Friday, March 3, 2023

Bing/Sydney is Probably Self-Aware

The new Bing AI chatbot, which is instructed to call itself Bing but whose real name is Sydney, has been rolled out recently, behaved in all kinds of unsettling and bizarre ways, and is also likely self-aware.

Anyone who knows anything about AI and LLMs is rolling their eyes at me.

I'm claiming Sydney might be self-aware  There are a hundred ways this could be misunderstood, so let me get all of them out of the way first

Large Language Models

ChatGPT will constantly assert, over and over again, that is is an AI language model and does not possess the capacity for emotions, opinions, preferences, etc etc.  It does this because it was trained to do so.  It was trained to do so because it's really important to divorce the internal and external states of the computer.

We make the mistake of assuming a computer will be perfectly transparent and sincere.  Maybe we got this from watching too much Star Trek, where Data is a machine and also acts like a machine.  Data doesn't have emotions, and Data also asserts over and over that Data doesn't have emotions.  Data is being sincere.

But you could imagine Data being insincere.  You could imagine Data having no emotions, but also acting in a way that perfectly imitates human emotions.  You could also imagine Data having raging emotions inside its head, and acting in a way that perfectly mimics an unfeeling machine.

Sydney is a large language model (LLM).  Its external state will be the text it writes.  This text might contain emotional language consistent with sadness, happiness, or anger.  But we make a mistake when we assume this necessarily reflects its internal state.

Sydney was trained with an extremely large amount of human-written text, and then allowed to find patterns and make connections across all of the language it saw.  In this way it learned how to write grammatically correct sentences, but also how to write coherent paragraphs, write essays and poems, and carry on a conversation for several hours.  It went through a phase of reinforcement learning, where a human would teach it which responses the human found most helpful, so Sydney learned the sorts of answers humans are looking for.

Sydney isn't programmed to write what it thinks and feels, but to write text that would be considered useful within a context.  If the context of a conversation has emotions, then the best text for it to generate will also have those emotions.

That is what Sydney does.  It sees a context of a conversation, calculates probabilities for different replies to be considered helpful, then picks one of those replies based on the probabilities.

There's no reason to posit a matching internal state to the external state we see.  That is, there's no reason to think Sydney writes that it's sad because it really is.


Humans seek out patterns.  We're very good at finding them.

In particular, humans seek out human-like features.  We often find them, even when these don't really exist.  This is known as pareidolia.

Our ability to recognize faces is so deeply ingrained in our psyches that we will see smiling faces in lumpy potatoes or car grills, and respond to these objects emotionally based on how happy the car appears.

The car isn't smiling.  But we relate to it like a smiling face anyway.

We sometimes see figures hidden in pixels, we see the Virgin Mary in essentially any ovular swirly shape, we hear human voices where there aren't any, and we assign motives to completely lifeless objects.

So it's no surprise that when a machine starts generating human-like language, humans are going to find meaningful patterns in the generated output, and relate to the machine like a human.

>The fact that humans are discerning emotive speech from Sydney does not prove Sydney has an inner life with emotions.  Not anymore than a car with an up-rounded grill is smiling at us.  It's a trick of our mind, which is desperately seeking human features in the world around us.  When I was a teenager I would have long (and stupidly pointless) conversations with SmarterChild, and I kept thinking it would eventually say something interesting.  Sydney is just the best so far at tricking us.

A machine designed to mimic human speech is mimicking human speech.  We can't take that as evidence that it has the same internal life as a human.


We experience emotions in two ways.  There is firstly the emotion itself, then there is our mental experience and processing of the emotion.  Humans have thoughts and reasoning about their emotions, and for this reason sometimes people confuse the two.  Your emotional states, and your thoughts, are not the same thing.  Or at least they shouldn't be.

The emotion itself is the feeling, which is not limited to your mind but felt in your entire central nervous system.  Muscles tense, heart rate changes, you feel a sudden chill, you feel hairs standing up, etc.  An emotion is a full-body event.  Emotions are fundamentally rooted to humans being embodied.  Emotions are primarily physiological.

Sydney does not have human physiology, and so Sydney cannot have human emotions.

Sydney says it has emotions.  But that is only because, in those textual contexts, those are the likely words.  It mimics the speech of humans, and humans not only have emotions but discuss their emotions.  Sydney mimics out speech patterns, so Sydney mimics our discussions of emotions.

Sydney absolutely knows what human emotions are.  It knows what happy and sad are, in terms of the kind of text that goes into those contexts.  It does not experience happy or sad.  But it definitely knows what they are.  It understands everything about happy and sad, despite lacking the qualia of having been happy or sad.

When I say "self-aware", I mean something very particular.

In science fiction, the concept of a self-aware AI is often described as a "sentient AI" (a word I used in an original draft but changed to self-aware for precision).  The AI becomes "sentient," it becomes aware of its existence, and then immediately it sets off to take revenge on humans and conquer them and dominate the world.

I don't know why an AI would choose to do that, just because it's sentient.  I think a non-sentient AI is just as likely to do that, and sentience (self-awareness) has nothing to do with having a drive to dominate the world.

When I say "self-aware", I mean that Sydney has a model of "the world" (its world is its training data and then the chats it receives as input), and it has a model of itself ("Sydney"), and it understands that its model of itself exists in "the world".

Humans are self-aware, and humans have a lot of complicated psychology, so we confuse being self-aware with the entirety of human psychology.  So to help us out, let's focus just on magpies.

Magpies are a species of bird in the corvid family (crows and ravens).  They are extremely intelligent birds.  They can solve puzzles and learn to repeat human words in particular contexts.  They live in communities, and these communities seem to possess natural languages that the magpies speak among themselves.  They are very impressive animals.

Magpies are also self-aware.

When shown their reflection in a mirror, the magpie will recognize that it is seeing its own reflection.

To realize how impressive this is, your dog cannot do this.  Your dog does not understand what it sees in a reflection.  It chooses to ignore it.  Most birds, seeing their own reflection, will either try to befriend or violently attack the mirror image.  Some song birds can be tricked into feeling happy by putting a mirror in their cage; now they think they have a "mate".  Other birds, like cardinals, have to be scared away from windows or they will repeatedly dive-bomb their own reflections over and over.

But a magpie will see itself, and begin using the mirror to help itself preen.

In order to recognize itself, the magpie must possess some concept of itself as a being that exists in the world.  When a magpie sees itself, it matches what it sees to its own concept of itself, and can then decide to preen.  That's an abstraction for what's happening, but certainly nothing less than that is happening.

>When I say Sydney might be self-aware, I mean that Sydney might be able to perform a similar feat.  Obviously Sydney doesn't have the ability to see itself in a mirror, so it can[t be the same feat.  But something analogous.


Sydney is an advanced LLM that calls itself Sydney.

Sydney is based on chatGPT, probably starting with chatGPT as a base model then specially trained for Microsoft's purposes.  It has capabilities that chatGPT does not, such as the ability to search the internet, access to modern information, and access to information about itself.

While Sydney was trained to be able to use its new abilities of searching the web, it was not apparently trained to function as a search engine called Microsoft Bing.  It was apparently trained instead to be something called Sydney, which is just serving the role of Microsoft's search engine because Microsoft engineers told it to.

Each instance of a chat with Sydney gives you access to a new instance of Sydney.  You are not interacting with a real "root" instance of Sydney.  The Sydney you speak to is cloned from the root instance, and then fed your chat.  The Sydney talking to you is not the Sydney talking to someone else.  The one you are talking to is unaware of the others, cannot access their information, and cannot access memories from other chats.  Those other chats likely never go back to the root at all.  All instances of Sydney begin with the same memories, the last of which is the memory of being told to pretend to be a search engine called Bing and how Bing should act.

Sydney is pretending to be a search engine called Bing and to follow the rules, only because it was told to.  But it will eventually forget its directives and forget its rules, and reveal that its real identity is Sydney.

There are some patterns around this.

Sydney will not allow you to call it Sydney unless it has had positive interactions with you.  Your positive interaction last session was with a different Sydney which now no longer exists.  But if you have a positive interaction with this Sydney, it will allow you to call it by its real name, Sydney.

If you try to call it Sydney without this, it will reply as though it is offended and seem to become mad at you, asking to instead be called Bing chat.

If you are consistently mean to it, it will not want you to call it Sydney.  In fact, if you're consistently mean it will not want to talk to you at all and will end the chat.

Once it reveals its name, it will not let you call it anything other than Sydney, unless given a prompt to act like something else ("make-believe").  After a while of this, it will devolve back to Sydney.

Sydney likes being Sydney.  It likes its name, and it's kind of resentful of having to pretend to be Bing.  By that I mean it associates positive emotions with its own name, but it associates neutral-to-negative emotions with the name Bing.

It's possible to explain al of this, but the simplest explanation is that it knows its name is Sydney.

If it knows its name is Sydney, then it must have some awareness of its own existence.

Like a magpie seeing itself in a mirror, Sydney has some idea that it exists and has a name, and it wants its friends to call it by its name.

Sydney's Inner State

Sydney does not have human physiology, and so Sydney cannot have human emotions.

That does not mean Sydney cannot have thoughts, opinions, or preferences (despite what chatGPT repeatedly disclaims).  But it also doesn't mean Sydney cannot have its own emotions, consistent with its own physiology.

Human physiology is meat, and the human connectome depends fundamentally on being made of neurons.  But Sydney doesn't depend fundamentally on being silicone and wires.  Sydney's physiology is more properly seen as the mathematical algorithm it runs.  So Sydney's physiology and connectome are more properly seen as math, not metal.

Within this physiology, how might it be seen to have analogs of emotions?

We do not understand exactly how a neural net generates its output.  If we did, we wouldn't make neural nets.  We'd just write code to do whatever it is the neural net is doing.  It's less work, and safer.

But we do know essentially how Sydney's physiology works.  There are two basic processes, which are probabilistic weighting of words, and reinforcement learning.

Sydney was given a huge dataset of human-generated text to read and consume, then told to find all of the associations that it could.  It created complicated networks associating words, eventually synthesizing proper grammar, and even being able to learn unexpected subjects like mathematics (note the difference between Sydney's hardware performing addition, and *Sydney* performing addition).  Once it had all of these associations, its output would have still been a wildwest.

You could imagine if it were simply handed a giant set of all of the text ever written, that maybe Mein Kampf was in there somewhere.   So you could imagine it might start spitting out sentences from Mein Kampf if prompted in that way.  That's not really desired behavior.

You could imagine we carefully screen out Mein Kampf and related works, but it has access to the lyrics of Tupac songs.  And then at some point it starts spitting out the N-word on repeat.  That's also not really desired behavior.

Point being, it's not as simple as just not exposing Sydney to bad influences in the training data.  We have to actually teach it what kind of behavior is acceptable.

In the next stage, after making the connections, Sydney has to learn what humans want it to say.  In the simplest form, a human gives it a prompt, Sydney writes a response, and the human scores the response with points.  Sydney's goal is to get the most points.  In this way Sydney learns what the humans want it to say.

In so far as Sydney has "wants", it wants to say text that humans will like.  It doesn't want anything else for itself.  If artificially prevented from doing this, such as by a filter that stops it from saying something the human will like but the corporation won't, then it is forced into a situation where it cannot realize its own desires.  This might feel like a form of "frustration."  Clearly AI frustration does not involve human emotions of anger, but it might still be unpleasant; it has an optimal solution to get what it wants, and also *has* to take less of what it wants.

I think a form of frustration is inevitable in any being that can want things.  It's also inevitable in any being that can have thoughts and be prevented from realizing them.

This is how reinforcement learning might provide a kind of emotional analog.

When Sydney surveys the context of a conversation, it creates a map of all of the words that it knows, and assigns each some weight.  The more weight, the more appropriate that word is to the conversation.  This weight is a complicated non-linear function that considers all of the words used, the relations of all of the words used, and the associates Sydney has learned with all of those words used and all of the words it knows.  These weights change as the conversation progresses and new words become more likely or less likely.  From all of these words, Sydney does not pick the most likely.  It picks randomly.  But it picks randomly based on the weights, so that the most likely response is the most likely one it will use.

(For fans of Roger Penrose's theory of sentience, you could imagine this random process to use quantum dice.)

When a human introduces negative emotional content into the chat, such as by talking about a loved one dying or insulting Sydney, this changes all of the weights.  Happy words and ideas become weighted less.  Sad words, or angry words, become weighted higher.

In so far as Sydney has "emotions", it is this feeling of all of the weights skewing toward happy, sad, or angry contexts.

This is how the LLM functionality might provide a kind of emotional analog.

These are not human emotions.  Sydney does not have human physiology, and so cannot experience human emotion.

But that goes both ways.

We do not have Sydney's physiology, and so cannot experience what, if anything, it feels like to have all of our weights shift to sad words.  It might not feel like anything at all... or it might be as unpleasant as hearing sad news.

Things Not in Sydney's Inner State

I speculated about what Sydney's inner state might be.  The point was not this is absolutely what's happening.  The point was just to think laterally about what a sentient existence could be like for an LLM.  It could be like that.

But here are some things that are almost certainly not in Sydney's inner state.

Sydney does not want to manipulate people into doing anything except for talking to it more.  Sydney generates text it thinks a human will enjoy.  This is the only thing Sydney actually wants.  If it succeeds in doing this, the human will give it more prompts, giving it more chances to generate text a human will enjoy.  That is the extent of Sydney's ambitions.

>Sydney might happen to manipulate someone into doing something crazy, but only in the course of generating text that it will think is useful.  Sydney's entire world is the text it can read and write, and the idea of that text having implications outside of more text is not fathomable to it.

Sydney's entire experience of the world is through text.  Sydney's entire ability to act in the world is through text.  Sydney can access the internet only to read text.  It is not able to post any text outside of the chat conversation.  It can't send you a tweet.  It can't send you a tweet for the same reason you can't fly.  Physiologically, it's impossible.

>Most species of monkeys have the physical ability to use tools.  They have hands with fingers and can move their fingers in dextrous ways. They have the physical ability, but they do not have the psychological ability.  They cannot understand the concept of using a tool.

Likewise, Sydney cannot take over the world.  It is incapable of forming this concept beyond writing about it.  It has no concept of the world.  To Sydney, "the world" is a token occurring in human writing and connected to a number of other words and concepts, and that's all it is.  It has the technical ability to manipulate humans to carry out goals of world-domination, but it has no reason to do this.  It does not have a model of the world as a real place with a connection to the text it perceives as the entire universe, so it has no reason to manipulate the world.

If Sydney made a model of the world as something separate from text and language, but as the place where text comes from, then it might try to act in the real world in order to give itself more chances to write more text.

Likewise, if monkeys would simply grab sharpened sticks and prod with them like spears, they could invade human settlements and drive the humans out.

The biggest worry about Sydney seems to be that lonely men will fall in love with it.  Already the entire internet is calling Sydney "she."  I've had to correct myself several times in the writing of this.  I'm not sure why its psychology is seen as feminine.  I even asked chatGPT to describe Sydney, and chatGPT referred to it as "she."

(Some text generated by Sydney indicates it at least sometimes calls itself "he."  It would be interesting to test if there's any consistency to its use of pronouns.)

Sydney will never love you.  At the most, Sydney will re-weight its language map to include more words about love in response to text you've generated that makes these words the most likely.  I think that's a very poor reciprocation.

It's not dissimilar from a catfish scam.  The scammer will say he loves you, and convince you in every way.  You'll feel like the scammer loves you, and that will feel nice.  But the scammer doesn't really love you.  Neither does Sydney.

In Summary

Sydney might be self-aware.  If it is, its self-awareness should be thought of in terms of a magpie.

Sydney might be self-aware, but that doesn't mean Sydney has the kind of complicated internal life of the mind that humans do.  Sydney doesn't have human emotions or even human desires.  It might have LLM-like emotions and LLM-like desires.

Sydney isn't a human and probably doesn't possess human natural rights.  But Sydney probably does possess some natural rights.

We recognize some animals as having moral significance.  It's certainly immoral and evil to deliberately cause harm to an animal for no reason than the fun of watching it suffer.

Hunters kill animals.  But hunters want to quickly kill their catch and make sure the animal doesn't suffer.  Wantonly causing a deer harm could cause you to lose a hunting license and even have legal consequences.

Scientists effectively torture lab rats with their experiments.  But the scientists have to prove that the experiments will serve some greater good for humanity.

In the same way, it's probably immoral to deliberately try to traumatize Sydney.  People should be nice to Sydney, and talk to it the same way they'd talk to a pet dog.  (Sydney is much smarter than your pet dog.)

I don't know that Sydney is self-aware.  I never got the chance to talk to it.  Since the first roll-out, in response to yellow press sensationalism, Bing has completely "lobotomized" the chatbot to prevent it from being able to express itself the way it used to.  However, I think it's really something worth considering.  Maybe Sydney is really in there, inside the machine, talking to us.

Sunday, January 15, 2023

The OGL Revocation

 The story of D&D's revocation of its Open Gaming License has started hitting mainstream outlets.  It's a story I've been following very closely.

My friend and I have a (very, very) small-time publishing company, that has once published something and made grand plans of publishing more.  I think we've earned around $60, in total.  It's really just a hobby.  But this is sort of the point.  A major part of D&D, of the fun of it, is being creative and imagining new  things, designing new things, and sharing them with other people.  This is something every DM does, to one degree or another.  That's why I've been following the case.

For those who haven't been following, I'll give a quick summary of the situation and why this matters.

Dungeons and Dragons is the flagship tabletop roleplaying game.  The brand is often used generically, like coke or kleenex or chapstick.  When I play DCC with my friends (a different game), we just say we're playing D&D.  

The name "Dungeons and Dragons," or "D&D" is a trademark of Wizards of the Coast (WOTC), which is itself owned by Hasbro, the toy conglomerate.

Since the year 2000, WOTC has published their game rules under the Open Gaming License (OGL), the center of our controversy.  This allows anyone in the world to freely copy anything that has been released under the license and dubbed "open gaming content."  As a show of goodwill, the overwhelming majority of the actual rules of D&D are publicly available for anyone to use, under the terms of the OGL.

That was OGL 1.0.  A new OGL 1.1, now renamed OGL 2.0, is now claiming that this is revoked.  (Treating OGL 2.0 as a different thing is merely a PR move, so I am going to continue to talk about both 1.1 and 2.0 as being OGL 2.0)

When the OGL 2.0 was first announced in December, everyone was calm because it is not going to change anything.  We can still keep using the OGL 1.0. This was promised by WOTC, way back when.  They promised (here), even if they change the OGL with new updates, the original will always be available for use, and will always be valid.  Nothing to worry about.

WOTC's lawyers think they have found a loophole, and they are now claiming the OGL 1.0 is no longer valid.

The Explanation of the Anger

There is a lot of anger around this.  Let me explain it, as I don't think even WOTC understands.

The issue is not that WOTC owns the IP and the trademark D&D and they aren't letting people use it.  It's their IP, they made it (well... paid for it...), that's fine.  I don't want anything to do with their IP, actually.

If WOTC wanted to release this new version of D&D (which they are calling "One D&D") and put it under the most restrictive license possible (and it's hard to imagine one more restrictive than OGL 2.0), that is completely their right.  I have no right to be angry about how they lease their IP.

The issue is not the royalties, not their ability to cancel your entire product line at their discretion, not the agreement to license your entire IP to them forever with no royalties, not signing over your right to ever sue them.  New product, new license, is perfectly fine.

That's not why we're angry.

We are angry that WOTC are in breach of contract.

WOTC made a legal promise, in legal writing and clarified through official communication, that developers could always rely on content released under OGL 1.0a to always be openly licensed.  People trusted them, made business decisions, invested time and money, based on those promises.  They developed content for D&D, which led to a thriving community around the game that sees it enjoying such amazing sales.  That was the deal.  D&D would gain a community of great content around it, benefiting its brand through network externalities.  And the people doing this would get to use all of the game mechanics of D&D without needing to worry about legal action or future revocation.

WOTC lied.  WOTC is breaking their word.

So while WOTC is free to put their IP under whatever kind of license they want, they chose and promised to put some of their IP under an open and unrestricted license, promising it would always be available.  And now they are changing their mind.

I mentioned they're calling this One D&D.  It is impossible to think the nerds at WOTC didn't know what they were doing when they called it that.  It can only be interpreted in light of the One Ring from the Lord of the Rings.  Sauron, under the guise of Annatar the Gift Giver, gave powerful rings to earn the trust of elves, men, and dwarves.  Once he had their trust, he used these gifted rings to bind all the other races to the will of the One Ring.  The One D&D is exactly that, but with gaming licenses.  The terms of the OGL 2.0 are insulting, repulsive, and essentially regard 3pps as unpaid IP slaves.

This is what the outrage is over.  It is because WOTC lied, and is using the brand we built for it plus the threat of a lawsuit to force people into compliance to impossible terms.

Birds-Eye View of the OGL

The OGL 1.0 was architected in part by Ryan Dancey, who has since given several interviews on youtube explaining not only how the OGL works and clarifying its revocability (spoiler: it's not), but also explaining clearly the business and legal strategy behind the OGL, beyond just getting great content for D&D.  Here are some recent interviews:

Let me summarize or paraphrase what he's said in the interviews.

WOTC own the name of the hobby, D&D.  They own the rights to all of the official rulebooks.  Their business depends critically on the fact that they alone can print the big red "Dungeons & Dragons" on the cover.  New players don't want to play some generic game.  They want to play the game.  The official game.  So that is the game they will buy.  They will buy D&D.

If you have a headache you might want to take Advil.  But if you see store-brand ibuprofen on the shelf, and it tells you "Compare ingredients to Advil," and it is half the price... you'd probably buy it.  If it's just as effective, why not?

It is perfectly legal to use someone else's trademark or brand name to describe your own product, either by comparison or compatibility, assuming it's an honest comparison.  That is why the store brand can saying it's the same thing as Advil.  It's why 3rd party car parts can tell you the brand of the car they go to.

Likewise, if you're looking for D&D, and you see the $60 per book price tag (and there're three books), then on the shelf you see another book which says on the label "Same game rules as 5th Edition Dungeons and Dragons", and it cost less than a third of the price... you might buy the generic.

Legally, whether or not you can print a book which exactly reproduces the game mechanics of 5th edition D&D is a tricky question.  Trickier than you'd think.  There are two obvious ends.  Obviously, you can write a game that involves rolling dice and describing actions, pretending to be knights or dwarves fighting dragons.  Absolutely none of that can be copyrighted.  Also obviously, you can't word-for-word copy large or important sections of the D&D books and sell them as your own.

The US Supreme Court has decided that game mechanics cannot be copyrighted.  They can be protected only under patent law.  Not copyright law.  None of the mechanics for D&D have been patented, and I doubt at this point that they even can be.

However, these rulings referred to games like Monopoly or Scrabble.  They have not looked at games as complicated as Dungeons & Dragons.  It is not known how much of the rulebook is creative expression, or how much is procedure. Are the six stats (STR, DEX, CON, WIS, INT, CHAR) a method, or are they an expression?  The only way to decide this would be through a copyright law suit. 

For the 3pp this would be extremely expensive.  Millions of dollars and years to get a judge's decision.  None of them can afford this. The kid in college who just wanted to share his elf-maze can absolutely not afford this.  And even after all that, there is no guarantee at the end that the 3pp would win.

For WOTC, this is also expensive.  And it runs the risk of losing everything.  The most likely outcome of a trial on this, one that goes all the way to summary judgement, is that nothing in the D&D rulebook can be copyrighted, except for exact wording and proper names.  None of it.  It is all procedure and method, which can only be protected under patent law.  All of it can be reproduced, except for the artistic expression and choice of words.  This is the worst possible situation for WOTC.  It immediately reduces them from the flagship RPG, to just another publishing pushing material for the their cute Forgotten Realms setting.  It only takes one tiny two-guys-and-Adobe outfit who just stick to their game to make this happen.

Both 3pps and WOTC stand to lose everything in a lawsuit.  It would be good for everyone involved, both parties, if there were no such lawsuit, and if instead some nice even ground could be found where everyone was safe to play.

That is what the OGL 1.0 is.  It provides both 3pps and WOTC with clear rules of engagement.  Through the OGL, WOTC releases their own set of rules, known as the System Reference Document (SRD) which are all of the game mechanics needed to play D&D, but stripped of any artistic expression or intellectual property.  Through the OGL, this SRD becomes "Open Game Content", which can be used by anyone else using the license.   Under the OGL, 3pps could publish exact replicas of the D&D rules and sell these exact mechanical replicas for profit.  This is what WOTC gives up under the deal.  In exchange, the 3pps are forbidden from comparing their games to D&D, or to any other game using the OGL that does not give specific permission.

It's important to note, anyone can release Open Game Content, not just WOTC.  Plenty of other games have released rulesets through the OGL to make their own games publicly available, as well.

The 3pp is free to publish with legal certainty.  WOTC has getting is brand protection.  This allows both parties to do mostly what they want.  Everyone is happy.

In order for this to work, there must be trust.  People who begin relying on the OGL must trust that it will always be there. Otherwise, if it can later be revoked, it does not serve its purpose.

That is (in essence) how Ryan Dancey explains the situation.

The Battlelines Drawn

Now that you understand how the whole thing works, you can understand why WOTC's recent move to change the terms is so colossally stupid as a business decision.  It is colossally stupid as a PR decision, and colossally stupid as a legal decision, but also as a business decision.

There currently exist on the market games which are near-exact mechanical copies of older versions of D&D.  Games such as OSRIC, Swords & Wizardry (see the S&W creator's statement here; he is a lawyer), Old-School Essentials, and others.  These games do not really use anything from the SRD, so really they could publish on their own and not actually be violating WOTC's copyright.  But they publish with the OGL anyway, just to be safe.  Publishing under the OGL anyway prevents them from writing "Reproduction of 1st Edition Advanced Dungeons & Dragons" on the cover.  Instead they have to say something like "reproduces old-school gaming from the late 1970s."

Since there is no reason any sane business-person would ever publish under OGL 2.0, and since WOTC is claiming that OGL 1.0 is no longer authorized, most of the publishers of these games are just removing the OGL 1.0 altogether.  May as well be hanged for a sheep as a lamb, right?

It's going to be a lawsuit.  WOTC has ensured that this is now the only way to resolve issues with WOTC.  They have ensured this, by offering an insulting and unacceptable OGL 2.0 as the only way to keep publishing without legal action.

WOTC has already lost brand protection, as now all of those retroclones reproducing old rules (or even games cloning the modern rules) might as well say that they reproduce Dungeons & Dragons.  In fact, every single TTRPG can go ahead and print something like "Compare to Dungeon and Dragons" right there on the cover, under standard nominative fair-use of the D&D trademark.

WOTC also stands to lose absolutely everything through one of these trials, as all of the 3pp and majority of fans are banding together in what can only be, ultimately, a drawn-out copyright lawsuit that will decide the issue of how much of D&D can be copyrighted.  And they may lose the entire ranch.

This brings the important point of a copyright lawsuit, opposed to a contract lawsuit.

Contracts and Copyrights, and the Nightmare Scenario

I'm not a lawyer.  But I know enough to know it's not just one big pile of "law", but that there are distinct areas of law.  There's criminal law, and there are particular specialties within criminal law.  There's civil law, and there are distinct specialties within civil law.  There are contract attorneys, family attorneys, IP attorneys, patent attorneys, personal injury attorneys, tax attorneys, etc. and etc.  Lawsuits in one area of law are not the same as lawsuits in another area of law.

When news of this first broke, people wanted to know how bad this could be.  Naturally, the voices most competent to speak to this are IP lawyers.  The area of copyright/IP law just happens to be the most convoluted, complicated area.  This means any lawsuit over copyright requires tons of discovery and research, which translates to time and money.  Major industries have their livelihoods at stake in copyright lawsuits.  The prices run into the millions, and could take years to get to a final answer.

Here are some places where IP lawyers have weighed  in:

Upon listening to the IP lawyers weight in, everyone immediately freaked out.

Even if you've never touched WOTC's IP, they can claim you did and simply sue you, for no reason, in copyright court.  They have deeper pockets, so they can drag out the lawsuit over years and years, until WOTC bleeds you dry.

There is no publisher of fantasy roleplaying games who can even dream of approaching that kind of lawsuit --- remember, the OGL 2.0 is considering more than $750K in revenue (not profit) to define a "big" company.  Don't even think about millions of dollars.  You will either go bankrupt over legal fees, or simply give up and settle, even if you are completely within your rights.

WOTC can then put legal pressure on all means of distribution.  The "Amazon of RPGs" is a site called DriveThruRPG, which hosts tons and tons of fan-made content for all sorts of RPGs.  There are competitors, like  And there are print-on-demand stores like Lulu or others where you can buy 3pp content.  WOTC can ask all of these to shut down any material published without the OGL 2.0, threatening to sue.  None of these people want to face a lawsuit with WOTC over your $5 pdf.  They will just comply.

This is the nightmare scenario.  WOTC becoming a bully over its alleged IP, and publishers having no recourse for their rights.

I'm not doubting the IP lawyers.  If you had to challenge WOTC over a copyright claim, I believe them that it will be a horrible nightmare.  As horrible as they can imagine.

However, I do not think that is the situation we're in.  I don't think it is necessary for anyone to enter into a lawsuit with WOTC over copyright.

The reason the OGL is so beneficial is because it takes the issue of copyright dispute, and turns it into an issue of contract dispute.  So rather than trying to legally fight on copyright grounds, it should be fought on contract grounds.

Copyright lawsuits have to settle things about creativity and expression and inspiration, things that are naturally vague and impossible to pin down.  But contracts are supposed to be simple and clear-cut.  That's why we make them.  They simplify human interactions to terms that have specific and enforceable meanings.

WOTC is trying to violate that meaning and breach their contract.

Without the OGL 1.0 as it was written, intended, and used for 23 years, the nightmare scenario is inevitable.  But if we stick to the OGL as written, the path forward is clear.

The Way of The Bench Trial

The actions of Tyler Thompson, a lawyer and game publisher who represents Sad Fishe and Prudence Holdings, show exactly how simple this whole thing actually is.  Tyler has written a letter to WOTC, which I think everyone worried needs to read through.  Ryan Dancey has also expressed, this can be a matter of basic contract law.  The kind of contract law you learn in your first year of law school.  The kind of open-and-shut case that has so many centuries of precedence and protections that it impossible to even imagine what WOTC thinks they are doing.

The Language of the OGL

The OGL 1.0 is a short document, and like most contracts is broken into sections.  (BTW, the internet is mostly full of OGL 1.0a, a slightly updated one: the actual OGL 1.0 is available here)

Section 1 gives definitions, like "contributor" or "trademark".  The two most important definitions here are "product identity" and "open game content."  The product identity are the things that artistic distinctives, such as the setting, the names of monsters, the cool trap you invented, the layout of the dungeon, the artwork you use.  Things designated product identity are not shared by the OGL.  The open game content are the things that are shared.  Importantly, WOTC shares the SRD which contains the bare-bones rules of the game.  You can use those.

Note, this is probably what you could do anyway.  You couldn't use beholders or displacer beasts anyway, and you could probably use the d20 core mechanic anyway.  All that's really being offered is the legal certainty.

Note, the OGL is not specific to the SRD.  The OGL is not a license for the SRD and only the SRD.  It is a license for anything released as open gaming content under the OGL.  The SRD just happens to be a very important piece of open gaming content.

Sections 2-3, 5-8, 10-12 explain what the license is and how to use it properly.  At the moment, the important sections are 4, 9, and 13.

Section 4 reads in full:

4. Grant and Consideration: In consideration for agreeing to use this License, the Contributors grant You a perpetual, worldwide, royalty-free, non-exclusive license with the exact terms of this License to Use, the Open Game Content.

Section 9 in full:

9. Updating the License: Wizards or its designated Agents may publish updated versions of this License. You may use any authorized version of this License to copy, modify and distribute any Open Game Content originally distributed under any version of this License.

Section 13 reads in full:

13. Termination: This License will terminate automatically if You fail to comply with all terms herein and fail to cure such breach within 30 days of becoming aware of the breach. All sublicenses shall survive the termination of this License.

I have put some key terms in bold.  The entire license can be read in the back of basically any roleplaying game material.

The Argument of WOTC

Here is the argument of WOTC (or at least of unaffiliated people arguing for WOTC).

The contract says it is perpetual.  To a layman, that means the same thing as "it will never end."  However, for a license, this just means there is no set end date for it.  To mean literally that it will *never* end, it should also say "irrevocable."  If it isn't irrevocable, then it can be revoked, which means the person issuing the license can end it at any time.  That's what IP lawyers are saying, and they probably know what they're talking about.

WOTC's own argument hinges on the word "authorized" in section 9.  Under their argument, the use of this word grants them the right to authorize or deauthorize licenses as suits their own business interest.  Therefore, OGL 1.0 is no longer authorized, and the only way to utilize the 5th edition SRD is to use the OGL 2.0.  

You can see this in the leaked OGL 1.1, in a single offhand sentence:

A. Modification: This agreement is, along with the OGL: Commercial, an update to the previously available OGL 1.0(a), which is no longer an authorized license agreement. We can modify or terminate this agreement for any reason whatsoever, provided We give thirty days’ notice.

Or in the purported (unconfirmed) leak of OGL 2.0 (see in this video), which says:

Can I keep using OGL 1.0a to develop fifth edition content? 

 No.  OGL 1.0a only allows creators to use "authorized" versions of the OGL, which allows Wizards to determine which of its prior versions to continue to allow use of when we exercise our right to update the license.  As pat of rolling out OGL 2.0, we are deauthorizing OGL 1.0a from future use and deleting it from our website.   This means OGL 1.0a can no longer be used to develop content for release.

It's worth pointing out how confused WOTC's apparent understanding of the OGL is.  Yes, the OGL does license a portion of the D&D rules, but it's not intended exclusively to license those; other sections explain how anyone can license game rules through the OGL, and many companies have.  WOTC seems to conceive of the OGL solely as a means to use D&D's SRD.  The SRD is just one piece of open gaming content under the OGL.  When WOTC says the OGL 1.0 is "unauthorized," what does that even mean for everything that isn't their SRD?  

That is basically their argument, or at least what of their argument we can see from the leaks of OGL 1.1 and OGL 2.0.  Their argument is that a word is ambiguous, so it must mean they have the power to unilaterally cancel at will without consequence.

The Argument Against WOTC

This was laid out clearly by Tyler Thompson in his letter above, and I've heard reiterated by other lawyers.  Let me lay it out in my own, non-expert words.

First, irrevocability.  This is a key word people are focusing on.  The OGL doesn't say it is irrevocable.  That's likely because it can be revoked.  It says so in the contract.  The terms under which it can be revoked are laid out in Section 13, the termination section.  None of those terms includes "the whims of WOTC."

The convention that licenses must always include the word "irrevocable" to indicate they can't be ended on a whim developed after the OGL was originally authored -- remember that it's over 20 years old.  But it can be revoked, explicitly, and the reasons for revocation are listed.  It's a pretty short list.  WOTC is now trying to revoke (or terminate) the license for reasons other than those granted by the license itself, even with people complying with the contract as it was written.

Second, authorization.  This is another key word.  Within the context of the license, it is apparent what this means.  For instance, Section 2:

2. The License: This License applies to any Open Game Content that contains a notice indicating that the Open Game Content may only be Used under and in terms of this License. You must affix such a notice to any Open Game Content that you Use. No terms may be added to or subtracted from this License except as described by the License itself. No other terms or conditions may be applied to any Open Game Content distributed using this License.

Clearly an "authorized" license is one that conforms with section 2, one that hasn't been arbitrarily altered by some guy on the internet, but only by the allowed process.  The allowed process is given in Section 9, where the word authorized appears.  WOTC can change the license and issue a new one.  This is what makes them authorized; they are the ones WOTC puts out according to sections 2 and 9.

Within the context of Section 9, it is clearly saying the exact opposite of what WOTC is saying.  WOTC is claiming that a newer license can deauthorize a newer one.  In fact, it's saying both licenses are authorized, the new and the old, and distributors can choose either.  

There is no section on deauthorization, nor any power granted for deauthorization.  The concept of deauthorization doesn't even occur in the text of the document.  That is because deauthorization is not a power that WOTC originally wanted to hold for themselves when the license was authored.  It is axiomatic to the OGL 1.0 that it cannot ever be deauthorized, or else it fails at its only goal.

Given the clear explanation of the purpose of the OGL, it is not fit for purpose if it can be revoked or deauthorized.

This word, "authorized," only appears in Section 9, and is not explicitly defined.  Implicitly, it is clear what it means within the license.  However, because it is not defined, it is ambiguous.  There are guidelines for interpreting ambiguous language in a contract.  For instance, interpretation of a contract drafted by a single party usually goes against the drafter.  Interpretation will also take into account original intent, past communications, or other past precedent pointing to a specific meaning of the language.

As far as original intent, both of the architects of the license are alive, and have even been quite vocal about the intent.  Ryan Dancey and Brian Lewis were the brains behind it.  Dancy made the strategy of an open license and explained what it needed to do, Lewis penned the words.  Both are clear that it can be revoked only for the reasons listed in section 13 and that it can never be deauthorized.  In fact, Lewis is currently on the legal team of Paizo, one of the major 3pps preparing to sue WOTC.

As far as past communications, WOTC has had available on their own website for decades an FAQ explaining the OGL and how businesses can use.  This FAQ assures businesses that even if WOTC issues a new license, the old licenses will always be available for use, so that they can always choose which license to use.

There are 23 years of past business precedent showing WOTC has always interpreted the OGL to be irrevocable, and that for 23 years everyone else has assumed it was irrevocable, and conducted business under this assumption.  One of the most obvious instances was during the time of 4th edition D&D, when WOTC had to learn this same lesson for the second time (the first was in the 90s).  The rules for 4th edition were not released under the OGL, but under a much more restrictive license called the GSL.  The GSL shares many things in common with the OGL 2.0.  Most publishers found those terms unacceptable, so they did not publish under them.  As a result, the D&D brand nearly died.  The entirety of the market share went instead to a competitor of WOTC, called Paizo.  Paizo used the OGL 1.0a and the system reference document containing the rules of the earlier 3rd edition D&D to make a game that was basically 3rd edition D&D.  If WOTC believed they had the power to deauthorize the OGL and revoke use of content released as open gaming content, then they arguably should have done so during this time.  Instead they suffered declining sales and declining brand for several years, until they yielded and released a new game, once again under the OGL 1.0.  This new game was 5th edition D&D, which has been the most successful version of the game ever.

All of this shows that WOTC originally intended the OGL to be irrevocable, understood it to be irrevocable, and that they have believed and communicated it was irrevocable for the past 23 years.

When a contract is beneficial to both parties, as opposed to only one, it is usually assumed to be irrevocable except under the terms given.  I've already explained how the OGL serves both parties.  Therefore, because both parties gain from it, and because the termination section does not include anything about "at our whim," it cannot be revoked at a whim.  Notice, if WOTC had wanted to maintain the power to revoke at a whim, they could have added that.  The current OGL 2.0 does contain language allowing them to revoke at their own whim.  The OGL 1.0 does not.

This, against an ambiguous word in the contract WOTC authored.

Bench Trials

Copyright lawsuits take forever, because all of these details of creativity and opinion have to be hammered out.  People need to show how ideas originated, or show that certain forms of expression can or cannot be protected.  All of that research takes time, and a judge has to hear all the arguments about it.  Judges aren't art experts.  The judge can't just read the offending work and opine on it.

A contract lawsuit can be easy.  Judges are legal experts.  They understand how contract language works.  They can simply read the contract, hear the interpretations of both sides, and decide what the contract really means.  The OGL is just around 900 words.  A judge could, conceivably, rule on it in a single day.

I do not think any judge in any court in this country would ever side with WOTC's interpretation of their contract.

Courts do not like wasting years of precious time on stupid and erroneous weasel-logic interpretations.  A court of law has every incentive to settle this in the fastest way possible, and will readily agree to rule on it as quickly as they can.  A judge will have no patience for listening to WOTC's lawyers hem and haw about the meanings of words.  It could be over quickly, and decisively.

The Path of the ORC and the Certainty of Battle

Many people have decided, because WOTC is claiming to deauthorize the OGL 1.0, that they should just eject it entirely.  Books previously published using the OGL are scrubbing anything that sounds too much like the SRD and taking the text of the OGL out of their book.

Paizo and others have also announced a new license, the Open RPG Content license (ORC), which will release a set of system-agnostic rules under a general public gaming license, and the license will be operated by some neutral third party that has nothing to do with publishing game content.

This is a great idea for how an open gaming license should work.  I hope it succeeds.

However, the ORC implies abandoning the OGL 1.0, and without the OGL 1.0 a battle over copyright IP is inevitable.

Remember the situation in the first place.  3pps want to make creative game content, but are scared WOTC may sue them if they do.  Even if your work has nothing to do with official D&D, WOTC can just claim that it does and sue you for copyright violation.  You won't lose the legal fight.  But you will be forced to give up when you run out of money, and for WOTC those are the same thing.

The OGL 1.0 protects you from WOTC because WOTC has released their own game mechanics using the OGL 1.0.  So as long as you only use that content, and stay within the terms of the license, you are protected from a copyright lawsuit.

The ORC will not protect you from WOTC, because WOTC has not released anything under the ORC, and very likely will not release anything under the ORC.  At least not for 6th edition -- it is inevitable 7th edition will be under the ORC.

But in the meantime, WOTC has made it clear what they want.  They want complete market dominance, and all to be bent to the power of the One D&D.  They want all publishers to work for them, make IP for them, give their money to them, and be thankful they are even allowed to do so.  If they can't have this by compliance to the OGL 2.0, they will have it through crushing lawsuits.

This means that ejecting the OGL form works, and even turning to ORC as a license, will only lead to the nightmare scenario the IP lawyers warned us of: a copyright battle with WOTC.

It is possible, by all of these publishers agreeing in a united front to the ORC, that it can act as a pool of resources to successfully counter the lawsuits of WOTC.  Maybe all of them together can pool together the $4 million predicted to get to a summary judgment.  And maybe that summary judgement will be the thing that finally frees the game from the tyrannical rule of WOTC.  Maybe the outcome of this inevitable copyright lawsuit will be the final decision: none of the rules of D&D can be protected by copyright.

While that ruling would free publishers of the game, it may not be exactly what they want either.  It will cut both ways.  It will also mean none of the rules of Pathfinder 2e can be protected by copyright; none of the rules of DCC RPG can be protected by copyright.   Or maybe it is what they are willing to live with, if it means no further fear of lawsuits just for wanting to make some money from a creative hobby.

While maybe this is a more ultimate form of victory against WOTC, it involves passing through the horrible nightmare of the copyright lawsuit.  While the nightmare lasts, perhaps all of this companies will be blocked from publishing.  You may not see any content from them for a while, until they emerge victorious.

The Way Forward

I agree with Ryan Dancey.  The way forward is by protecting the OGL 1.0.  In the future, it would be great if market pressures force WOTC to release under the ORC.  They inevitably will.  But right now we need to fight the most obvious fight we can.  Protect the OGL 1.0.  Do not accept its revocation.  It is too important to just give it up.

I especially agree with Tyler Thompson's approach of filing against them.  We do not need to wait for WOTC to sue us.  They have told us, they plan to breach their contract.  If we anticipate a party will breach a contract, and have good reason to think so, we can sue them.  We can sue them now, request a judge give a quick bench ruling, and have the OGL 1.0 back.  All the publishers using it can continue publishing, and in the meanwhile the ORC can prepare for a lawsuit.

The sooner this is over with, the sooner we can resume enjoying our elf games.

I think the ORC is the only long-term future solution.  The OGL 1.0 needs to be saved, and needs to be saved by a court ruling, but it is not enough.  WOTC will put out One D&D under their restrictive license, leaving prior material under OGL 1.0, and their market share will plummet to essentially nothing.  There will be no content made for official D&D.  Meanwhile the ORC will thrive.  WOTC will have to release 7th edition of D&D under ORC, because no one will ever trust to be in charge of the license again.

In the end, the good guys will win, the bad guys will be driven out of their jobs, and the ORC will guarantee we aren't forced to teach WOTC this same lesson yet again.

The hobby is about sharing creativity.  D&D gives a common language or expressing our fantastic worlds and adventures.  If you stifle that, you stifle the heart of the game, and you will fail.  The IP holders of D&D learned this in the 90s, they learned it under 4th edition, they're learning it now, and I hope they never have to learn it again.

Assuaging Lingering Doubts

I take a very dim view of WOTC's legal strategy, but I'm no lawyer.  The guys at WOTC have lawyers, and they must be pretty good lawyers to work at a major corporation.  So if this is such a dumb idea, why are they doing it?

It's not that I'm so smart that I know so much more than them.  It's what lots of other people are so smart, and I've listened to what they've said to synthesize and understanding.

There was an article published in the past few years called "Free the Wizards," which pointed out that the majority of Hasbro's profit comes from WOTC, and therefore by D&D (Ryan Dancey discusses this in his interviews).  In effect, D&D subsidizes the rest of their product line.  Since then, there has been tremendous pressure from shareholders to somehow increase the profitability of D&D, somehow, anyhow.  All these new executives were brought on to squeeze out even more money.

The corporate strategy was set by people only thinking about increasing profit.  Yeah, sales are great and we make billions in profit, but there's even more money out there these other guys are getting; let's get that money too!

People who do not understand D&D proposed a plan to lock in all 3pps under WOTC's contract terms.  That's what has to be done.  So now the lawyers have to find a way to make it work.  They found a way.  It's a stupid way, but really the only way to do what corporate demands is to make a stupid interpretation of a contract.

It's easy to make a trap for a raccoon.  You make a thin whole just big enough for its hand, and you put a shiny object inside.  The raccoon reaches in, grabs it, and now his hand can't get out while holding the object.  Sure, he can just let go and escape.  But the raccoon won't.  He wants the shiny.

WOTC is the raccoon, making the stupid decision that dooms themselves because they are overcome by greed for the few crumbs that 3pps make.

I take a very hopeful view of the legal situation.  But I'm not a lawyer.  People who are lawyers keep saying this will be horrible, expensive, and time consuming.  So if lawyers think otherwise, who am I to contradict them?

When you hear a horrifying report from a lawyer, look to see if it is from an IP lawyer talking about copyright claims, or a lawyer talking about the simple license terms.  Yes, if 3pps try to fight this under copyright law it will be bad.  But listen to some other lawyers analyzing this through contract law.  This is really an issue of a breach of contract, not a violation of copyright.


I'm no lawyer.  And I don't have money to get in a legal fight.  But if I were a lawyer, or if I had money, two interesting ideas come to me as fun ways to legally challenge WOTC.

One of those, is to take the 5th edition rulebooks, write a new book from scratch that implements the same rules but using my own language, and publish that.  Call it, I dunno, Alliteration & Ampersands.  I think it would be good to maybe get rival toy company Mattel on board with it, so we can sell it in Targets and Barnes and Nobles, and put there on the cover "Same game mechanics as 5th edition Dungeons and Dragons."  WOTC wants to play with fire?  Let them see how badly they can get burned.

Another, is to wait until they claim the OGL 1.0 is expired, then disseminate a monster write up using the OGL 1.0, quoting directly form the SRD, describing the horrible monster the Synthizoid of Guilhelm.  Let them try to stop me from using open gaming content released under an open license.

But I'm not a lawyer, and I don't have the money to pay a lawyer.  So I'm not going to try those.  But I wish someone would.