, so this dillution works both ways. On higher difficulties more of these cards get shuffled into your deck. Trying a "thin" strategy with no countermeasures in such a fight (high damage output to end the fight fast?) can get you railroaded. You also start out with 10 starters that you want to dillute because virtually any card is better than them. Cards with "draw" and "exhaust" written on them also let you workaround the problem of drawing starter or negative effect cards. Picking up one or two versatile 2-cost cards early (attack or block card) can give you a good way to use your energy even if you draw mostly starters or negative effect cards while avoiding excessive deck dillution later on.
StS does offer you worse cards at the beginning of the game, and limited ways to remove or upgrade these cards, so picking up too many cards can burden you later on.
Really to be good at StS deckbuilding mostly requires you to know which enemies will come up when and building your deck sufficiently to survive them without shooting yourself in the foot long term too much. Jorbs is usually focused on building his deck to beat the hardest fight in the game while absorbing the minimum deadweight to survive until that point. Watching a stream can give you a rough idea of what a good "pace" for adding new cards into your deck is.
Jorbs can reliably reach the heart with a much more challenging deck than you or I can. He's also playing a game mode that imposes challenges that don't exist for new players. Just shoving a lot of good cards in your deck might be pretty good advice for new players up until about ascension ten.
...Slay the Spire has nothing in common with roguelikes?
It's always intrigued me seeing other people pick up the game and master it much more quickly than I did. There's some people that get their A20 heart kill just 100 hours into the game.
I really enjoy the feeling of slowly but steadily gaining knowledge and intuition about the game - in the 500 hours of play there's been no shortage ah-ha moments throughout. Truly one of the best $/entertainment value games I've ever purchased.
You're possibly bringing baggage from other deck builders, mentioning combos and better cards is an example of this.
Generally you want to judge cards for what they do for you right now (in the next few fights) to maximise rewards and build long term value. Try not to build into "just in case" combos or synergies or you bleed HP before it shows up. The exception is if you are confident your deck can handle the hallway fights then you might take picks for the boss, and if you can handle that then you can start building for the future.
Apart from that, general pickup rules are: Act 1 is front-loaded damage, Act 2 is AOE damage with some damage scaling for the boss, Act 3 is scaling offense and defense (without damaging your other factors too much). You usually want to be plugging holes rather than building on your strengths but there are always exceptions.
One example: A single dark orb can be enough to kill Champ on Act 2 as defect. Any more is excessive.
I have fun though, so that’s nice.
The heart itself, well it's definitely challenging, but it only works for very specific builds. I went to some effort to beat it once on each character and basically have no intention of doing so again.
If you want a meme deck, A1 is the sweet-spot.
I don't know if the latest version changed this, but the green key was definitely also a sacrifice -- if you look at the map, paths that allow you to get the green key have one less fire.
Suggest you google how to unlock the heart and replay whatever needs to be done. It's really not intuitive if you already got all the characters and cards.
Similar to how every (properly designed) sudoku is solvable without guessing, the fun isn't in having solved it, its figuring out how to solve it.
I'm terrible at strategizing. To this day I can't figure out anyone other than the Ironclad with that power that grows his strength every turn.
It's also interesting that there are deck-building games which remove deck-thinning as as strategy e.g. in mystic vale your deck has a fixed number of cards, you can only fill advancement slots in your base deck[0], and you can't remove advancements. You can also acquire "permanents" (the vale cards).
[0] every card has 0~1 slots filled from the base deck, and position matter, you can't put two advancements in slot 2 of a card
There's a couple of reasons you might want to hang on to them in Slay the Spire, depending on what events come up, but getting rid of them is pretty important in this game as well.
Even something as basic as all the cards costing energy to spend and you have only a certain amount of energy every turn (that can be increased by other methods), forcing you to prioritize what to play, as opposed to just playing everything you draw and only the order matters, like in most deckbuilders, changes the feel quite a bit.
They were able to take advantage of the video game concept (allowing cards to morph within an encounter and over the course of multiple encounters) and put it to good effect. Technically you could do that with a board game with stickers or something, but it's a lot more effective here, and only permanent for a run as opposed to forever.
That being said, they did take a ton of ideas from existing games. It just feels a lot more inventive than most deckbuilding games I've played, where most of them feel like some combination of Ascension and Dominion, with little to differentiate it.
Innate of course doesn't have a precedent in Dominion, since you don't keep your deck around between games. Ethereal could - there's no technical reason you couldn't print a Dominion card with text like, "During clean up, trash this card if you would discard it from your hand". I don't think there are any such cards, and I suspect they'd be generally worse in Dominion than Ethereal is in StS. Actually, having a deck of cards using in junking attacks with that text separate from ruins and curses might be interesting - curses that trash themselves, so you could hand them out a lot more aggressively without destabilizing the entire game.
Dominion's traveler cards were Donald X's attempt to make cards that change during the game, though of course Slay the Spire can push that idea much further.
Artifacts and States are a little bit like the Watcher's stances, but less deck-defining. You generally don't build a deck around having a particular artifact or state, but you can use them to get a little extra juice.
I agree, limiting your plays each turn (either with mana or dominion's limited actions) does a lot to change the texture of a game. The Ascension family of deckbuilders where you can just slam your whole hand down every turn tend to leave me underwhelmed.
I don't mean to rag on StS for being unoriginal, quite the opposite, it's great.
I was a little hyperbolic in saying that virtually all of its ideas come from existing physical games, but the core gameplay itself does not stray far from its inspirations. Most of the gameplay elements you listed are present in mechanically identical forms in pre-existing games where possible (obviously the game being digital allows for mechanics that physical cards do not). Innate, Retain, Ethereal, upgraded cards (ex Pirate Ship), cost reduction and stances all exist just between Dominion and Thunderstone.
Not to be too pedantic but, Dominion also has an action economy, its just all actions (until later expansions) cost 1 resource. Another deckbuilding game, Mage Knight, also has an advanced resource-to-use-cards system and now that I reflect on it is probably the closest game to StS.
I'm not trying to argue that StS is un-inventive, they did a very good job of bringing rogue-like elements and persistent upgrades into the deckbuilding genre and a VERY, VERY good job at adding an addictive "one more run" feeling to the genre.
Now you're really making me want to check out Mage Knight. I was curious about it before, but maybe I should finally pick up a copy.
I hope you like playing games solo or have a group of friends just as excited about Mage Knight as you because it is a lot of rules with instructions that only make it harder to learn.
It's really not that much more complex than StS but you have to keep track or rules and perform game tasks (like refilling shops and playing out enemy encounters) that the video game would perform for you.
That's really where digital games shine, tracking lots of small numbers, states, and actions which would be hell to do by hand. That's e.g. why hearthstone can have persistent damage while that'd be super annoying in MTG: you'd pretty much need a -0/-x die per creature which would get very old very fast.
They left when StS took off, and Netrunner has since been discontinued (although there's still an active fan-run Tournament scene with fan-published cards).
If you add powerful cards the power of the deck increases, but if you remove weak cards it also increases, and the consistency increases further: if you have 15 cards 10 of which are starter every hand will average 4 starter and 1 new card, but if you have 10 cards 5 of which are starter it’ll be split 3/2 ou 2/3 on average.
This is also why Snecko Eye (the relic that gives +2 draw every turn and randomizes card cost) is arguably the best boss relic. Even if you only get to play one 3 cost card, it's more likely to be the best card in your deck.
I can't agree with this. Snecko Eye is very strong, but playing a 3-cost card will cripple you. It's strong because you draw a lot of cheap cards -- the lesson of Snecko Eye is that playing several cards is better than playing one card, even if the cards you got to play were chosen randomly.
So it's a good strategy still, it's just that most of the time you can improve your deck faster by adding good cards than by removing bad ones.
After the first few runs through the Spire, you'll generally know what you will be facing, and what you need to do is pick cards that cover your bases against those potential upcoming fights.
All of my most successful winning decks have been built off of synergies though.
To be fair though, it's not as computable (I guess?) as FreeCell (just 1 unwinnable game?) or Minesweeper.
each are board game designers, and Vlaada was play testing one of Ignacy's designs:
https://boardgamesthattellstories.wordpress.com/2013/02/10/t...
I met Ignacy once several years ago, at Gen Con. He's a really nice guy.
That's not at all rare in other games.
I think Ignacy wanted there to be ambiguity there.
You can solve the "players know what's coming" problem easily, and some games do ("before you begin, take three disasters and two windfalls and shuffle them together"), but that adds a large logistics overhead that may not be worth it.
The update shows you many moves you've made and how many moves (optimal solution) you have to make on that level.
https://play.google.com/store/apps/details?id=com.victorribe...
The update should be out really soon.
If random shuffling of your deck can cause you to take 40 damage from Lagavaulin, and the game is still 100% winnable, then that 40 damage doesn’t matter at all. At which point the game is just not interesting, because most of the time you don’t take the 40 damage and the rest of the game is way too easy and low risk. With procedural generation, in order for most games to be challenging and interesting, you have to have games in which the game just stomps you.
Which is to add that there is definitely a balance, and if you are going to have designed levels, I think it is more fun to have designed experiences. But I can see the case for both, and agree that I expect some games to stomp me by design to be part of the fun.
I'm assuming there is a guiding principal here, but I don't know it.
> With procedural generation, in order for most games to be challenging and interesting, you have to have games in which the game just stomps you.
Only if you have a chance of winning or you're given the chance to detect the situation and run away from the fight until you do have a chance of winning. Otherwise it's just annoying.
I play such games regularly. I'm just choosing place to die, then I try to stay alive as long as possible. Sometimes I even win, just because of pure luck. Then I win more often.
This is not the definition of winnable that the OP (and having listened to him discuss this on-stream, I'm pretty sure Jorbs as well) is using.
From the article,
> “Can a perfect player win every game of Slay the Spire?” (alternatively, “Is there a strategy that achieves a 100% win rate in Slay the Spire?”)
"100% winnable" means that you are _guaranteed_ to win if you do the right things, not that you have a _chance_ to win under optimal play.
What you're saying is the same thing as saying "a coin toss is 100% winnable, because you have a chance of winning if you do the right things". I don't think that's a useful way of defining "100% winnable" in the context of this discussion.
(There is a loose conversational sense in which your interpretation could be valid; 100%=absolutely, "sure, 100% it's winnable" but these discussions are all using the technical narrow interpretation.)
> What you're saying is the same thing as saying "a coin toss is 100% winnable, because you have a chance of winning if you do the right things
This is a misunderstanding and completely the opposite of what I'm saying because there's no such thing as "the right things" for coin tosses other than "toss coin" unless you're cheating, because the coin toss is meant to approximate a random selection. Only being able to win by chance is literally the opposite of being 100% winnable. It means you'll lose many tosses without doing anything wrong unless you cheat. That's why tossing a coin is not a fun game.
That's exactly what he was arguing too. Having a chance to win means that if you do exactly the right things, you can win. It doesn't means that you may win if you do exactly the right thing... that's just absurd saying may about something fixed.
You would enjoy a game where it is litteraly impossible to win? Whatever the moves you decide to do? You just lost once you got a card but still have to keep going... even though it's already decided that whatever move you choose to do, you will lose? I know I wouldn't enjoy it. There has to be a way, a specific choice of move, the rights one, that will make me win, otherwise, I'm losing my time.
This is the opposite opinion from what the first comment in this thread was saying though right? They actually do want there to be some seeds that are just impossible.
[0] - This means at atomic components of the game, not necessarily the whole game. There are infinite games that have no ultimate win condition where the objective is to just go as far as possible, but if the progression is fundamentally unfair (i.e. if components are randomly unwinnable) then it will be also fundamentally unsatisfying.
many people who play a lot of the game prefer to play at higher difficulties. for me the main reason is that at low difficulty a significant % of my time is spent going through the motions in a run that's already won, where almost any choice will still win and so my relevance as a player disappears, and my enjoyment of the game increases as the % of the time that my decisions appear relevant increases.
i've played ~5,000 hours of the game and thanks to high difficulties i'd guess at least 4,000 of those hours have been engaging gameplay, whereas if the game had never released higher difficulties maybe only 1,500/5,000 would have been interesting.
it sucks that some runs just end up being losses for pretty much any reasonable player, but at least even in losses there is usually engaging decision-making up until the end, since you can almost always find some out to play toward which will let you win with a bit of luck, and you can't tell ahead of time that you won't succeed at hitting that out. easy wins where you get enough stuff to beat the endboss in 10 minutes but still have 30+ minutes of gameplay left are less desirable in my experience.
it's a nice design goal for a game to just always be engaging, but very few games like this come close to achieving that goal. i'd say ascension 20 spire is closer than anything else i've ever played.
I used to be a competitive gamer myself, and going through the motions where a win is guaranteed no matter what you do is definitely the other face of fundamental badness in game design. I fully agree that the most important thing in game design is for players to feel like what they do matters, and that means winning when they earn it, not winning when they don't earn it, and not making people slog through the motions when the outcome is decided. It's part of why we don't like playing against opponents who are too far below us in skill level. Trashing an enemy without effort just isn't fun for long.
Some enemies have fully hardcoded patterns, or hardcoded patterns with multiple choices like you say, like the Heart. That's maybe less "semi-random" and more "hardcoded with a bit of randomness".
(I'm not sure which is more common between the two actually, I don't pay that close attention to attack patterns of normal enemies.)
I should play StS again, but when I get the craving for a deck builder these days I'm playing Monster Train.
The 40 damage does matter, but you as a player decide as part of your risk management strategy whether its worth potentially being crippled, or killed, by such a monster. It also means a roguelike can introduce unbeatable creatures (at least at your current strength), or overly high risk-reward scenarios, and it doesn't mean the game is actually over.
Slay the Spire on the other hand is a battle simulator -- a series of fights until you win. You don't get an option in the matter, because there is no grander goal. Which means the same monster that is a choice in a roguelike, is simply game over in StS.
Jorbs is right about StS; he is not right about procedural generation. This is a direct result of the game's mechanics.
Anyway, a lot of classic (Berlin) roguelikes put you in situations where it is impossible to escape combat. If the enemy you’re facing is faster than you and you have no tools of escape, combat is probably unavoidable.
So there are plenty of situations in a classic roguelike which are simply game over because you saw a monster (likely out of depth) you had no chance of beating. It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
Its primary value for me is to differentiate between roguelikes and roguelites, where the latter term seems to have been given too much weight and controversy (roguelike communities trying to hold onto their name too strongly, as to become unseemly); referring to berlin interpretation seems like a less sensitive workaround. Roguelite is a dumb name anyways. The differentiation is still very useful though, because the roguelike genre is a very interesting one to be able to reference. (and so thoroughly distinct from roguelites, that the short distance in naming really isn't very appropriate -- similar to NoSQL vs SQL DBs)
> So there are plenty of situations in a classic roguelike which are simply game over because you saw a monster (likely out of depth) you had no chance of beating. It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
This is true; the general design trends towards combat encounters (roguelikes really just being the DnD combat & dungeoneering mechanics ripped out and pushed on computers, dropping most of the other elements), but that's more out of design choice than out of any necessity to keep the game "interesting".
That is, you can trivially conceive of interesting situations and mechanisms that are not directly combat-related, and you can also find examples of them implemented (like the orb-chasing of DCSS orb spiders, which force an awkward dance -- the challenge is from limiting positioning, and the damage is what gives it weight, but it is in no way unavoidable).
Rougelikes aren't interesting because there's always a chance you might end up in an unavoidable death -- they're interesting because there's always a chance you'll push yourself into an unavoidable death.
> It may be unlikely if the game throws lots of escape tools at you, but there’s always the chance you don’t get any of them on a given seed.
The core point of my argument is that you could force a minimal set of escape tools to the player, and the game would still be interesting. The games aren't actually dependent on having unavoidable threats, and being able to win each and every encounter, because the goals are different from StS. The threat can be there, without the unavoidable loss.
That the procedural generation may produce unwinnable games is not a requirement of procgen games -- it is a just an artifact of the implementation. But for StS and its kin, I think Jorbs is right -- potentially unwinnable games may be a requirement of that style of game.
it's cool as a theoretical exercise, but the answer ultimately doesn't matter for almost any human player of the game, and isn't a relevant thing for considering how the game should be designed and balanced. if we finally brute force every option from a difficult seed two years from now and find out that you actually can't beat it in any way that's not going to change how good or enjoyable the game is for people.
the more problematic thing is all the seeds where a reasonable player dies because "good" play dies. any seed you can eventually beat by retrying and taking bad cards/playing cards in bad orders until one of the reshuffles or card rewards lines up perfectly to get you past your hump is functionally problematic in all the ways that a truly unbeatable seed would be theoretically problematic. playing spire at a high level requires accepting that you run into those sorts of seeds sometimes.
> 2. “For every seed of Slay the Spire, is there a sequence of decisions that results in a win?”
Given that there are only 2^64 seeds, the answer to these questions may actually be the same. The optimal strategy for a perfect player would likely involve determining the state of the RNG (which only requires witnessing about 64 bits of RNG output) at which point you know exactly what will happen for the rest of the game.
I'm not very familiar with Slay the Spire so maybe I'm misunderstanding though.
Really what the question is asking is is there a strategy that for any theoretically valid game of slay the spire would, if followed, result in a win.
Given that the article did prove that there are theoretically unwinnable seeds with A20 silent, the answer is definitely not that am optimal player could win 100% of the time on all rule sets. Does change the question to "For a given StS rule set, what is the winrate the maximally efficient strategy would achieve."
That approach was in fact used for "Sploosh Kaboom", a RNG minigame in the Windwaker: https://www.youtube.com/watch?v=1hs451PfFzQ
The thing about "Sploosh Kaboom" is that the end state of the board, essentially a hash, is all you need to know to figure out the seed. That's thanks to the precomputed table embedded in the helper program.
Unlike reversing a hash, however, you actually want to crack the code using as little information as possible! Nobody wants to solve for the seed after they've already revealed the whole board, after all. The truly novel part of this particular tool is in how it leverages calculus to narrow down the seed pool by exploiting the RNG that produced the board seed. Each entry in the precomputed table is also time-indexed, since the RNG step count can be loosely approximated based upon the game's runtime. As the tool eliminates seeds from the pool, it's also gaining confidence in the exact cycle count the RNG was at when it produced the seed value for the board, which, in turn, tightens up the distribution curve of the remaining seeds.
Nintendo doesn't want people to upload courses that are impossible, so it requires you to prove any course you want to upload is possible by beating it yourself first. Of course it can't check the course is fun this way, resulting in courses that are "garbage" in the parlance of popular streamers. It also can't prevent the course design depending upon knowledge the designer has but which is concealed from players, such as so-called "pick a pipe" where there are clearly apparently plausible routes forward, all but one of which leads to death, and the use of "dev exits"† where the creator hid some simple way to beat the level while an innocent player will try to do what may actually be impossible.
If there was true randomness then even a course intended just to be quite difficult might randomly sometimes be impossible due to unpredictable behaviour.
†Unlike "pick a pipe" a "dev exit" could be a necessary part of some very sophisticated courses where there intentionally is no ordinary way to win from specific situations without dying, real players are expected to die and continue playing successfully but the upload test does not allow this.
Relevant from yesterday: https://robertheaton.com/preventing-impossible-game-levels-u...
https://forgottenarbiter.github.io/Correlated-Randomness/
I enjoy games for several different reasons:
- The mental challenge of trying to win, and develop winning strategies in general. Knowing a perfect strategy, or knowing that some games are genuinely unwinnable, would diminish this. E.g. "Plague, Inc."
- Immersion in a compelling, fictitious, narrative, epic story. This pleasure is probably unaffected. E.g. the "Civilization" franchise or "Endless Space 2".
- The excitement of uncertainty. This could actually be increased by knowing that some random seeds make a game unwinnable, especially if one can't know that at the beginning of the game. E.g. "FTL".
At 1% unwinnability, I think that almost nobody would notice the existence of unwinnable games merely by playing, but having the knowledge of the exact percentage of unwinnable games might start to diminish enjoyment.
Solitaire, where 20% of games are provably unwinnable from the start, absolutely suffers from this effect, because even when a player loses a winnable game they will be tempted to blame the game rather than themselves, fostering resentment.
If a game was provably unwinnable 80% of the time, then that would be either a damning indictment of its design or an indication that the value of that game is to be found in some aspect other than winning.
(I say "unwinnable" because it is possible to win that game; in Civ 6 you would just need to build a military district on the far side of the mountain and then you could begin spawning units there, at which point you're now in actually in a very secure defensive position!)
But it would definitely diminish many gamers' enjoyment, as you say.
Dark Souls is an example where some players just aren't interested in the difficulty (even though the game is winnable), and some players relish it. Perhaps "win % for the median player" and other populations are more practically useful lenses for the discussion of general enjoyment.
The appeal of games like Dark Souls (and Slay the Spire!) at high difficulties arises from the knowledge that you _will_ win if you are clever and observant enough.
It's the learning process that keeps players coming back to these games, not the Game Over screen.
I don't know about that, but I think that the analysis can be (for some people) more interesting/fun than the game.
For the case presented in the article, I'd actually feel much less compelled to play StS if I knew there were genuinely unwinnable seeds. It takes away the player's agency to a higher degree than it should and having a run where nothing you do will get you a win means you can't learn from your mistakes or enjoy a win - two of the main purposes. As for having fun with a run, well, being on a doomed ride isn't that fun either, at least compared to a regular StS run.
Interestingly, this is also common in typical U.S. house construction. Moulding, caulk, and paint cover a multitude of sins.
The things I enjoy in a game very much align with yours, but I have several close friends who find solving the optimisation problem a game represents to be the most rewarding part of the gameplay experience.
'Min-maxing' stats and deeply understanding and exploiting game logic are all very enjoyable parts of the game to them. Immersion and storytelling are often secondary to those puzzles.
To players with those interests and goals, uncertainty is a puzzle to solve and the ultimate goal of the mental challenge may very well be to find the perfect strategy.
Some people enjoy breaking a game wide open, whether through degenerate strategies, hacking, careful character design, or something else. That's where they get their fun.
Once they do that, they might be done with the game, but it was probably the best way to maximise their fun (most of us eventually move on from a game for various reasons anyway).
For intuition around this, look at how much chess has evolved over the past 400 years. Look at the history for speedruns or the ABC challenge for super mario 64, or look at speedrun histories for other games like golden eye or super mario odessy.
Just based on the fact that this is a relatively obscure game with a lot of complex mechanics, I would estimate that with optimal play, the win rate should be over 99% for all characters. This probably requires the development of a large number of new techniques and play styles, and probably has many surprising optimizations that would seem ground-breaking today.
they can all get above 50% and watcher can definitely get way higher than that, but 99% is unthinkable because not enough of the ways you can die are solvable with the information available in a run.
https://github.com/ForgottenArbiter/CommunicationMod
Seems relevant. In life you can execute a 'perfect strategy' and still lose, so it goes in some games too.