How the AI Works in Pokemon Mystery Dungeon Rescue Team Red and Blue

Despite the words’ lack in the title, the two videos linked here, both made by Some Body, are all about roguelike behavior, and likely have implications for Chunsoft’s Mystery Dungeon engine generally, from which the Rescue Games derive.

In terms of depth, this post is rated 4 out of 5: highly detailed information for obsessed fans and game designers.

The first (28m):

And, the second (44m), it goes further into the weeds and is longer:

So, here’s a tl;dw overview of the first video. Despite the length, this is really only a brief summary! Some Body got their information by reverse engineering the games’ code, so it should be considered authoritative.

PMD has three times of actions, moving, attacking and using items. First they try to use an item–if there is no item to use, or the situation isn’t appropriate, or there’s a random component and they choose not to, they fall through to attacking. If there’s no one appropriate to attack, they fall through to moving or wandering. If they’re not pursuing a target and aren’t wandering, they wait in place.

Awake Pokemon try to reach a target: team members try to reach the leader (you)*, enemies try to reach a party member of yours. If they are following someone, they try to reach the target by default moving diagonally before moving orthogonally. This is good to know, and an effective strategy, since it’s harder to escape a cardinal-adjacent Pokemon than a diagonal-adjacent one. If a Pokemon has a target in sight but can’t move towards towards it, it doesn’t move.

(* Note: for teammates, this assumes the “Let’s Go Together” tactic is in effect. Generally, tactics settings are covered in the second video.)

No Pokemon can move towards a target they can’t see. Sight in Blue & Red Rescue Team is two spaces around them, or throughout a lit room they are in plus one space into corridors. Of course, invisible targets can’t be seen, even if they’re nearby. Note, a quirk of the Mystery Dungeon series generally: when standing in the first space of a corridor, you can only see slightly into the room, but everyone in the room can see you. While your default sight range in darkness is two spaces in the PMD1 games, instead of MD’s standard 1 space, you’re still a bit blind when moving into rooms. Notably, that two space distance around you is a square, so in corridors with bends in them you get a bit extra sight distance.

Now comes the interesting part (to people who are as obsessed with roguelikes as I am): what happens if a Pokemon loses sight of its target? In PMD1, it considers the last four locations the Pokemon was in, and tries to go to the one it was visible in most recently. Note in bent corridors, it becomes harder for a character to lose its target.

If the target is four turns outside of the follower’s sight, it has lost track of it, and the follower begins wandering randomly. This can happen if the Pokemon has never had a target (none has come into sight), or the target or follower teleports, the target moves over terrain the follower cannot cross, or the target moves away when the follower is occupied, or, due to the variety of events that can happen in the Mystery Dungeons, other ways.

Followers without targets wander randomly. When they spot a target, they cease moving randomly and pursue it. But if still wandering, in rooms, they pick a random exit, go to it and go down the corridor. In a corridor, they follow it until they reach a room (then entering it), or they reach an intersection. At an intersection, we see an interesting behavior: PMD1 occured before Chunsoft switched over to making wanderering monsters pick random directions at corridor intersections! In later Mystery Dungeon games, including later Pokemon Mystery Dungeons, wandering monsters go straight in intersections if they can. This is behavior that can be relied upon, but not in PMD1.

Outmatched Pokemon can decide to flee, essentially, moving away from their targets instead of towards. In rooms, they pick the exit furthermost from their pursuer, unless they moves them towards that pursuer; then they just try to get away as best they can, likely remaining in the room. A quirk of this: sometimes a fleeing monster breaks for an an exit that is more distant from the target, but not away from at attacker, giving it a free hit. The circumstances around this are complicated: the explanation begins at 7:16 in the first video.

For attacking, Pokemon have up to four moves, and a normal “attack.” This generic attack is not part of the main Pokemon game series. It was present in the first two PMD games, but after that became less effective. In the fourth and fifth PMD games, the normal attack only does five points of damage, and in the Switch remake of Rescue Team, it does no damage at all; it’s only a tool for passing time. But we’re still in the realm of PMD1, where “normal attacks” are not only useful but frequently used, because they don’t consume any PP.

Attacks are chosen based on a weighted average of all the usable moves. Each move has its own weight value; the normal attack weight’s varies according to the number of other moves available.

Ranged attacks are an interesting case. If a Pokemon has a ranged attack, and an enemy that can be attacked at a distance, it triggers the attack routine, where it picks a move from those available, but then only actually performs the move if the attack can reach its target. This can result in an attacker passing up opportunities to attack while an opponent approaches it. Out of fairness, room-range attack moves are only used by the AI when adjacent to an enemy.

Items have a bunch of minutiae associated with their use by the AI, but a lot of it is pretty ordinary. A few highlights: teammates can throw held negative status equipment at enemies, wild Pokemon start using items at Level 16, and there is only one Orb that wild Pokemon can use, and teammates can’t use it: the Rollcall Orb, for them, summons a number of other wild Pokemon into adjacency with them.

Caves of Qud Intro Videos

Caves of Qud, after over a decade of development, finally reached a 1.0 release and has, for now at least, become the toast of the more-enlightened gaming internet. Of course there will people who will look at its time-based graphics and look down on it, and go back to their games of Call of World of Fortnight Among Us Craft Duty League. But if you’re here, then there’s a good chance that you get what’s special about roguelikes. And not just roguelikes, but classic roguelikes: heavily randomized, turn- and tile-based, and challenging. Hence, Caves of Qud.

Honestly, the roguelike scene is so large now that no one person could reasonably be expected to keep track of all of it. But there is no need to; others hold aloft that particular torch. Here’s a couple of videos, then, on getting started in Caves of Qud.

Publisher Kitfox Games (who also publish the Steam release of Dwarf Fortress) sponsored a video with “Getting Started” right in its title. Here it is (18 minutes):

It contains information on the different modes, the best starting location for beginners (Joppa), basic controls, navigating around the starting town, how to get around the world map (reminiscent of Alphaman!), how to spend kill points, how to read things, how to examine Artifacts, how to experiment with things (even if it gets you killed sometimes), how to steal things, performing the water ritual, and some combat tips.

Another, slightly longer at 24 minutes, intro video is by Rogue Rat:

It covers ranged weapons, the town of Joppa, Truekin, what to do when you get lost, some different skills to learn, gaining levels from giving books to a specific NPC, using its Crawl-type Autoexplore feature and other topics. Rogue Rat did a longer, more basic, intro video (34m) last year that went over many of the same topics as the first video here.

My Talk on Mystery Dungeon for Roguelike Celebration

They haven’t broken the talks apart into individual videos yet, but in the meantime you can see my presentation overview of the 31, give or take a couple depending on precise definition, games in the Mystery Dungeon series here, queued up to the proper starting point in the 8+ hour video. The talk portion is about half an hour long (with a couple of interruptions due to the router I was on being a bit flaky).

Here’s an embed, but note that WordPress doesn’t seem to accept the link for embedding with the time code linking directly to my talk, so you’ll have to skip ahead yourself to 6:10:18 to get to it. Or you could watch some of the other very interesting talks on the way there! Either way!

Roguelike Celebration 2024

Starting this Saturday at noon US Eastern time (9 AM Pacific, 5 PM Greenwich, 7 PM CEST) is Roguelike Celebration 2024! I’m presenting half an hour on the Mystery Dungeon games this year, at 3:15 PM Pacific/6:15 Eastern/11:15 Greenwich/1:15 AM Sunday CEST. Whew, the roundness of the Earth makes it difficult to express times!

It’s being held entirely online again this year, and offers a fun social space to explore that’s kind of like a MUD! Roguelike Celebration’s schedule is here, and you can get tickets for $30 for the whole event here. They usually set aside some free tickets for people who can’t afford the fee, although you might have to check around to find them.

Roguelike Celebration is nominally about roguelikes and procedural generation, but I think it’s interesting from a wide variety of gaming perspectives, and every year I find several talks that are incredibly interesting. Past years have offered presentations from people who worked on games as diverse as Kingdom of Loathing and Blaseball. Here are the talks being offered this year:

Saturday

  • Harry Solomons: Trampling on Ghosts: Hauntology and Permadeath
  • Cezar Capacle: Enhancing Narrative Through Randomness and Complications
  • Max Bottega: Keeping Art Direction interesting in a procedurally generated world
  • Stanley W. Baxton: Bringing Real-World Occultism into Your Games Without Accidentally Being Racist
  • Jeff Emtman and Martin Austwick: Neutrinowatch – the podcast that plays itself
  • Nic: Braided Narratives: Or How I Learned to Stop Worrying and Love Linear Stories
  • Pandamander: “Out of Book,” The Psychology of Why Roguelikes Keep Us Playing
  • ?: Inverse Terrain Solver
  • Adrian: Probably Impossible – NecroDancer’s network code
  • John Harris: A Trip Through The Mystery Dungeons (psst: this is mine)
  • Marlowe Dobbe: A Swarm of Monsters is Hard To Build: Generating Visual Concepts for Enemies in Roguelikes
  • John Bond: Doors or no Doors: How Roguelike games take you places
  • Dan Norder: Chase, The BASIC Language Proto-Roguelike

Sunday

  • Yong Zhen Zhou: Who Controls the Controller? Thinking about physical player interactions outside a digital game
  • Tabea Iseli: Animal Crossing meets Roguelite Dungeon Crawler – The surprising genre mixture behind Grimoire Groves
  • Philomena Schwab: 100,000 wishlists in 3 months – Weird roguelikes are taking over the world
  • Kaysa Konopljak: Going legit with DotA: How to transform a thousand authors into four
  • Alexander Birke: Practical procedural world and story generation in Sea Of Rifts, a naval roguelike RPG
  • Robin Mendoza: The Use of Knowledge in the Labyrinth: The price mechanism as a storytelling device
  • Ollie: The Right Variety – Understanding and Visualising the Output Diversity of Your Generators
  • Eiríkr: Uxn – Permacomputing & Roguelikes
  • Brian Cronin: Black Box Sim for Roguelikes
  • Isaac Io Schankler: Orb Pondering Simulator LIVE!
  • ?: 7 Layer Dip, Multi-Layered Narratives For Roguelikes
  • Emily Halina: New Levels from a Single Example via Tree-based Reconstructive Partitioning (TRP)
  • Courtney: Cheating the System (By Design!) for Epic Combos
  • Joe: Magic in Game Design: (Mis)Directing the Player’s Attention
  • Tyler Coleman: Finding your 80/20 Rule with Proc-Gen
  • Nat: Procedurality and the Primes

A Japanese Youtuber Plays Rogue

I only have the barest understanding of Japanese, and the auto-translation on this video is pretty bad*, but I still found this Japanese Youtuber’s experience with the Steam release of Epyx Rogue to be interesting (27 minutes):

They keep using terms from Chunsoft’s Mystery Dungeon games, especially Torneko no Daibouken from the Super Famicom, but seem to have a good sense of how those items connect to and were inspired by Rogue.

Your armor weakens, oh my! “All the F words in the world were about to come out.”

They also die a lot. Because Rogue doesn’t want you to win. It was made for a community of players who would play it over and over, and were competing on shared scoreboards on university machines, and indefinite play makes for a poor measure of player skill. Standing and trading blows with every monster is a bad strategy in Rogue in the long run. Instead, it helps to run from strong enemies, to build up more hit points so as to defeat them, and sometimes in order to escape them to the next floor. Rogue’s monsters grow in strength as you descend fairly quickly, and the player is usually not far ahead of them in the power curve. Then around the time Trolls show up they’re roughly an even match, and they keep getting tougher. The point where the monsters become stronger than the player is different every game, and depends a lot on which items the player has found and has identified, but it always comes eventually. They eventually get pretty far, dying on their fifth attempt to a Griffin on Level 18.

*
“Water supply texture: Say goodbye to the smell of raw oysters.”
“The Dora doll’s twisty honey positive is getting warmer.”
“I miss the days when I used to go Hee Hee in Centauros.”
“Let’s quickly wash and throw away the rotten plastic bottles we drank from.”
Tell me more, auto translate bot!

@Play: Which Is Better, Ring Mail or Splint Mail?

@Play‘ is a frequently-appearing column which discusses the history, present, and future of the roguelike dungeon exploring genre.

Gary Gygax was a weird person. I won’t get into his life or history or, the casual misogyny of AD&D character creation, or the Random Harlot Table. But he did know a lot about medieval weaponry and armor, and to some degree this obsessive interest seeped out and infected a whole generation of nerds.

How useful is this armor in protecting someone? Five. It is five useful. (Image from National Museum in Krakow)

I know which is generally better: leather armor, studded leather armor, ring mail, chain mail, splint mail, plate mail or plate armor. I know that, although in life each is different, battles are random, and there’s countless factors that might determine who would win in a fight, the order in which I have given them is roughly how effective they are, because it’s the order that Armor Class increases, sorry decreases, in classic Dungeons & Dragons.

While the list of armors is presented, in practically every Player’s Handbook, with their effects on protection right there in order, unless you’re steeped in the material, it is not obvious, just from reading the names of the items, which is supposed to be better than another.

Splint Mail was rare in Europe during the medieval period. It’s also really hard to Google Image Search for without ending up with pictures taken from D&D material!(Image from Wikipedia)

This is a considerable roadblock, and one I struggled with for a while, when I first tried to learn to play Rogue, because that game expects you to know how effective each piece of armor is. You start out with Ring Mail +1. You find a suit of Splint Mail. Should you switch? People who play nearly any classic roguelike are going to run against this eventually. Even now, some games just expect you to know the relative strengths of each.

If you decide to take the chance and try it on, to Rogue’s (and Nethack’s) credit, it tells you immediately how effective the armor is on the status line, and you can compare its value to your past item. To Rogue’s (and Nethack’s) detriment though, if the new armor is cursed, you’re stuck with it, until you can lift the curse (to a new player, unlikely) or die (very likely). And then, unless you’ve been taking notes, you’ll still probably forget the relationship between the two items, meaning you’ll have to guess their relative value again later, and deal with the same risk.

Classic D&D tended to give short shrift to the intricacies of real-life armor use, simplifying a complex topic beyond perhaps what was appropriate. AD&D attempted to remedy that by going overboard, giving each armor ratings according to its bulkiness, how much of the wearer’s body it covered, how much it weighed and how it restricted movement. Gygax’s tendency towards simulation is responsible for some of the most interesting parts of the game, but it didn’t help him here I think.

Most classic roguelikes, at least, use the “bag of Armor Class” approach to armor, which is probably for the best. Nethack probably goes to far in the Gygaxian direction. If you find Plate Mail in Nethack, you’re almost entirely better off just leaving it on the ground, even despite armor’s huge value, because it’s simply too heavy. Even if you can carry it without dipping into Burdened status, or, heaven help you, Stressed, its mass and bulk lowers the number of other items you can carry before you reach Stressed, and carrying many other items is of great importance. This is the secret reason that the various colors of Dragon Scale Mail are so powerful in Nethack: it’s not that they have the highest best AC in the game (though they do), it’s that they’re also really light! Even if you don’t get the color you want, it takes concern about the weight of armor completely off your list of worries.

The use of armor underwent revision throughout D&D’s development. (This page lists the changes in detail.) For reference, the relative quality of D&D, and thus roguelike, armor goes like this.

NameNew-Style Ascending Armor ClassOld-Style Descending Armor Class
Leather Armor28
Studded Leather & Ring Mail37
Scale Mail46
Chain Mail55
Splint Mail & Banded Mail64
Plate Mail73
Plate Armor82

Why the difference in values? Up until the 3rd edition of D&D, Armor Class started at 10 and counted down as it improved. 3E updated a lot of the game’s math, and changed the combat formula so that AC was a bonus to the defender’s chance to be missed instead of a penalty to the attacker’s chance to strike. Because of that, now it starts at 10 and counts up. The changeover was a whole to-do, I assure you, but now two editions later we barely look back. Back in that day others were confused by the system too, and even Rogue used an ascending armor score. But Nethack, to this day, uses original D&D’s decreasing armor class system.

If you compare those values to those used in 5th Edition, you’ll notice that even the new-style numbers don’t match up completely. As I said, while the relative strengths have remained consistent, if not constant, the numbers continue to change slightly between versions.

That concludes this introductory level class. You’re dismissed! If you’re looking into the relative effects of different polearms… that’s the graduate-level seminar, down the hall.

Episode 3 of BS Shiren the Wanderer Recovered

The Mystery Dungeon series of Japanese roguelikes, which includes the Shiren the Wanderer games, has a fair number of obscure entries. There’s “The Rainbow Labyrinth,” a mobile entry that toyed with adding F2P features and never made it out of beta. There’s a few other mobile remakes of early titles that can’t be obtained or played now due to their platforms being discontinued. And back on Super Famicom, one of the very first Mystery Dungeon games, a spinoff and modification of Furai no Shiren, was released for Nintendo’s Satellaview add-on.

Most Satellaview titles are extremely obscure now, with their only remaining remnants that aren’t languishing in a vault somewhere inside Nintendo (if they even exist there) being saved data files on aging flash memory cartridges in the possession of diehard Nintendo players and collectors in Japan. Satellaview was treated as a way of distributing disposable software, games and other programs that were tied to a specific date or time, so there are a good number of lost items for it, and many will probably never be recovered.

Entropy and bitrot are huge problems with computer software of all types, and it’s shocking how little most companies, even Nintendo themselves sometimes, seem to think about recording essential parts of their past. So any successful reclaiming of old data from the land of howling hungry ghosts is good.

Image from Satellablog

That’s why I’m remarking here that Satellablog, dedicated to recovering and making playable as much old Satellaview software as they can, has managed to obtain a copy of Episode 3, of the Satellaview version of Shiren the Wanderer, “Save Surara” or “Save Surala” depending on the tastes of the person romanizing the title. That means episodes 2, 3 and 4 have been found, leaving only the first episode.

Save Surara was a Soundlink title, like the releases of BS Legend of Zelda. That means they were intended to be played at the same time as a special audio broadcast, and contained events that were timesynced with that broadcast. Without the broadcast (which are usually lost now), Soundlink games can’t be entirely played as originally intended, but it’s still better than nothing.

Here is video of Episode 3 in action. It’s about 49 minutes long. It’ll have to be modified to get it into a state where people who aren’t into romhacking will be able to play it themselves:

With three episodes recovered, there’s still hope that someone in Japan saved a copy of Episode 1 on a forgotten flashcart resting in a closet somewhere. Frog bless all of you awesome hardware horders over there!

C64 Dungeon Play and Lost World Demonstration

Another personal project post! I have done more work in making David Caruso II’s obscure Commodore 64 CRPG Dungeon, published in the issues of the disk magazine LOADSTAR more than once, presentable to current-day audiences. Although it certainly has its limits, there are some aspect to it that are unique, even forward-thinking. We posted about Dungeon here before. To remind everyone, we sell Dungeon on my (rodneylives’) itch.io page for $5, with the blessing of rights-holder and LOADSTAR owner Fender Tucker.

There are a few bugs in Dungeon, now basically impossible to fix, that I’m trying to track down and document, and I’m also working on improving the documentation, as well as provide some useful goodies with the system, like a disk of monsters, equipment and magic items. That’s useful because Dungeon has a special feature where it’ll take the monsters and items on a “Data Disk,” and scatter them around a dungeon map of its own creation. It calls these randomized adventures “Lost Worlds.”

Lost Worlds operate as a kind of quasi-roguelike. The Dungeon software creates a random map and places random items around it, but once created it becomes a Dungeon adventure that any created character can explore as many times as they like. While it doesn’t have roguelike tactical combat gameplay or random item identification, it does have a form of permadeath. Characters only get three lives to advance their level as far as they can go.

Lost Worlds are interesting places to explore, but there are some bugs in them. It is possible, in fact pretty easy, to get stuck in a part of the dungeon from which one can’t escape. Sometimes a one-way door leads into an area that can’t be escaped, and sometimes a passage-blocking trap will strand the player’s character in a dead-end. And once in a while a Lost World is downright unfinishable, its goal item disconnected from the parts of the dungeon the player can even reach.

While there are spells (Passwall and Teleport) that can release a trapped character, if they aren’t available the character is not completely lost. If you turn off the C64 (or close the emulator), then return to the Guild screen, the character will be marked as GONE. Over time, measured in loads of the Guild menu, the character will eventually find their way back on their own. It takes quite a while for this to happen though: I counted 15 loads, saving the game each time, before a GONE character returned.

This video (23 minutes) is is something I recorded myself as a demonstration of both Dungeon’s gameplay, and its Lost World adventure generation. It uses a set of 30 low-level monsters and items based on the stats of the old Basic edition of D&D, and a set of magic items I created for usefulness and to show off Dungeon’s spell set.

So, why would someone want to play this game, when there’s so many other newer CRPGs out there to play?

  • The idea of rolling up a character and taking them through scenarios made by other people, to try to get their level up as high as they can get before they die three times, is great. My hope, perhaps misplaced, is this release will inspire other people to make dungeons for others to play, and I look forward to seeing them myself.
  • The magic system of Dungeon, while it doesn’t allow for characters to learn spells themselves, is unique in that most of the spells are utility spells! There are spells for passing through walls, for teleporting anywhere on the map, for revealing terrain, for seeing in darkness, for giving oneself a damage shield, for locating the goal item, for disarming traps, and more. There is only one direct damage attack spell! Spells are more like tools than something you use to pound through the enemies.
  • The dungeon model allows for dark areas, traps that block exits, two-way and one-way teleporters, secret doors, one-way doors, and decorating dungeon maps with PETSCII graphics. The simplicity of the dungeons, all of them fitting on one screen, works in Dungeon’s favor. No dungeon can be too large since they must all fit within the bounds of the map grid.

There are unique design considerations for making Lost Worlds too. Even though the computer creates the maps unaided, since it populates them from the monsters, items and traps that are on the Data Disk, the difficulty of the resulting dungeon is affected. The various doodads are distributed without apparent heed for what they are; I wonder if the generator actually cares for their identities or if it just checks how many of each type are on the disk, so as not to exceed that number.

If there are more easy monsters, more powerful items, and more weak traps on the disk then the dungeon will be easier due to their corresponding numbers being greater, and vice versa. It occurs to me that one of the flaws in the dungeon generation I mentioned could be alleviated, by not giving it one of the wall creating traps that could trap a player in a dead-end, but that also makes the dungeon a bit less interesting, so I’ve left it in the mix I use.

I recognize that, if I let myself, this might become a Dungeon blog. Rest assured, I’m not going to take it that far. But I really hope that some people give Dungeon a chance. While sure it has its inspirations (one person on Mastodon said it reminds them of Phantasie, a somewhat less obscure early CRPG), I think it’s pretty unique, and deserves for more people to have a look at it. I’m particularly pleased how well the sample monsters and items I made work in the Lost World framework, and I’m trying to think of ways that it might be improved. More on this later… but, not immediately, I think.

Shiren 6: What Happens When You Finish The Final Dungeon

I’ve been playing a lot of Shiren the Wanderer: The Mystery Dungeon of Serpentcoil Island lately. Partly in preparation to add a chapter on it to my Mystery Dungeon book, partly because I like Mystery Dungeon games. I streamed my playthrough of finishing the main dungeon (on my first attempt!) here.

Here is the title screen (which is a spoiler for after finishing the main dungeon, although that is really only a short way into the game):

After you finish every other task in the game, including finishing the final 99 Floor “megadungeon” where most items are unidentified, the title screen changes to add a nice rainbow:

I forgot to get a picture with the title in place. I can’t go back and get it now because of what followed….

There is one more thing to do at that point though. That is to play through the megadungeon again, but finding 12 “Celestial Stones” that severely restrict your inventory by the end.

Well, I’m not sure if they really counted on anyone doing that? There doesn’t seem to be much reward for it. It doesn’t go remarked upon by anyone in the game. But it does change one thing: the title screen. Here it is:

I like the red “IN SPACE” stamp! Sadly, all the graphics in the actual game still show an island floating in the atmosphere, and not in orbit. I wonder if they plan on doing something with this in an update? That seems like a lot of extra work for the benefit of not a lot of people.

Looking through my screenshots, I found this illustration that can be unlocked for behind the main menu, showing Shiren stumbling upon a Monster House:

There’s a lot more to say about Shiren 6, after I gather up my thoughts about it….

The CRPG Addict Reaches Nethack 3.1

He’s been at this since the days of GameSetWatch’s run of @Play, but the CRPG Addict has finally reached Nethack 3.1, the game where Nethack reaches most of its final form. It’s true that it has gained features since then (especially weapon skills and splitting apart race and role from each other), but it was the version that introduced the current-day structure of the dungeon, added the many role-specific Quests, made the Wizard’s Tower a three level stronghold instead of just a little place in the mazes of Gehennom, put in the Bell, Book and Candle subquest, handed the Amulet of Yendor to the High Priest of Moloch, and put in the Elemental Planes and the current form of the Astral Plane.

Here are all his Nethack 3.1 posts to date:

Game 504: Nethack [3.1 series] (1993) – Beginning adventures

Nethack [3.1]: Blessed and Cursed – Discovering exercise, death by battery drain

Nethack [3.1]: Rust and Ruin – Beginning of very good game

Nethack [3.1]: Quest for Glory – Middle dungeon levels, Rogue level, Quest branch

Nethack [3.1]: Wish List – Exploring the lower regions of the main dungeon

Nethack [3.1]: Beyond This Place Of Wrath And Tears: from the Castle to Fort Ludios to entering Gehennom to killing Baalzebub

Nethack [3.1]: Nothing Lasts Forever – from killing the Wizard for the first time to getting the Amulet up to escaping the main dungeon for the Elemental Planes

@Play: Glorious Adventure in the Mystery Dungeon

@Play‘ is a frequently-appearing column which discusses the history, present, and future of the roguelike dungeon exploring genre.

It’s the shortest @Play column ever!

What is happening here? This is the newest Mystery Dungeon game, Shiren the Wander: The Mystery Dungeon of Serpentcoil Island. It’s actually a great deal of fun, a sharply-designed entry in the long-running classic roguelike series.

This isn’t “roguelike” like half the games on Steam. This is a true roguelike, even if it doesn’t have ASCII graphics: a turn-based RPG with substantial randomized elements, that demands that you live (or not) by your tactics, strategy and wits. I don’t begrudge others appropriation of the term, but it does mean I have to now use the qualifier classic when I want to discuss the old style. Really, it’s better to call games not in the original style roguelite.

The dungeon depicted is Heart of Serpentcoil Island, the traditional end megadungeon that most Mystery Dungeon games have. After finishing the “main” dungeon, and playing a lot of extra bonus dungeons that each show off a specific element of the game’s engine, there’s the megadungeon: a 99-level gauntlet of terror where you enter at level 1. None of the items you’ve collected throughout the rest of the game will help you here. You must start from scratch with just a riceball. You don’t even have a weapon or a shield to begin with: everything you have, you must find along the way.

The game doesn’t pull many punches in this dungeon, as you can see. At experience level 1, every space in a room (other than the entrances: that’s a secret tip for you!) could contain a game-ending trap. The only consolation is that they’re really quite rare! I was exceptionally unlucky in this run.

Additionally, the uses of many of the items, the scrolls, grasses, pots, bracelets and incense found in the dungeon, are unknown: their effects must be discovered, through means both blatant and subtle, for yourself. Some of them will be essential to your survival, let alone success; others, like the Ill-fated Seed, you really want to avoid using.

It’s a ludicrous test of knowledge and skill, and a fitting capstone to the game. If the experience shown in the video seems like it might put you off, at least it shows conclusively that the game isn’t taking it easy on you. If you win, and it can be won, it’s a great accomplishment. I’m still working on this one myself; I’ll let you know how it goes.

Shiren 6 Main Dungeon Completion

Whew! At around 3:30 AM last night, I finally was able to play Shiren the Wanderer 6: The Mystery Dungeon of Serpentcoil Island. Five hours later I finished it on my first attempt, with no deaths. I had six Revival Herbs in inventory at the end. (There is a story battle against the boss at the beginning that you’re supposed to lose. I don’t know, maybe there’s a tricky way to finish it? I gave it a good try.)

I did this on my rarely-used Twitch channel. I’ve put the recordings of the play in a video collection, which you can see here.

I’m not going to say this means the game is easy. I can finish Super Famicom Shiren on one try too. I’ve been playing roguelikes for, oooh, over 30 years now? Shiren 6 falls a lot closer to the first Shiren games than Shiren 4 and 5, and I couldn’t be happier about that. I plan on written a full review later, after I’ve recovered from staying up all night playing this game.

I’m sorry that this isn’t more generally interesting, but I’m pretty jazzed!