[Review] Into the Maze of Dreams (OSE); Ai-generated shit tier digital litter

[Adventure]
Into the Maze of Dreams (2023)

Luiz Eduardo Ricon (Hexplore Publishing)
“First Tier Characters” (lvl 1 to 5)

Shit Shit Shit. Digital litter. Garbage. Dregs from the latrines of the mentally ill. I am without words, incensed to my very soul, my fury is olympean and burns with the heat of the very fires of Hell! This adventure is low-effort garbage that is worthless for everyone and I will run you through the process of discovery so you won’t fall for its ilk. Please refer to bolded text for indicators of shit quality.

36 pages. It begins with a page full of ads, then title page, then credits which is spammed full of links to social media and drivethru. First page: First tier characters is not a concept in core OSE. So why is this terminology being used? Text is bloated with redundancies to pad out the page count and give you the idea you are getting something. The adventure claims compatibility with generic OSR and boasts that you can easily adapt it to Pathfinder, Five Torches deep, Shadowdark, Maid the RPG, Vampire the Masquerade, Monopoly, Call of Duty Black Ops 2 and Mahjong. The common OSR wisdom states that most old school gaming systems are interchangeable. On the one hand, this is true, you can run a B/X adventure in AD&D. On the other hand, each system has a host of different assumptions and subtle differences that will make a real difference in an adventure with substance that is made to actually be played. Extra spells, differences in statt limits, altered cleave rules etc. All that might require adjustment in order to make it run. “Can be adapted,” is bullshit. Everything ‘can be adapted.’ I can ‘adapt’ my adventure to run in mahjong. Claiming broad compatibility with every system under the sun is empty marketing bullshit.

Second point: A trigger warning. + safety tools + session zero. Trigger warnings are a pseudo-scientific concept, brought into popular culture by malignant agents seeking an excuse to dictate its shape, using weaponized empathy and moral camouflage. This has been used to great effect although by now the ruse is mostly apparent. The idea that someone playing an adventure could get traumatized by something as vague as ‘psychedelic elements’ is hard to credit so including it is primarily a means of signalling conformity with the dominant cultural narrative, and this tells you its target audience is composed of the most conformist, least critical, most zoned out dupes, and this message is included to re-assure them: ‘Don’t worry my fellow heckin wokerinos, I’m just like you. Buy this adventure and let us together fight the evil nazis/putler/drumpff/coof/global warming/the dictionary/chuds!
As long as the tranquility of their carefully tended bubble reality is not disturbed, this audience will generally buy anything with bright primary colors on them. Bring in the heckin’ dollarinos! Once you have signalled that you are willing to appease whatever histrionic demands are put forth by the more unstable members of such communities (this is colloquially known as ‘paying the troongeld’ or ‘Xandering’), provided you don’t become the target of a cancellation attempt by an inadvertent comment or simply random chance, you are basically good, and no effort need be expanded on expensive innovation to improve the quality of the product. The moderators of such communities will generally squash any sort of critical voices bemoaning the incompetence of the designer and low quality and effort of the product, with the exception of charges of insufficient narrative conformity. The takeaway point is that in any commerical ecosystem there are different strategies for different audiences, and Adventure Quality is only one particular attribute one can focus on.

What does this page really communicate? Yet another fucking overview. But the adventure is only 36 extremely padded pages, 8-11 is the city (which consists of a list of what appear to be ChatGPT generated NPCs), 12-20 is the dreams leaks, which is a single encounter followed by several possible hooks (again, these look AI-generated), and 20-30 is the dungeon, which consists of seven rooms. Do I really need a seperate overview page to help me navigate that? No but it helps pad out the page count. We are using random tables? Wow, what a novel concept. The OSR has been around for 20 years, I have never heard of this before!

This is another good indicator of a scam. It’s not that there is not ever a good reason to use ToTM (say, in places with unlimited boundaries like the planes or with simple encounters during a hexcrawl), but try to imagine, for a moment, all the complexity, all the decisions and factors that are encompassed in something like this:

And then collapse that into this.

Theoretically there is no reason that all that interaction, all that information, all these extra factors, cannot be compressed into those points. After all, how many angels can sit on the head of a pin? In practice, this almost never happens, so Theatre of the Mind is usually an indicator of lack of complexity and a corresponding lack of effort.

Long-winded explanations of patently obvious things to further pad out a worthless adventure. Observe generic adventure hooks:

By now I had begun to suspect this text was AI-generated. There is something about its lack of substance, its generic, grammatically pristine but soulless sentences that can provide only surface mimicry of the structure of the actual adventure. Notice the generic descriptions, the lack of specificity. Mechanistic adjectives, applied with metronome precision. Grindstone has dark catacombs in which are located mysterious secrets and deadly dangers. GEEEE. Nothing specific. How many people, how much is the inn, can I get retainers etc.

Notice exact paragraph length, generic descriptions, lack of specific information (alignment, levels, etc), lack of specific gameplay permutations (Elysia can provide the players with ‘crucial leads’ and ‘hidden knowledge’. Nothing about how this relates to the events of the particular adventure. Another page is exactly like this.

After the DM has wasted the players time with scenes (oldschool games don’t have scenes), we may now begin the adventure proper with…generic bullshit. Please note that none of the strange occurences have any sort of measureable gameable effects and are just there to further pad out the adventure. Also: ‘This distortion can cause confusion and disorient those unfamiliar with the Sleepers slumbering influence.’ Its almost as if the author was some sort of elaborate language model that uses big data to predict the next word in a sentence without having any sort of awareness of the meaning of any of the words it is using.

Observe the subtle indication that you can actually skip the single encounter, which comprises 1/3 of the adventure, and move on to the last part if you want to do a ‘more streamlined, one-shot kind of adventure.‘ So this encounter is also padding. I feel it is worth discussing because it is, at least, an encounter, thereby implying something is happening, something the author conceived of that can be interacted with. In this case, we are informed we can heal ‘the guards’ (no stats) to ‘reduce the amount of enemies we will face’ who ‘arrive in waves of 1d4, with a minimum of 1.’ Excellent coverage, I was wondering what would happen if I had rolled a 0 on the 1d4, well done, chapeau! And then we are belatedly informed that if we ‘healed the guards’ the waves are reduced by 1, supposedly referring back to the minimum of 1. Why in the name of garl glittergold you would not simply give the guards some fucking stats is beyond me, and how many guards does Grindstone have? FUCKING TRASH GARBAGE. AN APE COULD SHIT A BETTER ADVENTURE. After that excitement, the players are rewarded with 50 gp per horse. This is the totally new and awesome monster btw:

Next three pages are all fucking long hooks AGAIN HALFWAY INTO THE ADVENTURE. ‘Offers them a request with the promise of a generous reward.’ Literal word salad padded nonsense. Completely irrelevant. Three fucking hooks to get players to investigate the 7 room dungeon, something that could be handled in a paragraph at the start of the adventure.

Two more hooks like this, braindead. Section where you hike up the mountain. Notice lack of gameplay. A paragraph of description. Nothing happens. No interaction with the rules. No random encounters. Nothing.

Something should happen but it does not. Again, nothing. Encounter on the top of the mountain. No gameplay. No duration. No map. No design.

20 pages in we reach the Maze of dreams. 8 rooms. Do we think they will have gameplay in them? We step through the magic fissure under the waterfall and are transported. 20 pages of complete nothing in we reach the maze of dreams.

Room 1, nothing happens. Notice reference back ot the lonely man section, which is approximately two pages ago. So much had happened that the author was unsure if you, the idiot that bought this trash, still remembered the thing that was described a page ago. ?


Room 2. A puzzle. Since there is no penalty or time limit to the test, anyone with knowledge of primary colors or sufficient persistence should eventually be able to solve this puzzle, although I am a bit concerned that the adventure does not include tips in case one is running it for the blind or achromatopsic.

Room 3. The hallway of distortion. I would have also accepted ‘walk forward with your eyes closed.’ This would be one typical encounter in a 34 room dungeon taking up 7 pages. I suspect also it was AI generated, but we shall see.

Notice use of doors already? What gives? I thought we were doing theatre of the mind? Now we suddenly have a choice? We are treated to generic combat with a generic antagonist. Notice fucking bullet point descriptions of rooms content, priming the user for a 2 paragraph room with 3 doors and a combat.

Gee thanks for the permission. I really want to use all these wonderfully creative monsters like ‘mist monster’ and ‘fire horse’ in my campaign, but I was not sure if Luiz Eduardo Ricon, author and possibly non-sentient AI language model, would permit me to do so.

This looks like game design because the PCs might suffer some sort of random effect with game mechanics but since there is no procedure for movement through the abstract, 0-dimensional dungeon the 1 hour time limit is arbitrary. In addition, some of these effects have no listed duration. The strange creatures are of course not defined, and the room is pointless.

This is the second room 5, but since this is a place of non-Euclidean geometry, this can happen. There is now a maze, and in order to traverse the maze, you must roll high enough times. No decision that the players make will affect their progress through the maze. Horrible.

2 pages for a pointless random table and a set piece combat. There is a sort of theoretical game design because you must use the item from this room in room 7 to collapse the dungeon. Why is the number of statues random btw? You incompetent poltroon. Can you not even fathom the purpose of random encounter numbers when you penned this imbecilic doggerel and put it up for sale you stinking subhuman ape?

I’m cautiously optimistic that 6 statues with two attacks per round, sleep immunity and AC 4 against a 1st level OSE party of zoned-out coomers could be a potential TPK so perhaps there is hope to be found after all.


The last room you have to dunk the item in a pool and then run away because the cave will collapse. In order to simulate a sense of urgency it apes Holy Mountain Shaker and you must roll on a table every room for a generic hazard, only unlike Holy Mountain Shaker, which is not a good B/X adventure but at least it was inspired and it tried something interesting, it fucking sucks. The PCs must give some sort of description of what they do and you just roll randomly. If they get enough 3 out of 6s, they make it, otherwise they die. That’s it, adventure over. I think I am going to be sick.

The party is given Garvil Thorblend’s treasure if they ‘helped him’ for a total of 4700 gp and a magic item, which is suprisingly generous.

This adventure, and others like it, are copper bestsellers on drivethru. Complete, utter, shite that someone with the aid of randomly generated text can put together in an hour that is lapped up by uninformed buyers. One well deserved Star, an OSR Crusade against Hexplore Publishing and a nelsonian Ha-Haaah to the customers who were tricked by this junk.

*

Postscriptum: Perhaps a useful illustration of the game design chops at work would be to consider the path of optimum risk/reward. If the party wishes to avoid tangling with the spooky scary monsters or die from random chance, their best recourse would be to simply escort Gavrin to the first room of the dungeon, watch him dissapear, then get the location of his strongbox, and thus gain all the treasure in the adventure. The difference between this approach and actually exploring the dungeon is approximately 600 xp, and avoids the chance of TPKs from high numbers of mist fiends (which have a blinding breath weapon attack) or Statues (which are only 35 xp despite having AC 4, 3 HD, sleep immunity and 2 1d6 attacks per round).













25 thoughts on “[Review] Into the Maze of Dreams (OSE); Ai-generated shit tier digital litter

  1. What a crock.

    I currently work on LLM outputs and you’ve described the non-style they produce perfectly. I think they have a place in our beloved hobby (parsing overwritten, fluff-bloated rules text for one thing) but the creative heavy lifting is not it. Similarly, Theatre of the Mind is not without its merits elsewhere in RPG land but here, in the donjons of the OSRmen, it cannot bear the bulk of the play.

    Perhaps the Trigger Warnings bit intrudes here as part of the LLM manifestation? Modern RPGs have trigger warnings, ergo this has to have one, whether it’s warranted or not. The idea of a Session 0 before a one-and-done module is daft, and I say this as someone who finds Session 0 useful to bring everyone onto the same genre co-ordinates, brief on house rules, build characters that’ll function together, and so on.

    Like

    1. I’ve been running ChatGPT for months doing adventure, plot, and other RPG ideas through it. Not a single time in all the hundreds of hours of me banging away at it has it ever proferred up a “session 0” or “X-Card” routine whatsoever. 

      Even the art reeks of Midjourney color schemes with the contrasting colors. The garbage room 5 (the first one) where the monster mimic’s the PLAYERS not their characters is something that does not belong. The point of an adventure is to seperate the person from their avatar, not combine them or EVER use the actual player.  This is why so many of the modern “OSR” authors fail miserably as they can’t separate themselves from the PC they play.

      I live for Escapist fantasy of giant bus like lizards being killed by pointy eared magic elves and dwarves with axes. 

      I long for the day we can get an AGI based on Blooey…

      Like

      1. Duly noted. Lazy human choice it is, then.

        Even in storygameland, divestment and distance is required. Overidentification with one’s character, inability to separate one’s real self from the fictive universe, is where madness lies.

        Like

  2. 4700 gp for a level 1 party is insufficient. Why do you pretend that is generous. They still aren’t leveling.

    Like

  3. I feel like this was created on a dare. “I bet I can get to Copper with a totally generic garbage AI-generated OSE adventure.” “Nuh-uh, it’s not THAT bad out there yet!” “Well, let’s see…” It almost seems like the kind of thing that Prince himself would put out as a grand jest.

    It does make me truly sad to hear that there are sufficient people to purchase, read and play this sort of thing. Prince, tell me this is all a bad dream, or a good joke.

    Like

    1. Gaze upon the ruins that lie beneath the big tent.

      I suspect its AI-generated trash, but it might just be horribly uncreative trash writing. Its a Brazillian OSR special, apprently, so I could have foreseen it was trash.

      Like

      1. I almost left an “I’m offended” comment on a blog! You sir are good at this. Please continue to do what you do.

        Like

      2. It kills only animals, Let us say I suggest you may be human. Steady! I warn you not to try jerking away. I am old, but my hand can drive this needle into your neck before you can escape me.

        Liked by 1 person

  4. Oh boy.

    I like that the choice of bold text works in exactly the opposite way: instead of helping it hinders. If you only read the bold text, say, as a way to parse the important points of the “adventure”, you get close to Gary, the (sole) vault dweller of Vault 108 in Fallout New Vegas (spoiler alert, he is gone into the deep end and communicates by emoting one word only ~win a No Award for guessing the word).
    For example, page 15 gives you this:
    “Magda Silverhorn magic-user Flaming Steeds Mt Somnus Magda Magda Magda Magda Grindstone Mt Somnus Dreamwater Cascade Magda Magda Dreamwater Cascade Mt Somnus”
    Positively Shakespearean, amirite?
    The entire adventure is pinnacle design in genuine meta-nega-design, ie if you want to write an adventure, then do the exact opposite of what this adventure does; every single design choice made in the text is a “No No” for elfgame design.

    Now, on another front, I also have to say that that Xandering link is something.
    I suggest replacing “Xandering” in the text-link by “Neckbeard caught with his hands in the cookie jar provides non-pology”. (*)
    JA’s mealy-mouthed, conceited diarrhoea of text says nothing good about him. But for purposes of clarifying to anyone that he is truly a POS, it’s flawless. And we are to believe he is a respected Developer, no less, at Atlas Games. As a gamer, he should know that when one finds oneself in a hole, one should stop digging. Legend!

    (*) Yeah, I noticed he shaved his beard. It is a shame, verily. That beard had more personality and honesty than him.

    Like

  5. I completely missed that Justin Alexander wandered into the crosshairs of internet drama. It only got worse for him after apologizing (although referring to non-linear dungeon design as “Xandering” is a bold move).

    How fun to see you and Melan get name-checked as part of “that crowd” to be “contend[ed] with.”

    Like

    1. Yeah I tried contacting him about posting deranged accusations of the sort without ever trying to contact any of us but I have no recourse to assume he was just too absolutely devasted at the pathetic stooge he had reduced himself to to be able to lift his fingers and reply. Feel free to ask him of course.

      If the contention is that I believe that we should gatekeep lunatics, circus freaks and sociopaths from this section of the hobby then by all means put me in the crosshairs. But I will repeat my earlier observation, that if my objective was to gatekeep the merely mentally ill, promoting the works of Jaquays, Stuart, Scrapprincess (now cured, bless him) or Emmy Allen is probably not the best way to go about that.

      Like

  6. Okay, this shit pushed me over the threshold so I decided to check if chatGPT (3.5) can be used in any meaningful way to help write a decent – not exceptional, but at least decent – module for home use. I am a bit disappointed.

    I never expected it to be too useful, just to give me a springboard of ideas to build upon – very much like you would use a set of random tables, like Tome of Adventure Design -, and to do some of the algorithmic dirty work. I asked it to generate a dungeon (the ruined swamp fortress of the radioactive lizard men) of 20 rooms with 50% of it empty save for decoration, 12,5% with monsters, 12,5% with monsters and treasure, 12,5% with treasury only and 12,5% with a trick or trap. I also asked it to specify treasure value in gp, and that the sum of the treasure should amount to 40k. Finally, I asked it to give the dungeon a 1930’s sword&sorcery vibe.

    Oh boy. It was funny and promising at the first hour, but became increasingly frustrating. chatGPT is really dumb. I could barely make it properly understand the simplest of requests, and when it finally got something right I had a hard time reproducing the outcome later, with basically the same prompt. Or maybe I am just unable to properly use it. Anyways, it seems to have problems understanding basic mathematics and things like ‘treasure’, and the patterns in its responses become obvious quite fast.

    Anyways, I was able to lay down the foundations of a dungeon quite fast, but when I started to ask for details in each room, I got such a great dumpload of contradicting bullshit information that it would need very heavy editorial work to boil down to something neat.

    Verdict: it had a few ideas that I could build upon. Some of these ideas i even liked, because it surprised me. But it took quite a while, I had to ask many specific questions, and if I really wanted to use the outcome, I would have to revise everything. I would have to edit the text very heavily to make it useable, rework the encounter types and numbers, detail the traps, treasure and magic items, draw a map and connect the rooms in some way, think of the whole place and how the inhabitants relate to and affect each other, etc. Maybe I could use it to kickstart the creation of an adventure if I am very uninspired and really have to make something for tomorrow’s game, but I am not sure this would be the most time-effective tool.

    Like

    1. I did a similar experiment a couple months back to have it write a “bear themed” adventure; I was actually rather pleased with the creativity the AI displayed; however, A) it worked best with a small (12-15 encounter) dungeon, B) it has no real design sense as far as mechanics (it’s more an ‘idea generator’) and C) it can’t do maps.

      As such the TIME it would take to render an AI conceived dungeon as actually usable makes it counter-productive as a tool. If one already has a strong concept, it’s best to just pen it yourself.

      I mean…unless producing shovelware is your side hustle.

      Like

Leave a comment