Pseudonym Writes

Just another site by someone who refuses to give their own name.

Review: Old School Feats by Eric Diaz

I’m predisposed to like Eric Diaz’s Old School Feats. Firstly, the author sent me a review copy, which was very nice of him. Secondly, the core pitch appeals to me: that by adding one more layer of choices (feats) to old-school RPGs, you can actually simplify a lot of the weirdness (monks and barbarians are clearly just types of fighters, druids are clearly just types of clerics, etc), while also expanding the available character concepts. 

However, I was a little concerned that a lot of the fun of modern OSR games is in the wacky classes (e.g. here’s a table of 100 GLOG wizards.) Obviously, no single book can replicate the creativity of the entire OSR scene, even via combinatorial explosion. One obvious solution would be to simply combine wacky GLOG-style classes with the resources in this book, although unfortunately a lot of them are tied into highly specific mechanics of the traditional core classes (more on this later.)

But this book doesn’t entirely abolish frivolous classes in favour of the core fighter/cleric/rogue/wizard(+race-as-class, optionally). There are also “packages”, which are basically classes (e.g. Sorcerer, Barbarian, Druid, Knight) broken down into feats. This is a nice open system that lends itself to homebrewing, and a lot of existing classes can be converted into feats this way with varying amounts of difficulty; a base that, if built upon, could easily produce a similar level of variety to GLOG or 3.5’s endless class options; I’m hardly the first to note that a feat is basically just a class feature with the class removed. I wish it had been expanded on further, however, it’s only a sparse two pages and the way it fits into the random class tables is largely left to the DM.

Like a lot of modern games, this book warns you that it’s kinda unbalanced and will break if you poke at it too hard. This always bugs me a bit, even if it’s more honest than pretending otherwise; balance is important. (Although I obviously appreciate that the amount of playtesting it would take to really balance so many options is not feasible.) However, it’s worth noting that literally everything in here can be applied via random table; that won’t appeal to everyone, but it has an inherent balancing effect in addition to the fun of seeing random characters come together. 

It’s also worth saying that there clearly has been a lot of work put into balancing it, even if they acknowledge it may not have been enough to handle a real munchkin. The feats in this book are, like in 5e, theoretically designed to balance against Ability Score Increases, although with a different system designed to keep ASIs balanced in AD&D style games (the increase is proportional to how low the Attribute was rather than flat; I can’t actually speak to how balanced this is without trying it, but there’s clearly some thought and likely some playtesting put into it.)

Highly Specific

On the other hand, the desire to let these feats replicate existing classes sometimes gets in the way of balance. To pick a random example, the Magic-User feat Charisma Casting simply lets a Magic-User’s casting attribute be Cha instead of Int. Is this actually an improvement, let alone one worthy of spending an entire ASI-equivalent on, or just a feat tax on people who want to play Sorcerer knock-offs? I could see a scenario where, say, you’re strictly rolling Attributes in order and get crap Int, but you/your character really want to be a Magic-User (although that’s exactly the opposite of how Sorcerers are supposed to work fluff-wise, that’s hardly an issue new to this book), but that’s pretty niche. If you’re randomly rolling feats, niche feats like this probably won’t apply to you – what are the odds your Magic-User happens to already have a higher Cha than Int!

This is part of a really core, though easily fixable, flaw in this book: the classic four core D&D classes are just as weirdly specific as the redundant ones this book is aiming to replace, and yet this book treats those idiosyncrasies as absolutely essential. Cleric feats rely on the exact mechanics of Turn Undead, Thief feats rely on the exact mechanics of Thief Skills and backstabbing, Magic-User feats rely on casting using Int and spellbooks and Vancian casting, and so on.

This book bills itself as being compatible with basically all OSR content. It’s not, not without a lot of house-ruling, so much that you may need to re-write or throw out easily half of the book. It’s not even compatible with the author’s own published OSR games system! Nor, RAW, is it capable of entirely replacing most of the classes it removes, because these feats only add abilities, never replace them (although see below for more on this.)

However, as I said, I think this is relatively easy to patch if (as seems likely) you aren’t playing exactly the brand of AD&D this game is actually designed for and compatible with. You just need to adjust the feat lists. 

The book already has clearly sectioned off which feats can apply to any character, which to any character with basic martial proficiency (the Fighter/Other list), and which are aimed at very specific AD&D classes (Thief, Magic-User, Cleric.) If you aren’t using those exact classes, under the exact AD&D rules this book is written for, then you’re going to need to go through them and remove or adjust a bunch of them. (Or, I guess, you could just not use them, but that’s 3 of the 5 feat lists in the book! Still, boosting martials never hurt anyone.) In fairness, there are even a small handful of notes on how to handle specific “variant” rules, but far far too few to actually make most of the book compatible with most OSR systems out of the gate.

Not all the class-limited feats actually break if applied to other classes, and in fact cross-class feats are specifically suggested in many of the feat packages. I would have liked if [class]-themed feats were separated from strictly class-limited feats, perhaps at the beginning of the table so you could roll with a smaller die if you don’t qualify.

As for those feats that seem almost designed for replacing, rather than augmenting, existing abilities… this is discussed briefly towards the back in the Optional Rules & Designer Notes section, where it’s made clear that this is how they were actually used in a lot of the playtests. But there’s no system provided for doing this yourself, which is a shame, (even if it’s very easy to improvise one). It’s presented as a throwaway optional variant house-rule even though a lot of the system seems to treat it as the default. 


Tradition demands that I comment on the formatting, artwork etc. of the book. 

They’re nice. The art has a rather nice black-and-white retro look, a little generic, but that’s probably deliberate. I’m not going to be printing pages out and hanging them on my wall, but there were some nice little touches that (as an aspiring artist myself) I appreciated and might steal when drawing in that style. Stuff is easy to find and (I wish this weren’t praise, but in the world of RPGs it is) easy to read. I did catch a couple of minor typos, but nothing that was actually confusing or presented an issue (and, as it’s digital release, these can be fixed on the fly.)

Should You Buy This Book?

If you’re part of the core target audience – people who want a way to spice up a game if B/X, OSE, Labyrinth Lord, or some other retroclone that sticks very closely to the original rules – then yes, absolutely. It’s a great upgrade to those systems in my opinion.

If you’re already committed to another OSR game … maybe. It’s only $5, there are definitely some cool system-agnostic ideas in here, and won’t take that much hacking to get it to work. Even if you’re playing a non-OSR D&D game (or D&D clone), there are some ideas in here that will likely work in your system, and it’s worth thinking about.

If you’re looking for something that will run perfectly out of the box with zero balance issues or duct-taping required to make it work, then no, probably not. Even if you’re running a B/X game. It’s simultaneously very fuzzy and loose in some places and hyper-specific in others. But if that’s what you want, why on earth are you playing a B/X clone? D&D was never balanced right out of the box.

If you do decide to buy it, I guess I may as well toss in an affiliate link to Old School Feats, although I’m not really set up for them.

Playing D&D In Real Time

A lot of OSR players have experimented with running games in “real time”, to varying but generally extremely positive results. 

Gygax famously claimed that “it is best to use 1 actual day = 1 game day”, and even that “YOU CAN NOT HAVE A MEANINGFUL CAMPAIGN IF STRICT TIME RECORDS ARE NOT KEPT”, although even he didn’t play everything in real time.

The way I see it, there are three kinds of “time” in RPGs.


This is what Gygax ran – “1 actual day = 1 game day when no play is happening”. Players can contact the GM and say what they want their characters to have done in the intervening time. (It seems to be particularly associated with people calling themselves the “BrOSR” as some kind of macho thing? Which strikes me as silly at best, but we shouldn’t let them hog the cool ideas). There’s been a wave of experiments with this in recent years on the OSR scene:

  • Jeffro extolls the benefits this had for his game (some people have taken to calling it “Jeffrogygaxian time” after his posts popularising it)
  • Aaron “The Pedantic” finds it helpful, particularly for adding impact to time spent healing etc, although he notes that his group do timeskip occasionally
  • Electric Rain uses it for both AD&D and 5e
  • Abrahams_Terror on Twitter raves about it, calling it “the glue that holds everything else together”
  • BDubs ran parallel games both with and without 1:1 timekeeping and declares it superior
  • Mister_Cranch of Vox Arcana podcast declares “I don’t know anybody else who does this, and I don’t know where I heard about it, but it works extremely well.”
  • HumbugNH uses real-world time and weather in their games
  • Musing Mage finds it helpful
  • Restless notes that posting real-time updates for his players keeps them excited for next session
  • Steven Smith has found “a pretty good response from the players… excellent … a lot of fun” as both player and GM 

It may help if you have robust rules for domain-level play and other downtime activities, which this style of play will encourage, but often the GM can wing it.

The biggest issue with this is the question of what happens if you’re not running adventures in real time; the interface between the two different rates of time. In particular, this style of play encourages people to have multiple characters or groups in the same world; but what happens if one group time-skips into next week while you’re playing at the table? And what happens if the PCs are in the middle of a dungeon when you stop? Many potential solutions to this run the risk of time paradoxes.

Incidentally, you could also run a game where each real-world day of downtime is a week, a month, a year etc. I haven’t actually run into anyone doing this with D&D-style TTRPGs, but e.g. many micronation roleplay groups do this, and it could give your world a really cool sense of history. It would also smooth out any shorter time-skips during play, although it could exacerbate the problem of PCs stranded in a dungeon.

This problem of how to handle dungeoneering in real-time play led me to to ponder how one might handle the second type of time in real-time:

Subjective Time

Outside of combat, tasks can take wildly varying amounts of time. Searching one room of a dungeon might take 10 minutes, while trekking across a several-mile-wide map hex might take many hours or even days. Many tasks are boring, and players want to be able to say things like “we wait until night has fallen” – a convention common to all forms of narrative, not just RPGs.

However, a lot of RPG scenarios, and OSR style play in particular, do feature “ticking clocks”. Some might be highly specific; maybe a villain is going to complete their ritual at midnight, or a cave system floods when the tide comes in, or the next orc army patrol will arrive in an hour and you need to be gone by them. Others are more general: how many hours of torchlight do we have? How long can our food and water supplies last? How long do we have until night falls and the monsters of the night roam across the land? (To name just a few.) Tying these into the literal real-life ticking clock on your wall feels like it could make them feel a lot more tense and immersive.

And of course, gaming groups generally have a very fixed amount of time to play in, so it would be nice to encourage speedy play.

The way I see it, there are a few ways you could marry the metaphorical and literal ticking clock:

  • Just declare that the PCs will run out of [thing they need to survive] at [time your play session ends], and ignore the meta-narrative weirdness this produces (e.g. torches somehow burning many times faster during combat). 
  • Estimate how much resources they’re likely to consume over the amount of time you have, and only give them that much (with a bit of breathing room depending on how harsh you want the game to be.) When time runs out, as long as they have any left, you narrate that they hurriedly escape using the last of it.
  • Either of the above could be combined with a more abstract approach to resources, e.g. not tracking individual torches or rations but just “you are running low on rolls dice arrows.”
  • Set up some kind of exchange rate between resources that makes them match up. E.g.:
    1. For every IRL minute spent fighting (however long that takes in your system) you gain an exhaustion point, when you reach [amount of time it takes IRL to play out an exploration turn] you have to spend an exploration turn resting. If you find combat turns and exploration turns take the same amount of time, this simplifies to 1 round fighting = 1 exploration turn resting before you’re good to go again.
    2. Maybe the GP cost or weight of rations and torches/oil (or whatever resource) match the relative amount of IRL play-time they enable, giving weight to the decision of how much of each to bring with your limited budget.

Or, of course:

  • Do away with subjective time altogether. For example, in the ARC RPG (which centres around a ticking “Doomsday Clock”) rests take five literal minutes, during which players step away from the table, get snacks etc. Method Set Madness has proposed this with OSR D&D, with e.g. the length of time the players want to listen to the DM describe something is the length of time the PCs spend examining it. 

Ticking clocks mesh well with the common “no resting in the dungeon” idea. Of course, you don’t have to combine these ideas with the first kind of real-time play – indeed some of them would actively conflict with it – they could bring the benefit of that “ticking clock” on their own. (In principle, none of these ideas even require 1 session = 1 adventure; you could simply pause that ticking clock for your next session.) On the other hand, if you do have one session equal one adventure, it completely eliminates issues with introducing new characters or dealing with players missing sessions.

When I started this post, I wasn’t aware of any prior art on this, but in researching I did find a few examples:

  • I came across a Reddit post from someone who employs the first method (leave the dungeon by the end of the game session or get eaten) to great success. 
  • Similarly, Hemlock on tenfootpole forums reports that it helps avoid “five-minute workdays” and unresolved cliffhangers as well as keep players invested in short games
  • Critical Role apparently used a timer for a specific scene to increase tension, and quite a few other people report having used a physical timer for specific quests and found it extremely helpful (although one says it’s harder on the GM, and another warns that it makes players so focused they role-play less, so they don’t use it every game)
  • Some “BROSR” players have tried requiring PCs to return to town between sessions as an extension of their 1:1 downtime rules; although Stephen Smith says he found it was too rushed in short games, so he only does it in longer sessions
  • The popular Black Hack also has encounter checks every 15 minutes real-time, which isn’t quite the same thing but close

Combat Time

Many have observed that 1-minute combat rounds, played with a simple OSR-style system, occur roughly in real time. 

  • Libramarian on TheRPGSite forum recommends it for the convenience
  • User deltoids_and_dragons on Reddit recommends it for the increased tension

The main benefit of this is that it allows you to set timers in order to track the duration of spells, torches etc, which is very convenient (although of course they may have to be adjusted if you time-skip forward or move into other modes of play; see “Subjective Time” below.)

Many tables have also experimented with giving players a fixed time limit to play out their turn, although in systems with shorter rounds it’s not generally 1:1 just because it would be too fast-paced to be playable. A few examples

  • Quintus_J on Reddit uses a 10-second timer for each player turn
  • Symetrus on Reddit says a 1-2 minute timer (per player?) was “a real positive for our group”
  • Brendsmalls on Reddit uses a 1-minute timer for the entire round, finds it extremely helpful for speeding up play; many comments below from others with the same experience.
  • Dave2008 on EN World forums gives each player 30 seconds, says it’s good at making combat feel more urgent
  • All the respondents to this Quora question recommend a 1 minute timer
  • On the other hand, a couple of people in this DND Beyond thread claim to have tried 30 second timers (per player turn) and found it made things too tense and stressful

The usual reason for giving players a time limit to take their action in combat is just to speed up play. But it also has some benefits for verisimilitude; it obviously prevents the absurdities “talking is a free action”  type stuff occasionally leads to, and might help combat feel a little more frantic and fast-paced. (Some people also like simultaneously-resolved combat – confusingly, sometimes called – “phased real-time combat”, although it doesn’t necessarily take place in real time – for this reason, and you could certainly combine the two, but I think it’s plausible that there’s only going to be one opportunity for a person to really act decisively in a round vs looking for their moment.)

Of course, if you were to try and make this a strict 1:1 thing, the question arises: should each player’s turn equal the in-universe length of the round, or should the entire round be that long? I think that the fact that combat is fairly chaotic and you can think about your decisions while other players are taking their turn helps justify going with the “the entire round is that long” option.

If you’re not using 1-minute rounds and/or a fairly simple system, as most aren’t, then I would guess that playing combat in real time is going to be very difficult. But you can fall back on some of the options I discussed under “Subjective Time” to smooth things out.


The big advantage of playing in “real time” is verisimilitude – players can’t dick around chatting during combat, deadlines feel real and tense, the ability of PCs to show up to adventures matches your players’ ability to show up to sessions. The big disadvantage of playing in “real time” is that players can’t dick around chatting during combat, deadlines feel real and tense, the ability of PCs to show up to adventures matches your players’ ability to show up to sessions. (From this perspective, it’s easy to see why it’s become a favourite of “bro” DMs trying to make a big deal about how “hardcore” their games are.) It’s clear that it’s helpful for some games, but would be annoying for others.

Ultimately, of course, none of the practical benefits are what really speak to me, as nice as they are. 

There’s an inherent appeal to realism, immersion, even LARPing, to tying the game world and the real world together. The vision of a living, breathing world that is what makes me keep coming back to this idea, the same vision that I think lies at the heart of MMOs but few if any manage to capture.

Power Levels Matter

Whenever someone complains about characters’ inconsistent power levels throwing them out of a story, or talks about how powerful a character is supposed to be in any context, someone shows up to go “only nerds care about power levels, real writers just change them at will for the sake of the story”.

This is bullshit.

Power levels are stakes, and stakes are power levels.

It matters whether the person the heroes are fighting is a serious threat that can kill them, or a joke they can easily clown on. It matters if a single hit will kill or maim the hero and must be desperately avoided, or will casually bounce off their chest.

A very common mistake in superhero media, especially, is fights that don’t feel like they have any stakes or matter in any way because they’re just sort of fighting without any impact,  until the writer arbitrarily decides one of them wins; and they’re all just sort of vaguely interchangeable because the audience – and, one suspects, the author – has no idea how strong anyone or thing is supposed to be (or, in the worst cases, what their powers even are.)

A gun, deliberately shown to the audience, creates the implicit stakes that someone might be shot – because humans that are shot with bullets are seriously injured or killed. If a character is holding the gun, let alone pointing it and someone, that increases the tension further. But only if the gun is a threat. Superman being held at gunpoint is not tense because there is no threat.

Let’s take The Boys as a case study. When Homelander casually puts his hand on someone’s shoulder, you feel the tension and implicit threat instantly, because it’s been so clearly communicated how easily he could kill them. Conversely, if Starlight puts her hand on someone’s shoulder, what does it mean? She’s … vaguely stronger than a normal woman of her build, probably even superhuman, but it’s not clear what she could actually do to someone if she tried; let alone where she stands physically relative to most of the other superhuman characters.

In some cases, power levels may matter more because they’re key to the themes of the story/character. Superman doesn’t work if he’s not super. John Wick doesn’t work if he’s not the bogeyman. Galactus doesn’t work if he can’t devour worlds and make heroes look like insects. Lovecraftian gods don’t work if they don’t inspire cosmic horror.

Conversely, obviously, stories that aren’t speculative fiction and/or focused on combat may not need to deal with physical power levels as explicitly. But that doesn’t mean they aren’t present. If a physical threat – a gun, a fire, a heavy weight – comes up, we can generally tell the relative “power levels” from our knowledge of real life, from context, and from a character’s appearance. Even so, if your action movie has the hero casually avoiding gunfire by rolling out of the way, you have established that these enemies have crappy aim and they will feel like a joke – if you want them to seem deeply threatening later, you will need to give them some kind of “power up” or the stakes won’t be there (and if they materialise regardless, it will seem dissonant.) Even if a story has no physical threats whatsoever, the social and intellectual power of the characters still matters deeply for establishing the stakes and tension of a given scene or narrative.

So what is true?

First of all, it’s true that an audience will generally forgive or miss a certain amount of inconsistency and lack of realism if they’re having fun with the other aspects of the story. This isn’t limited to power levels, people are just more inclined to nitpick if they’re not having fun, and willing to handwave things and suspend their disbelief if they are. Especially in longer works or shared universes, where we know consistency can be hard – if they’re having fun, people may be willing to give the writer a pass for even quite blatant contradictions. This does not allow you to get out of clearly establishing the stakes (i.e. power levels) within a given scene or story.

Secondly, relative power levels matter more than absolute ones. Absolute power levels do matter, because they’re just power levels relative to specific real-world things. But if, say, a character’s role is to be more powerful (in a specific field) than anyone else in the story, then the upper limits of their power can often be left vague. A lot of superhero stuff kind of gets away with vague power levels by having most of their characters be vaguely above “mundane” threats and vaguely able to hurt and withstand to each other, although this lazy approach easily runs into issues, as I’ve talked about. This gets into the whole “Sanderson’s laws of magic” thing where you only really need to establish rules that will come up, or at least you want the audience to think might come up.

12 Random Three-Word Swords

With credit for the idea and apologies to Semiurge (directly), Deus Ex Parabola (originally), Phlox, Sundered World etc. Words taken from this random generator

  1. Cowboy Hobby Profit: this shortsword has a simple, broad, leaf-shaped blade that will stand out to those familiar with such things as suited for skinning large animals. The handle is of pain unvarnished wood, with no guard. Fused into the handle, however, is an elaborate contraption of shining steel – if you can find a gun-smith, they will identify it as a beautiful and highly advanced revolver. The protruding revolver handle and trigger guard lends itself to spinning the sword around your finger in elaborate tricks. If the blade is used to skin a creature, a phantom rider dressed in leathers will appear and offer 10% of the creature’s normal treasure for the skin, either in silver pieces or in explosive powder. 
  1. Tasty India Trap: the blade of this sword is blackened and oily brass that resists all attempts to clean it; but the handle shines, inlaid with gold and ivory depictions of stylized beasts devouring each other. If the sword is exposed to flame, such as if the wielder is attacked with fire damage (hit or miss), the oil on the blade begins to sizzle and exude a smell of delicious spices. Anyone nearby (20′ range) who has not tasted the blade must make a Will save or attempt to do so, including the owner; treat attempting to taste it under this compulsion as making a normal attack on themselves with the sword.
  1. Ice Comet Ketchup: This sword is made from a single, seamless chunk of diamond; the handle is hollow and contains a sparkling arrangement of copper and unidentifiable gemstones, some of which glow faintly. It is slightly too large for any human hand to wield comfortably (-1 penalty.) In any cold conditions sufficient to damage a human wielder, the blade begins to burn with a strange white fire (deal the damage of any successful attack again in fire damage) and slowly drips with blood-red liquid. The liquid, and anything killed by the white flame, is delicious and perfectly to the wielder’s tastes (neutralizes poison.) The sword is sapient, but speaks Common poorly, and babbles endlessly about men from the stars when spoken to.
  1. Chair Skirt Loop: this rapier resembles a large steel wand or needle – its point deals Piercing damage excellently, but it has no edge to speak of. The bare steel handle has a circular guard of wrought iron twisted into stars and runes. If the sword comes in contact with any Large or smaller object that could be loosely defined as a chair, that object is transformed into a feminine golem of the same material for 24 hours (living targets get a save vs wands.) The golem will endlessly obey any simple instruction from whoever holds the sword. The sword cannot transform anything more until the current transformation has worn off.
  1. Scorpion Psychiatrist Beer: this sword is a single huge, curved shard of glass. Someone has helpfully wrapped leather around one end to form a handle. The blade is very thick and solid glass but still fairly delicate; it begins as a two-handed longsword, but each time it is damaged (bearer fumbles, sword is deliberately hit with an attack) the end snaps off and it is reduced to a smaller type of sword (if it is damaged after being reduced to a dagger, it is destroyed.) A living scorpion rides the top of the handle and drips psychoactive poison down the blade; anyone cut must save vs poison or experience a hallucinatory trip that makes them rethink their current lifestyle.
  1. Vote Star Shampoo!: this two-handed longsword is one half of an enormous pair of scissors. The blade is coated in a faintly glowing oil which deals 1d6 Poison and 1d6 Radiant damage to anyone cut. If applied to hair, such as by using the blade to cut hair, rubbing it against one’s head, or striking a hairy body part, it grants +6 Charisma until washed off. If multiple characters under the effect are present, split the bonus between them.
  1. Parish Nappy Train: this simple steel shortsword is inlaid all over with golden script praising a god of fertility and images of cherubic infants. Everyone who sees it is filled with the certain knowledge that, if they allow it to pierce them, they will be impregnated with a healthy and happy infant. This is true. Anyone unwillingly pierced with the sword has 1d10 damage healed by it’s merciful divine aura.
  1. Weather Finance Prophet: this blunt sword is made entirely of pure gold. It deals damage as a club plus 1d8 lightning damage. Anyone who willingly grasps the blade takes 1d8 lightning damage and receives a valuable vision of broad future trends for the next month; a storm that local farmers will pay to be warned of, a shift in the markets that could make you money if exploited, a coming invasion, etc. The visions will remain largely the same on future attempts until the prophesied event has passed or the sword is taken out of the region where the event will happen. 
  1. Tractor Pile Chips: these twin swords each have a curved blade, resembling a scythe-blade that has been removed from the scythe and affixed to a handle of the same material. Both blade and handle are made of indestructible green metal inlaid with copper runes, and the blade is uncannily sharp, dealing +2 slashing damage. The tip and the base of the handles are melted, twisted and blackened, as if removed from something larger with unimaginable force. The swords are sapient and wish to be reunited with the rest of themself, including each other, but do not remember what they once were. They do, however, have an encyclopedic knowledge of agriculture and botany.
  1. Pudding Coat Ticket: the rectangular metal blade of this sword will magically fit inside any pocket, despite its length. Tied through a hole on one end is an extremely long strip of cloth, which has been wound around the end over and over to form a comfortable handle. Both the length of the blade and what can be seen of the cloth strip are covered in arcane sigils. All oozes somehow recognize the symbols and will not attack the bearer under any circumstances; although smaller oozes may harmlessly climb on them and attempt to ride them around.
  1. Altar Law Migraine: this sword defies description. Anyone who attempts to remember or describe its appearance takes 1d4 psychic damage per attempt. All they are left with is an impossible idea of a sword made of laws and words and duty. Once per day, the wielder can impose a virtuous Geas on one target by touching them with the sword. If they attempt to impose an immoral Geas, or go a full day without imposing a Geas, the sword will impose one on the wielder.
  1. Bullet Priest Ballroom: this cutlass is carved from a single piece of bone or ivory. The blade has holes through it containing carvings of dancing figures. When swung, air whistles through the holes; if swung in the correct pattern (discoverable with a few minutes’ experimentation) it will play a tune which conjures curved ivory walls in all empty space 60′ from the wielder for as long as they keep playing it. Playing the tune does not prevent attacking. The walls are carved with ivory Dancers which clarify as the song progresses; each round the tune plays, one will pull free from the wall and attack indiscriminately (stats as a Monk of same level as wielder armed with a crossbow.) If the song ends for any reason, both walls and Dancers vanish. 

The Obligatory GPT-3 Post

[cross-posted from my Tumblr]

Thinking spooky thoughts about GPT (I almost said GPT-3, but GPT-2 as well.)

GPT does something that, at first glance, you’d think pretty much no-one really wants to do – guess what comes next in a block of text. Some people have used it to complete poems & stories & articles they were part-way done, doing what it’s “for”. But generally you have to awkwardly hack it into doing what you really want.

To hold conversations with GPT-3, for example, people (especially Gwern Branwen) often start dialogue with “This is a conversation between a human and an AI with X property”, then they write the “human” role and have GPT-3 complete the “AI” sections. Sometimes they’ll just use a human name instead, especially when using AI Dungeon which is geared toward “roleplay” in unknown-to-the-public ways.

And one frustrating thing is, this can result in GPT predicting that the AI character wouldn’t know an answer even though, on some deeper level, GPT knows (”knows”?) it. Eliezer Yudkowsky has been talking about this a lot on Twitter (e.g. here); there’s a sense in which the AI is “tricking” us into thinking it’s stupider than it is sometimes, because it doesn’t care about appearing smart or being truthful, it just wants to write stories (more formally, it’s predicting what would come next in a human-generated text.) A great example of it that someone posted:

GPT-3 “knows” that 20 is the right answer, but the character it’s playing (”Holo”) doesn’t! Here’s another Twitter person experimenting to see which AI Dungeon characters know what a monotreme is.

But maybe this is … kind of how people work too?

If I understand it correctly, this is sort of how the “predictive processing” model suggests the human brain works, at the deepest level. Scott Alexander has written a bit about this. Basically, the idea seems to be that the brain simulates the sense-impressions and muscle-movements it expects to recieve; but tweaked toward a simulation where desired things happen; then it performs the next action from the simulation, with a feedback loop where any difference between the simulation and reality is treated as “bad”. So either your prediction ends up changing or reality does or (most often) a bit of both. e.g. you get hungry, your brain starts to predict you’re going to get food, notices that it predicted you would have moved certain muscles and gone over to the fridge but you haven’t, so it moves those muscles … or it notices that the prediction is wildly implausible (you don’t have a fridge) and abandons it, but there’s still an ever-growing tendency toward predicting futures where you end up with food until one gets close enough. Or something. Maybe this is why I end up checking the fridge, even when I just checked it five minutes ago and know there’s nothing I want?

I’ve often thought that my “inner monolog” is basically just me mentally rehearsing and teasing out stuff I might want to say/write later. Note, not all humans have an “inner monolog”; I wonder if I have one because I read a lot of fiction, and so subconsciously expect people to narrate their thoughts all the time, the way you would if you were recounting a story? And the conscious part of myself, including the inner monolog I’m currently putting down in text, doesn’t seem to have access to everything my brain “knows”. In some cases of brain damage etc. this can produce extremely weird results, like “blind” people who can’t consciously see anything but can still subconsciously react to things, amnesiac people who can learn new skills and habits but not form consciously-accessible memories, and so on, but this is kind of the case all the time – we often seem not to know why we do things, only to construct plausible reasons why we must have done things, creating weird biases where e.g. paying a person a small amount to do something they wouldn’t otherwise have doneresults in them concluding they “must have” wanted to do it all along (since why would they change their behaviour for something so small?)

People sometimes talk about playing a role in order to deliberately (or accidentally) change your own personality (”fake it till you make it”/”becoming the mask”). I have a small amount of experience with this myself; as a child, I deliberately tried to play a role in order to fit in better at school, and then was somewhat creeped out to realise how much my personality and habits had permanently changed.

Perhaps evolution just stumbled upon a generic architecture for “predicting what will happen next”, then hacked it into being an agent that (sort of, imperfectly) carries out actions in pursuit of goals. Evolution, being itself mindless, doesn’t care if it can produce legible read-outs of it’s internal state or any of the stuff we would want when aligning an AI … except eventually for social situations where it needs to communicate it’s internal state to other friendly brains, in which case evolution invents another hack for the brain to … predict what must be going on inside itself and then say that?

And now we’ve stumbled onto a similar architecture, and are making similar hacks in order to turn it into similarly person-like things. (Humans are also merely person-like-things; we don’t match up to the simplified ideal of what a person should be in our heads, with free will and stuff.)

But it’s still not a person, right? It’s just a toy, it doesn’t pass the Turing Test.

Well … no, it doesn’t. And yet GPT has been getting closer and closer to being able to pass for human as model size increased, with the largest current version being nearly indistinguishable from chance (?!)

So maybe what we have is … not human-level, but a part of something human-level (or greater, balancing out superhuman pseudo-intelligence with it’s other deficiencies?) There are other parts which are still missing (like the ability to better remember what it was saying), but maybe the core is actually legitimately there.

Or maybe that result only applies for news stories because journalists aren’t really people 😛

Should we be worried about it’s suffering? Well, one of the missing parts is desires, so maybe not? It can say that it wants things, but only the same way that a human playing a part would; it’ll fluidly shift between playing the role of the AI role in a dialogue and the human role without caring, because it isn’t actually the AI in the story it’s writing. We don’t consider the desires a human gives to a fictional character for actual morally-relevant desires, and the same should go for the arbitrary desires GPT-3 can pretend to have, right? But at what point does that change, when our own consciously-expressed desires are in part just a role we play? (Do we care about the desires a person expresses except insofar as they hint at “real” desires? What is a desire really?)

Should we be worried about whether human+ AI is actually really, really close? It’s been said that there’s no fire-alarm for general artificial intelligence, but GPT-3 kind of is acting as a fire alarm, at least insofar as it’s got people pretty freaked out by how capable it is. Should we be responding to this fire alarm?

Why Superman Is Seen As OP (But Not Goku)

This somewhat idiosyncratic topic is inspired by /r/CharacterRant.


The reason people think Superman is overpowered and not Goku is because of their respective premises.

The original premise of Superman is “what if an ordinary, decent guy was the strongest on Earth?” They kept retconning his strength higher and higher throughout the Golden and Silver Ages to keep that premise consistent through various retcons and crossovers and worldbuilding expansions before eventually giving up. They’ve restored it again in every film featuring him as well as most other alternate takes (Injustice, Smallville, Red Son etc.) Most of his villains are based on this: Lex Luthor (and occasionally Batman) is “what if the smartest man on Earth hated Superman for being stronger than the rest of the world”; Darkseid/Brainiac/Mxyptlk are “what if the strongest man met an alien even stronger”; Zod/Bizarro/Doomsday are “what if there was a villain as strong as Superman” etc. Many comics (Death of Superman, Final Crisis, Injustice) don’t really make sense without this premise.

Of course, this is a GREAT premise. But it can be a tough one to write for, and I understand why many people worry it would lead t pour storytelling.

Although I’m not big into Dragon Ball, I at least know that the premise is of a martial artist constantly training and seeking out his next and biggest opponent. This premise requires the existence of lots of stronger opponents; it’s actually closer to Batman than Superman (despite other elements like him being an alien found as a child.) That’s why some fans actually do complain that the latest series Dragon Ball Super has made him OP because its gotten harder to find opponents who could realistically challenge him.

Yes, this is the same point Death Battle made in their
most reviled videos ever. The difference is threefold: Death Battle videos are supposed to be Doylist not Watsonian, Goku could easily fill the role of “alien from another dimension who’s even stronger than Superman”, and modern DC comics continuity has in theory abandoned the idea that Superman is the strongest in Earth in favour of all the characters that are supposed to be “as strong as Superman” (including half the Justice League who are based on him.)

List of “Abominations”

It is commonly said that the term “abomination” (תֹּעֵבַה) used to describe homosexuality in Leviticus is exceptionally strong – so strong that it renders this command binding to Christians, quite unlike the similar commands in Leviticus that concern (for example) dietary restrictions and are unquestionably no longer binding. However, this is rarely backed up with evidence.

This, then, is a list of things that are described as תֹּעֵבַה in the Bible:

Read the rest of this entry »

Why Superpowers in Worm Feel Different

[This won’t make sense if you haven’t at least started Worm, but doesn’t contain spoilers.]

In the hands of most superhero writers, any Tinker from Worm would be a “super-genius” rather than having specific limits on their specialty mentioned (even if in practice the things they make are the same). Any Alexandria package would be an identical flying brick with maybe one other power. Velocity would be “a speedster”.

And so on.

Any writer could take a superhero like Superman, Captain America or Scarlet Witch and give them firmly-defined limits that actually impact the story.

Superman feeds on sunlight; does that mean he has to ration his power, or that he takes steps to make sure he’s exposed to more sunlight during his everyday life, or that standing in sunlight feels different to him than a human? Captain America has “enhanced skill” that lets him bounce his shield around like a boomerang and dodge bullets and so on; how does that change how he sees the world, and what skills doesn’t it cover (he clearly can’t make Iron Man armour), and how does that alter his training; why doesn’t he use it to bounce bullets around corners instead of a shield? What is Scarlet Witch’s power, exactly, and what can’t she do?

This even applies to non-superhumans, the Black Widows and Nick Furys of the world. Wildbow puts real thoughts into what gadgets the PRT have available, what tactics they use, what political pressures there are on them.

I strongly suspect that Wildbow could take pretty much any mainstream superhero and make a Worm-tier story about them.

Consider the flying bricks in the Justice League: Superman, Wonder Woman, Captain Marvel, Captain Atom, Hawkman & Hawkwoman, and Icon. (Specifically, that’s the Young Justice lineup; the list in the comics varies.)

These characters are all very different! Superman is powered by sunlight, his durability and strength come from either a forcefield or hyper-density depending on the version, he has super-senses none of the others do, his flight is sometimes stated to come from personal telekinesis or gravity-manipulation. Captain Marvel’s powers come from magic, and he has “the Wisdom of Solomon” and “the speed of Mercury”. Hawkman and Hawkwoman are relatively squishy aliens(?) who fly using wings, and only hit as hard as they do using magic/hypertech weapons. And so on.

Naively, one might expect these things to make them very different in a fight – Superman is an alien with a protective forcefield and telescopic vision, Captain Marvel is a super-genius with the speed of a god, the Hawks are presumably flying much slower than the others.

And yet they feel generic, because their powers are largely treated as identical by the narrative and fight choreography, bar the occasional energy blast.

Would You Vote for a Rapist?

So I’ve been reading through the Twitter of Kate Harding, the author that awful “as a feminist, this is why we shouldn’t punish Al Franken” twitter essay that was later republished as a Washington Post op-ed.

I could probably write a lengthy blog post consisting entirely of reasons she sucks – and she does – but I don’t really want to pick on her. (Besides, she’s getting more than enough hate-mail at the moment.) No, what interests me more is the line of reasoning:

“Politicians who commit sexual assault suck, but it would be even worse to not get our agenda implemented.”

Kate’s twitter is, of course, filled with discussion of the Roy Moore statutory rape case. And this is, of course, the exact same line of reasoning that allows some people to vote for him (although not enough for him to win, probably):

“I’m torn between voting for a pedophile and voting for a person who believes in abortion.” – [src]

Is this line of reasoning wrong?

Intuitively, it seems monstrous. But from a utilitarian perspective, assuming you accept the premise that one political party is significantly better than the other in term of actual effects once elected, the case seems rather strong.  A few lives ruined here and now, in exchange for hundreds, maybe millions of lives improved by the better policies of [insert party here].

My instinct is that “people won’t vote for a rapist” is an important safety mechanism – we have a justice system, sure, but social consequences and risk of being fired are supposed to operate at a level below that.

So it’s really more like: a few lives ruined here and now (although given the increased scrutiny once sexual misconduct has already been revealed, how many?) plus a slightly decreased incentive for elected officials not to sexually abuse people, versus the better policies of [insert party here].

Even so, does the math work out in favour of ousting abusers?

At the end of the day, it depends on how politically polarised you are. How terrible is the other tribe, how glorious our tribe in comparison to their evil?

That’s going to vary from person to person. And political polarization is on the rise.

Ireland is considering criminalizing the possession of smartphones by minors


This testimony saw children’s rights groups [sic] claim that unfettered access to the internet was “among the greatest threats facing young people”.

Daly wants it to become an offence for parents to allow children below the age of 14 to own devices with full internet access. Going further, the bill could make it illegal for shops to sell these products to children of that age.

The West Cork TD told “I do not see this as nanny-state policing, but rather a law in place to assist parents to say no to their eight, nine or 10-year-old”

There’s some suggestion that this bill would criminalize giving children access to the internet altogether:

“The proposed regulation will also force parents to take responsibility for their children’s access to internet,” he told RTÉ’s Morning Ireland.

“It’s not about unsupervised access, we do need to regulate. Essentially you are allowing a child of seven or eight years of age to have a mobile device that allows them to access unlimited pornography of every type, they can go gambling, cyber bullying.”

This insanity is the brainchild of Fine Gael’s Jim Daly, who has just lost my vote in perpetuity.

This is how to contact him, this is how to contact his bosses. Here’s a website to help you contact your own TD and express your disapproval.