itch.io

Some more statistics!

The rarest type of Evil combination is Pooka with 2 non-minion minions. I have only recorded one of those, and I secretly wonder if it was a mistype and that game doesn’t exist?

I also haven’t played a Lilis game which has both a minion and a puppeteer. I wonder if that combination is impossible?

---

Once you get into regular ascensions, on average there are 2.52 evils per round yielding 885 points.

Beginners average about 970 points per 7 round 0 loss ascension, which works out to 12 extra points per round (85 in 7 rounds) which means on average each evil is killed with 0.5 cards still face down. In short, beginners are flipping all the cards before trying to kill their first evil.

An expert playing with the intention of 0 loss (no random strikes!) but doesn’t mind taking chances averages roughly 1300 points per 7 round 0 loss ascension. They score 59 extra points per round ascension (416 in 7 rounds), which means on average each evil is killed while 2.3 cards are still face down. An expert is often taking their first evil out with only half or a few more of the cards flipped. (The standard deviation for a 7 round ascension is about 197 points, so the evil composition matters a lot!)

I confirmed the minimum score per ascension is 350 points: 7 perfect solo Pooka games.

And by restart scumming for only 4 Evil Puppet games and losing 3 on purpose after killing 3 evils, I was able to score 3090 points in one ascension. I only tried once, so with some luck, I think a score of 3250+ is possible.

---

My revised Evil ratios after tripling the amount of games is

Minion_____ 27% 86

Twin Minion 14% 45

Witch______ 24% 75

Shaman_____ 16% 52

Poisoner___ 17% 55

Chancellor_ 22% 69

Puppeteer__ 17% 53

Puppet_____ 16% 49

Baa________ 36% 115

Pooka______ 31% 97

Lilis______ 33% 105

If only itch’s search was a bit better at finding similar games!

Hard, great progression, great tutorialization.

I had to replay the first few levels over a period of a few days to really nail down the mechanics to avoid frustration, but it was worth it.

This is the first game since Gate’s Give Up the Ghost that I was fairly, yet severely challenged to figure out all the new mechanics.

Noa, the user who made this puzzle set also made a one-room brain teaser for the confounding calendar called “>” that is also worth checking out.

I played 100 games in standard Ascension 100+ and recorded how many of each demon and minion appears.

Minion: 24%

Twin Minion: 13%

Witch: 21%

Shaman: 12%

Poisoner: 18%

Chancellor: 35%

Puppeteer: 17%

Puppet: 16%

Baa: 33%

Pooka: 33%

Lilis: 34%

Chancellor and Witch are over-represented while Shaman is under represented.

I also broke down how often each minion appears with a particular demon

---

Minion with Baa: 4/33 12%

Twin Minion with Baa: 4/33 12%

Witch with Baa: 10/33 30%

Shaman with Baa: 4/33 12%

Poisoner with Baa: 4/33 12%

Chancellor with Baa: 12/33 36%

Puppeteer with Baa: 5/33 15%

Puppet with Baa: 5/33 15%

---

Minion with Pooka: 16/33 48%

Twin Minion with Pooka: 4/33 12%

Witch with Pooka: 5/33 15%

Shaman with Pooka: 4/33 12%

Poisoner with Pooka: 2/33 6%

Chancellor with Pooka: 8/33 24%

Puppeteer with Pooka: 1/33 3%

Puppet with Pooka: 1/33 3%

---

Minion with Lilis: 4/34 12%

Twin Minion with Lilis: 5/34 15%

Witch with Lilis: 6/34 18%

Shaman with Lilis: 4/34 12%

Poisoner with Lilis: 12/34 35%

Chancellor with Lilis: 15/34 44%

Puppeteer with Lilis: 11/34 32%

Puppet with Lilis: 10/34 29%

---

Pooka, Lilis, and Baa were evenly balanced at 1/3rd each.

Chancellor, Minion, and to a lesser degree Witch were more likely to be a “+1”: appear in games which already had another minion in them, contributing to their frequency. Shaman and Puppeteer/Puppet were more likely than other minions to be the only minion it their round.

Lilis overall had the most minions at 67 in 34 games, followed by Baa at 48 in 33 games, and Pooka had the fewest minions at 41 in 33 games.

If you are looking for a challenge, try to get a no loss, 7 game Ascension where [Final Score – (350 ∗ Ascension level)] is greater than 1600 or less than 750. Or get a 3 loss game with greater than 1800. (For Ascension > 4)

My bests are 1700 for 0 loss, 1820 on a 2 loss, and 690 and 700 for lowest ascension score. (Without scumming, surprisingly! If you are willing to scum you can definitely beat these!)

Another way to play is to reset ascension level and try to maximize or minimize the first three Ascensions.

With no random stabbing,

Ascension 1 1310 (0 loss max), 910 (0 loss min), and 1530 (3 loss)

Ascension 2 1960 (0 loss max), 1400 (0 loss min), and 2520 (3 loss)

Ascension 3 2750 (0 loss max), 1900 (0 loss min), haven’t done a 3 loss run here

Dreamers are the weakest villager role with the lowest usability in the game, second to knight, maybe.

A truthful Dreamer will tell you an evil’s true identity. If you targeted a villager, the Dreamer will tell you a random evil which is pretty much a waste of them.

If lying, the dreamer will always lie about an evil’s true identity, and tell you a random evil if you picked a villager.

I am on Ascension 130 and have a 100 village saved streak on my main account, which means I’ve played a thousand rounds or more of demon bluff. From that perspective, I will say that dreamers are very rarely useful.

A lot of the time, you mostly want them to identify as truthful or liars, and you can’t do that for sure until you have killed an evil or there is a wretch in play. That makes them a worse sort of confessor.

Sometimes with a certain board configuration, they have a percent chance of validating another character by naming someone as a demon that’s already dead when you know the dreamer is not corrupt.

They are most useful in Lilis games where their ability recharges at night, so you might be able to validate them and then get a use out of them.

If the dreamer is good in a one pooka game, they will always call themselves the pooka no matter if they are lying or corrupt. If a pooka, the dreamer will name an evil that does not exist.

I made a small collection of other “one room” games I liked. https://itch.io/c/6231571/one-room

Late Nine also has “Key Lime Slime” which is technically two rooms but basically a one room game.

So G-Fire-Flip, not G-Flower-Flip. https://rumkin.com/tools/cipher/rot13/

First, get all three stars you can normally get with plain Fire Flip which leaves you with the upper right star. Try not to accidentally fire stuff in the process.

By this point in the game, you have probably figured out there is no way to get to that upper right sector without Rot13(Trggvat ba n jnyy) There is no way to do that with fire or flip. Which leaves you with your other two powers. Rot13(Gung zrnaf lbh arrq gb qb fbzrguvat gubebhtuyl hapbairagvbany gung bayl tbng znxrf cbffvoyr. Nf tbng, lbh pna chfu abgrf, ebpxf, fgnef, naq cbjre beof. Gurer vf n jnl gb znxr guvf jbex, whfg fgneg ehyvat bhg guvatf gung qba’g jbex hagvy jung erznvaf zhfg or gur fbyhgvba, nf penml nf vg znl or.)

The solution takes a quite a few maneuvers to set up, so you might start to feel silly, but it takes less time to set up than G flower flip IMO. Also, you have a lot of options, so long as you don’t shove things into corners or fire by mistake.

The drunk and an evil can both share a disguise as a villager not in play. And hunter #2 is clearly lying because, if they were truthful, all four evils must be in spots #5, #6, and #7, and #8… but drunk is occupying one of those spots, so hunter can’t be telling the truth. Hunter #2 needs to go. Chances are, Hunter #2 is Lilis, which means you can flip the rest of the cards and still win because the health subtraction penalty stops working once Lilis is dead.

Here’s an example of a drunk and demon “double claim”:

https://img.itch.zone/aW1nLzI0NzYyMzE0LnBuZw==/original/8Dk%2Fzt.png

Mine was Rot13 (onobbakkkcnqqvaterzbirgurk)

Interesting enough another (cevzngr) related word solved too: (obabobkkkcnqqvaterzbirgurk)

Ascension 100!

My average Ascension Score - (Ascension lvl × 350) was 1032, with a high water mark of 1820 thanks to about 5 Puppet or Lilis rounds in one ascension with meant blind kills were super viable. (Edit: my average score the last 20 rounds was 1163, median 1155, so definitely improved over the course of things!)

Most dangerous card: Dreamer. The number one cause of lost games. Dreamers can’t be used effectively before an evil kill and using them before a kill makes them very tough to validate later. A dreamer on the board is like you don’t have a card at all. Doppel dreamer and Shaman dreamer games are over-represented in my lost game notes. Now that I am this far out perspective-wise, Dreamer is still one the card needs to be fixed to do more so it has pre-first-kill utility, or simply ditched completely. Secondary nod to the plague doctor for being a round wrecker at times.

Most dangerous evil: Chancellor, replacing a villager for an outcast weakens the board more than the puppeteer or corruptors. A nod to Witch and Baa as well.

‘Pooka party’ hands with just Pooka and 4/5 villagers were a delight when they came up - the frequency was about right too. BTW. The knitter should say “There is only 1 Evil” like scout instead of “Evils are not adjacent to each other” in single evil games (taking into account the Wretch).

I confirmed some edge mechanics I had been wondering about: Lilis kills a random evil, not preferentially the wretch, if only evils are facedown on nightfall. (I have screenshot proof). Stabbing a facedown real knight while a witch is blocking does not flip it. Witnesses can point to evils when lying.

I was a little sad game 6/7 of Ascension 66 and game 7/7 of Ascension 77 weren’t something silly (although I did get a seven card seventh game on ascension 77)

A66 6/7 could be a hand with 5 evils and one wretch in play.

A77 7/7 could be a chancellor, shaman, minion game where they all disguise as bombardiers with a real bombardier outcast and shaman dupes a jester and the final char is a judge. #1 Bombardier, #2 Bombardier, #3 Jester, #4 Jester, #5 Bombardier, #6 Bombardier, #7 Judge (Guaranteed winnable!)

The puppet didn’t lie. The puppet was a scout. The Scout’s testimony gives the distance between the specified Evil and the closest other Evil to it. The Scout’s mechanic is different from the hunter who is distance from self to nearest evil.

The scout gave testimony about the witch, who truly was one card away from #9, a demon.

(Caveat that the wretch can mess this calculus up, but not in this round)

Can you post a full screenshot? You seem to have one more evil to kill because…

#1: You do not have “the village is safe” popup yet which means there must be one more evil left to kill.

#2 Your villager count is 5, minus one for the puppet. So you have 4 real villagers. You have 6 apparent villagers on the board that I see, one of which is a doppelganger. That’s one too many unless you have a second outcast.

#9 medium appears to be evil, pointing at #2 who is a doppelganger. In that configuration 7, 8, 3, and 2 would all be telling the truth.

Not challenging, but the Stanley Parable-like joy of circumventing the narrator more than makes up for it. Another birb special!

Alright, exorcists, here’s a weird one for you. Lilis night attacked a real knight twice in a row, so I’ve got a full living 10 card village with a triplet of scouts and a triplet of mediums. Who has to go?

5 is the standard penalty for most villagers and outcasts. That’s almost all the characters.

2 when killing a drunk because the drunk acts like an evil and sometimes you can’t avoid spiking him trying to figure out who is actually evil.

9 for a corrupted knight

6 for a drunk disguised as a knight (special case)

∞ for a real bombardier

And Lilis subtracts 2 health every fourth flip which is why Lilis rounds can look a little weird healthwise.

This is from a run where I got 2 points lower than my max, but it should help.

There are no major errors in the gameplay logic right now (but the lack of tutorials is a problem, imho). People here are happy to sort what’s going on if you screenshot your game results.

The wretch is usually the culprit though for new players, fittingly enough.

Here’s an example of someone else’s problem with the wretch: https://itch.io/post/15061433

Puppet was a slayer, and a puppet slayer can still slay evils. They will slay their puppeteer or themselves. Puppets can also tattle on themselves as fortune teller, jester, bishop, empress, and witness.

The only time the puppet sabotages the ability of the card it is impersonating is puppet alchemist (always “I cured 0 corruptions”), puppet Baker (always “I was a baker” and no propagation), and puppet Knight (can be killed). Puppet confessor will be dizzy.

https://demonbluff.wiki.gg/wiki/Baker

I needed the wiki for the baker.

Basically the baker is a disease that propagates to one unflipped villager card every time a new baker is flipped until it hits a corrupted one. The corrupted one will transform, lie about what it used to be, but end the chain.

So it’s possible for a medium to say something like #7 is a real Knight, and then on flip #7 is a baker that says “I was a Knight” (or if corrupted lie and say “I was an Oracle.”)

“I am the original baker” is always a good, uncorrupted card, but may be a doppel.

Also alchemists who were bakered will still do their alchemy because alchemists are a round start ability, not an on flip ability.

Sometimes when there is a puppet, self-validation is the best strategy.

The only problem with puppet currently is a puppet knight does technically lie by saying “I can’t die.” That should be sneak changed on right click when knight is puppet as a freebie for anyone observant, like cabbage dreamer.

Puppet killing puppeteer ought to unlock a backstabbing-themed skin, imo.

Kinda like Lost in Lampyrid Fog, the puzzles are great and very responsive, the atmosphere is charming, but the platforming is painful and more annoying than fun.

In the upper right, there is a list of the characters that might appear in the round and you can right click to learn how their abilities work. Some like Wretch (always good so never kill, but seen as a truthful but evil minion by truthful characters) can cause unexpected behavior.

The rest is logic.

If you have Poisoner (+1 Corrupted villager adjacent to Evil) and Baa (Don’t know which outcast will be in play), and a choice between Drunk and Wretch for the one outcast and no wretch appears on flip, then you expect 4 total lies - one drunk, one corrupt, and 2 evils. At least two liars have to be adjacent. Then you plot out possible scenarios that would make the board consistent with 4 liars (if X is a lying evil, this guy must also be a liar).

Then you kill someone suspicious (unless you kill a bombardier or you are in a Lilis round, you get a “free” kill) and find out if you logicked wisely. If you were wrong, there’s a good chance you can use the new information (dead cards reveal if they were truthful or corrupt) to pivot into the correct solution, especially if you didn’t tap all the character abilities.

Also you can use dead character abilities, even if they were evil, which helps.

For some reason my saves don’t save for this game anymore, but I will have to give it a try!

I think I just had the best round of demon bluff from a pure logic standpoint. This feels like the purest single game you could play to test if you know the rules.

8 would have been the perfect opening strike because drunk and demon must be double claiming the same role, and 3 and 5 would both have to lie if 4 was evil and only one of them could be poisoned.

There’s no backtracking away from the bear. You can get the solution with the abilities you have, but it’s a bit of a difficulty curve jump straight from yellow rabbit.

If anyone wants my opinion about the "smoothest" route learning mechanics wise, I suggest...

Orange Fox ▶ Yellow Rabbit ▶ Purple Squirrel ▶ Red Bear ▶ Green Deer ▶ Cyan Otter ▶ Blue Owl

My first run, I wound up missing orange fox and going straight to yellow squirrel to red bear to cyan otter. I would not recommend that.

By elimination, yep. The other rabbits at least tried to use their powers, even if they weren’t skilled enough succeed. Beast master is so scared they pretend someone else has their power.

This is a good one! Maybe par of slightly less than architect (but def less than oracle) because of the specificity but it doesn’t combine with any existing character role which keeps the balance.

Maybe the dancer should name one odd and one even Evil if you wanted to increase the potency a bit. (Puppet is Odd, Puppeteer is Even) (No even evils and no odd evils would be pretty helpful if that happened, which might be too OP.)

Here’s that hundred village streak validated on one account. (Sorry it took me while, roadtrip!) I discovered a new nitpick. If you save enough villages, the total number in the all saved villages tab on the left goes too close to the edge for those wide numbers.

So what is fury status, exactly? Does it do the max possible damage for an item? (I see it gets rid of items, but does it do anything else?)

I should have screenshotted to confirm. I don’t recall now.

Great Update! There are punchy gloves now on the apparent character exclusives. Oh no, garbage items! You threw off my Deck duplicator groove! (Still bought two!)

And I finally saw Doc again! I forgot how tricky he is, especially with that one full heal that you have to freeze or lose a universe to.

Is his primal first aid supposed to do anything if his HP is not at 1? He still healed even though I had Exclusion Principal to make sure he didn’t have a 1 universe.

Canceling is not really using an ability. It would be a bit silly to include a loophole like that.

A victory isn’t legit until it is all on one account, and I bet someone else can verify before I do! I’ll work harder for that screenshot proof, and good luck on the Q1 launch schedule!

My endless streak, split between my login and unlogged-in cookies is now 100 villages in a row, so I am now comfortable having even more opinions.

---

More Bugs Observed

Oracle mode bug. The flip-order tracker should not reset after executing a character.

Text off screen.

Pooka Frequency Check? I like Pooka in my endless games, but I seem to get the true fluffiest evil 50% more than Baa. Goat guy seems under-represented.

---

Playstyles

All minion games. I don’t mind the demons, but it would be nice to regularly have all-minion games in hard mode to mix it up a little. The minions have a greater range of disguise and slightly more chaotic abilities on average so they are quite challenging.

---

Wording Fixes

Small fix to Medium. The medium should say “#3 is actually a doppelganger” instead of “#3 is a real doppelganger” to match the better phrasing of the drunk.

Evil Hover text. The Evil hover text has a few grammatical oddities. Maybe change it to “Character alignment. Eliminate all Evils to save the village. Evils Lie and Disguise, unless specific rules say otherwise.”
(Current text is “Character Alignment. Eliminate them. Most often Lie and Disguise. Eliminate all Evil to save the village.”) IMO, the plural noun “Evils” as a shorthand for “evil characters” is trivial to understand, used consistently in other parts of Demon Bluff, and fits the gamelike context.

Alchemist Activation. Alchemist’s card should mention that his ability triggers at the start of a match when he is still facedown after evils and outcasts have acted because he is a violation of the typical rule that most abilities trigger on flip.

Chancellor and Puppet. It would be helpful to note somewhere that Outcast and Evil numbers in the upper right Villager/Outcast/Minion/Demon tracker don’t reflect the totals in game because of these characters’ abilities.

Minor plurals. It’s the littlest of nitpicks, but 1 minion and 0 demons in the upper left tracker would be correct.

---

Outcast-related deck bloat

Hard games generally have one too many possible outcasts in the play deck. This makes Baa completely meaningless as a demon.

For instance, a 6 Villager, 1 Outcast, 2 Minions, 1 Demon game with Chancellor, Minion, and Pooka has 15 cards in the play deck (9 villagers (6 real + 3 potential disguises), 4 outcasts, 3 evils) for 10 cards on the field. The play deck should only have 3 outcasts to account for the one chancellor will add, and a possible outcast disguise.

In a 9 card game with Witch and Pooka and 2 outcasts, 4 outcasts show up in the play deck. In this type of game, I would expect play decks that alternately have 2 or 3 outcasts, varying the uncertainty.

Fixing the outcast deck bloat would make hard games marginally easier on average because of uncertainty reduction, but it benefits players who like more logical approaches as opposed to stab-and-find-outers (like me!). You could have a mix of both, too.

Baker. The baker is a benefit & drawback flawed character like an outcast, but they can’t be listed as one because their ability wouldn’t work. That said, internally, I really think the baker should be treated as an outcast for frequency purposes. Remove one outcast from the game if a baker is going to show up.

---

Even more characters that could be in the game

Gardener

(Villager, Good) (↶) Pick 1 character: Learn how many evils were still facedown after the selected card was first flipped over. [NO NIGHT REFRESH! BUT MAYBE NOCTURNAL REFRESH.] Hint: Oracle Mode numbers cards by the order they were originally flipped in.

Lazybones

(Outcast, Good) Disguises as a Villager or Outcast with an (↶) activating ability. Trying to use their ability reveals Lazybones’ true identity.

(↶) Select two/three cards from the pop-up deck of eligible roles. Lazybones randomly disguises as one of those characters and can use their ability. Hint: Truthful villagers always see Lazybones as Lazybones and an Outcast.

[The ability is Lazybones running off to the neighboring village to borrow someone else to do their job! And the eligible roles might be the villagers in the current deck, or cards that are less powerful. Lazybones keeps the yellow outcast card border when it borrows a new role as a reminder of what they are.]

[A card to throw off careful player calculus, but give them an opportunity to try to pick a custom villager in recompense.]

Nocturnal

(Outcast, Good†) Game Start: I disguise as a Good Villager currently in play. I am initially Good and Truthful.

If a Good character is Executed, I turn into a Lying Evil Demon that must be Executed! (Note: Evils who slay Goods do not turn Nocturnal.)

Hint: When Nocturnal turns Evil, cards with abilities that refresh during a Night Cycle instantly renew. Previous statements and actions Nocturnal made while Good are still seen as truths, but any new ones after a turn will be Lies.

“A sinister and restless bloodline, reawakened by innocent blood spilt.”

[A card to share the game with a drunk or doppelganger, best friend of the Chancellor. Might also need the ability to reactivate any one living card’s power if there are not enough night refresh roles on the board. I think disguising as a card not in play like a usual demon is a bit too tricky when there is corruption everywhere.]

Ghul

(??? Evil) If there are no dead characters, lose 1 health on every other card flip, starting with Flip 1. (e.g. 1,3,5,7,9)

[Not sure if this would be better or worse in play than the other one health lost per flip baddie with reduced penalty I suggested below. The goal is to encourage strategic flipping and early risk-taking. Should not show up with Lilis or any other health subtractors. (Although I think of Lilis as a slightly stronger Witch, like Pooka is a stronger Poisoner, and doesn’t need extra bleed damage because losing a good villager selectively is its own penalty.)]

6 damage.

My understanding is that, internally, the math goes like this:

5 dmg [standard penalty for innocent execution] 
+ 
4 dmg [corrupted knight penalty] 
+ 
-3 dmg [drunk anti-penalty]
=
6 net penalty

The wretch catches a lot of people because they are used to thinking of the wretch as a safe, good character because no evil ever disguises as them.

But remember the wretch is seen as a random evil minion (it might change within rounds) by everyone but a truthful dreamer, who sees them as a cabbage. That means Lilis won’t kill a wretch on a first night because she thinks they are an ally, for instance.

In this case, the oracle thought the wretch was a poisoner, which immediately validated the oracle as truthful (albeit kinda useless, lol).

It also means you have to be more careful with hunter’s testimony because “poisoner” could be the real poisoner or the wretch.

[there should be a healing item for low universe runs]

“Miracle” heals (if I recall) five HP in the universe of your choosing, which IMO is generally more useful the fewer universes you have. It occurs to me that I could have the shop be more likely to stock items that it thinks are more useful to your current run, which might be necessary given how “bloated” the item selection has become.

The thing about those single universe target items is they take up a whole deck slot. I feel like there a sweetspot that is “Once per floor, pay 4 energy for 20 healing divided evenly across all universes you have, rounded down.” So if you have a four universe run, it’s super nifty, but a waste of a card in many universe situations.

---

Agree with your logic on wave collapse. It helps with the fun.

And also agreed on adjusting the shop inventory to match your current situation!

The Baker chains so long as the new Baker’s original role is not corrupted.

If not corrupted, the first Baker flipped (or Doppel baker) truthfully says “I am the original Baker”. It then targets a still facedown villager card to become a baker, who when flipped will try to baker another villager, creating a chain. The chain breaks if a baker tries to target a corrupted character, who will turn into a baker but lie about what their original role was and will not baker any more facedown cards.

A puppet who was a baker will claim “I was a baker.”

The strat in the game you just played would have been to shank the confessor, since there are no outcast disguisers to worry about (so bomby is real). If the confessor was corrupted because the poisoner was nearby (only 1 corruption in this game and they must be next to the poisoner), the poisoner must be the #5 baker claiming medium because 7 was telling the truth. And Baa would be the dreamer because demons can’t be the same role as a villager and the other bakers are clear.

Of course, shanking the confessor would have revealed Baa, meaning 7 was lying and you could validate the dreamer on Baa to reveal the dreamer was also lying. Then it becomes a fifty-fifty choice about who to shank (you still have 10 health, so you win).

Sometimes the best strategy is to figure out potential solutions and then spend the free kill on someone suspicious you think you would break the conundrum.

This is also why Baker games are tricky because the jester and medium might have given you more and better information.

I enjoyed the main quest and the three lantern hunt, but not even the path of observation let me escape the jarring gameplay switch, so I decided to call it a game with the standard ending.

Also the GUI navigation is a bit clunky. Not having a backscroll between clues, having to hit the inventory every time to light a candle rather than just having a hotkey for each item, or letting you shortcut by just X to interact once you have the useful item. And the clue to the green location should have probably been a collectable, so you wouldn’t have to memorize how it was done.

The cannon puzzle was excellent. I might have labelled each of the candles in the lighthouse puzzle with the symbol of their lighthouse to reduce backtracking later.