But then you play the games, and they're all so opaque. You have no idea what's going on, and the responses to your actions are so hard to grasp. But if you do figure it out, the model usually collapses into a linear, repeatable strategy and the illusion of depth disappears. You can see this happening from the start, with Gossip. Instead of noticing that his game didn't communicate and looking for points of accessibility, he plunged further forward into computer modelling. The failure is one of verisimilitude: The model is similar to a grounded truth on paper, but it's uninteresting to behold because it doesn't lead to a coherent whole. It just reflects the designer's thoughts on "this is how the world should work", which is something that can be found in any comments section.

Often, when Crawford lectured, he would go into evo-psych theories to build his claims: that is, he was confident that the answers he already accepted about the world and society were the correct ones, and the games were a matter of illustration. He was likewise confident that a shooting game would be less thoughtful than a turn-based strategy game because the moment-to-moment decisions were less complex, and the goal should be to portray completeness in the details.

I think he's aware of some of this, but he's a stubborn guy.

I played it as a cold war kid and was fascinated by it. Mid 80's, post War-Games, this game blew my mind. It simulated the world.

The lesson I remember was that conflict in the Cold War was not zero-sum. One side would win and one side would lose. There were (in this game) no win-win outcomes. But - and this is the key point - the value of each win or loss was unequally felt. For the US to back down in Indonesia was disappointing. To back down in West Germany was fatal.

Oh - and also the notion of graduated escalation & de-escalation. Playing the game well requried using escalation wisely. Sometimes you escalate (a bit) to see how they respond & judge the value of a conflict to your opponent. Sometimes you escalate (a lot) to signal to your opponent that a given conflict is very serious to you.

I don't know if I ever had _fun_ playing the game - but of the hundreds of games I played as a kid this one stuck with me.

All this with something like 64k of memory - brilliant!

What I mean by that is that there are game designers like Jonathan Blow who have their own theories on what is a great game and are extremely critical of the industry not following those theories, and then have released games that succeed at demonstrating those theories. In Jonathan Blow’s case, you can disagree with the man, but you can’t disagree with the fact that The Witness is a wildly original, successful game (1M+ copies sold) that has a cult following.

That does not seem to be the case for Crawford’s work. Lots of theories, lots of indictment for the industry doing it wrong, but no actual demonstration of what “doing it right” would mean.

Saying that no one gets it and civilization won’t be ready for many centuries (as the article I linked above does) feels like kind of a cheap rhetorical cop out.

For what it’s worth, I disagree with his indictment of the video game landscape as being narratively poor. Lots of video games with great interactive narratives out there, and there are many players who have been deeply moved by such games (of course, which games that might be varies from person to person).

I think a good antidote when one finds themselves in those thinking patterns is to listen to what others have to say, and not dismiss them as not getting it because they don’t follow your particular (unproven) theories.

I first attended CGDC in 1994. It was two years after his Dragon Speech (which I knew nothing about), but the highlight of the conference was his talk about the challenges of story-based games. The part I remember is how he modeled stories as branching trees with fan-out, foldback, tree-of-death, etc (he covers this in the "Architecture" chapter of this book "The Art of Interactive Design"). I didn't really follow Crawford's work on Erasmatron or later, but by the late 90's it sounded like his story model had changed from a tree to a graph network structure, like a finite state machine. While it was an improvement, I was a bit skeptical that this model was enough. Nevertheless, I spent a lot of time thinking about the problem. You see, he'd already infected me with his dream of interactive storytelling.

By the time I moved to California and took a job at EA/Maxis on The Sims 2.0 team, I had decided that true interactive storytelling (as I saw it) was not possible until game AI was sophisticated enough to enable autonomous NPC chatbots. So I put that dream aside while I pursued a career in software development.

Here we are, over a quarter of a century later, and that AI technology is here now. For those of us who have been waiting for this moment, it is almost miraculous. It might be the end of an era for Chris Crawford, but it is just the beginning of the AI-based interactive storytelling era.

I'm "only" in my 50s now, but I am in my 50s now. For a couple of decades I've had this grand vision about how to build the perfect algorithmic trading system. The practical expression of this has come in fits and starts.

During the Pandemic, I finally devoted some time to building a bespoke server and hosting it co-located facility near a high-quality market data provider. I chose to do this before the software was ready in order to "light a fire under my a$$" and help motivate me to move things along. By the end of 2021 I had written, tested, and deployed a set of specialized clients written in rust to consume the market data and perform basic parsing of the real-time information. I stored the transformed output of these processes in log files as a temporary data sink that could represent the data "someday" moving downstream in a custom stream processing platform that would perform trades via an online broker.

And then, it stalled. There always seemed to be good reasons why I couldn't commit to one implementation strategy or another. Would I pass the data downstreeam via pipes? With Kafka? Via shared memory? Would the parsed input be represented via protobuf records? A custom binary format? Apache Arrow/Parquet format? Would the algorithms be expressed in custom rust code? Using Pandas(numpy)/Polars? Would I focus on my imagined "insights" into how prices move or make the transition to Machine Learning by trying to produce "special sauce" custom features that could be fed into models of various types? In the meantime, I have only collected a high volume of daily log files.

There has been some progress, of course. Last year, with LLMs as my pair programmer, I added vector processing to the ingest components, which made them run even more efficiently. I certainly enjoy the sporadic work I dedicate to working on the system. But there always seems to be a fog of possibilities ahead of me obscuring my view towards a broker API getting called.

My family and closest friends look upon this as the "hot rod in his garage" which I sometimes work on and which will probably never get finished.

Someone who clearly wanted to make a difference, but mostly seems to have not just made games.

He made game tools, but then didn't actually use them to make games. And then he blamed everyone else for not being ready for what he was making.

Giving up after only one released work just seems like such a shame.

I have watched the dragon speech multiple times, and it makes me feel like you should put everything in what you are doing, as everything is your magnum opus. I have huge respect for artists and artisans that work like that. It is really difficult, and often not economically viable, but there are people that just `see` something and they want it to exist.

I often think about the dragon speech, when I am doing the same 'pick up 10 things' quest in World of Warcraft that I have been doing since 2005. Thousands of times.. I have picked up hundreds of thousands of things :) and I wonder, what game would WoW be had he succeeded.

Some of his reflections on growing old, remembering his first crush, and even just noodling about home improvement are incredibly beautiful too.

Those are all asides, but what I mean to say is that his other posts are worth reading.

https://www.erasmatazz.com/index.html

I was not familiar with Chris Crawford before this, though I think I'll look into him more. Reading his idea of People Games [1] I wish he was a younger man with a bit more time to revisit these ideas with new technology. There are new interactive mediums to discover with LLMs, and it's mediums that he's clearly been trying to create all this time...

Quoting his excerpt:

"I dreamed of the day when computer games would be a viable medium of artistic expression — an art form. I dreamed of computer games encompassing the broad range of human experience and emotion: computer games about tragedies, or self-sacrifice; games about duty and honor, patriotism; a satirical game about politics, or games about human folly; games about men's relationship to God or to Nature; games about the passionate love between a boy and girl, or the serene and mature love between husband and wife of decades; games about family relationships or death, mortality, games about a boy becoming a man, and a man realizing that he is no longer young; games about a man facing truth at high noon on a dusty main street, or a boy and his dog, and a prostitute with a heart of gold. All of these things and more were part of this dream, but by themselves they amounted to nothing, because all of these things have already been done by other art forms. There was no advantage, no purchase, nothing superior about this dream, it's just an old rehash. All we are doing with the computer, if all we do is just reinvent the wheel with poor grade materials, well, we don't have a dream worth pursuing. But there was a second part of this dream that catapulted it into the stratosphere. The second part is what made this dream important and worthy: that is interactivity.

"Let me explain to you why interactivity is so overwhelmingly important. Let me talk about the human brain. You know, our minds are not passive receptacles, they are active agents. It’s not as if we have a button on the side of our heads and they come along and push the button and the top of our heads flips open and then they take a pitcher full of knowledge and pour it into our skull, and then they close the top of our head, shake well and say, «Congratulations, you’re educated now!». [...] All the higher mammals learn by playing, by doing, by interacting [...]

"The interactive conversation is effective, but the expository lecture is efficient. That’s the trade-off we make. And over the centuries, we humans have learned that the gains in efficiency outweigh the losses in effectiveness. And therefore we choose expository methods. But the sacrifice remains real! We haven’t ever solved that problem. It’s been with us since the beginning of history. Every single artist has faced this, every communicator, every teacher, every novelist, every sculptor, every singer, every musician, every painter, every single artist through all of human history has been forced to sacrifice effectiveness for efficiency… until now. Because now we have a technology that changes everything. [...] That is the revolutionary nature of the computer. It allows us to automate interactivity to achieve both effectiveness and efficiency. That was the most important part of my dream."

[1] https://en.wikipedia.org/wiki/Chris_Crawford_(game_designer)...

I explored tabletop games after that, modding, expanding, and even building prototypes using 3DP.

When AI exploded 2 years ago I was curious how much of a game it could build and I had recently learned about the existence of Pygame. Ended up building a zero-player 2D space sim. It's stupid, but I appreciate it and there is some art in it.

https://github.com/derekburgess/simcraft

Then recently, a coworker asked if I had tried Codex yet. I hadn't, and while riding in a car had the idea of using Codex on my phone to make a game that I could then deploy on my phone and play. Wanted to explore something meta, so I made an AI dungeon master in the style of Zork. It got out of hand and you cant deploy it on a phone- but I did deploy it remotely on my home workstation and then played over ssh from my phone ;P

https://github.com/derekburgess/dungen

I appreciate games for their randomness and have always leaned into games that incorporate generative sandbox experiences.