|
Post by Chris Conley on Nov 3, 2016 15:13:37 GMT -8
I actually think that CoG is wonderful. It's just too bad that it's mired in a paradigm that will go nowhere. It's quite intentional, by the way. I managed to corner one of the founders of CoG at GDC and I asked him why they persisted with the dead-end approach of ChoiceScript. You know what he told me? They wanted to make a platform for authors to easily make games. Ideally, a published author (like Max Gladstone, for example) should be able to pick it up, learn it and make something neat, all without getting a 4 year degree in systems design. So, CoG is awesome, but it's not going anywhere because their creator audience is from a different medium. It's nice, but it's like recording theatre plays on film. A wonderful way for people who are writers at heart to earn extra cash, but a dead end. Oh, I know, CS is a mess. Doing anything complicated at all, you have to code it from scratch, because it's not built into the assumptions and interface of the engine—because there is none. It's basically BASIC with python-style indents and a couple of nice structures for writing multiple-choice. And it's ironic that targeting nonprogrammers is their goal, as so much of its design interferes, as I said before: IDElessness, errors being possible, no in-built handling of things like opinions and emotions and AI reaction scripting. I mean, I did that in my work (to a limited extent), but I had to code it all from scratch. I haven't had the chance to play B&L, no. I didn't have an iPhone during the couple of years it was available in the store (still don't, actually), and now it's dead because of an iOS update, and the devs don't have the resources to port it to a modern OS. I do see the value of having some graphics in the frame; my CoG release probably could have used one, because it seems like my resource system is just too fiddly to fit in a standard text-only format. And the old iteration on that design was even worse in that respect. Funny you say that. I just read this article about current developments. Google's procedural voice generation tech doesn't sound half bad. Cool. I'm definitely interested to see more from you.
|
|
|
Post by alexvostrov on Nov 3, 2016 16:34:16 GMT -8
Funny you say that. I just read this article about current developments. Google's procedural voice generation tech doesn't sound half bad. Thanks for the Guardian link. I'll check it out. Yes, I saw the deepmind voice synthesis article a few weeks ago. It's getting to be very impressive. They could probably train the neural network on voice samples with different emotions and get it to a pretty good uniform emotional expression (80%?). As a side-effect, they would also get an engine that recognizes how angry you are when you yell at their automated phone support system! The big challenge is subtle intonation, and well ... acting. There are lot of different ways to put emphasis on words to convey meaning. Imagine trying to offer a bribe while having plausible deniability. Cool. I'm definitely interested to see more from you. I figure that the way all of this will get settled is hard work. Chris will make Siboot, and it will do some interesting new stuff that I haven't thought of. I'll eventually release my thing and we'll compare notes. Everything else is just talk. The way you prove thing right now is by finishing something.
|
|
|
Post by billmaya on Nov 3, 2016 20:27:37 GMT -8
Back in 2009 I spent two years attempted to re-imagine the 1987 version of Trust & Betrayal: The Legacy of Siboot using Chris Crawford’s Storytron technology so I can say with some level of confidence that I’m probably one of a handful of people who delved deep enough into the technology to (a) realize how complicated it was and (b) figure out how to get things done (I’m the “Maya” in Chris Conley’s initial post). I was probably 6-9 months away from being able to release my work when Chris Crawford made the unilateral decision to sunset the technology, promising better things to come. That was in 2010. I ported Chris Crawford’s Java Gossip code to the iPad and iPhone as Teen Talk. That was in 2013. I so strongly believed in the verb-centric, process-intensive, visible-parsered approach that Chris Crawford has written about and I wanted to see it get its day in the sun so when the call went out for volunteers on Siboot 2.0 I signed up. I setup some project management infrastructure and I wrote the first draft of the Encounter Creation Guide with Chris Conley but, after several months on the project, I got tired of Chris Crawford's personality and management style and resigned from the team. That was in 2014 Bona fides aside, I look forward to the decade when Siboot 2.0 is released. I do not know how Chris Crawford's strategy of reaching out to young games will pan out. I will play the game though. I’m interested in seeing the new face technology in action. I’m interested in how the encounters are integrated into the 1-on-1 PC/NPC interactions. I am concerned how much of an impact it will make one way or another since it's a re-release of a thirty year old game that sold maybe 5,000 copies but I remain hopefully optimistic. I lean toward’s Alex Vostrov’s idea of integrating social interaction of Storytron/Sappho type into existing genres. I second his idea of limiting interactions to a single dynamic. I question his idea of “jettison[ing] as much of the current mechanics as [he] can,” since I believe it is these mechanics that will provide context for the conversations. Imagine a Balance of Power where discussions with your advisors happen in the environment of country and historical selections. Imaging a Global Dilemma: Guns or Butter where your conversations with other rulers incorporates promises about exports of commodities, tariffs, troop movements, and industry and economic goals. Even in Excaliber, Chris Crawford’s 1983 Atari game, social interaction was just one of several mini-games. Game mechanics are part of the world and part of the language. If it's all Vetvel Like Kendra DoublePlus Good then it's going to be quite boring indeed. Back in 2009, on the Storytron board, Chris Crawford started a discussion entitled “Should Storytron Attempt to Adapt to Games?” My response was an unequivocal “Yes.” Back then I wrote: “As much as I believe in the potential of the Storytron technology, I don’t think it will achieve a significant market presence in it’s current incarnation. Storytron's strength, it's focus on flexible interpersonal interaction through the Deikto interface, is adversely offset by its primitive output. While Storytron 'listens' and 'thinks' quite well I don't think it 'speaks' as well as it could. Emoticubes were a start, dynamic emoticubes are a direction, but to truly engage and move readers (players?), Storytron needs the ability to express itself procedurally in the most appropriate media whether it's words, images, sounds, animation, or video. Implementing this might dilute the Sapir-Whorf nature of Storytron but I think the potential benefits outweigh the consequences. The best way to integrate Storytron technology into existing games is to productize the Storytron platform. Storytron, Inc. doesn't need to create the storyworld equivalent of Doom or Quake, they need to create the dramatic engine that any developer can drop into their C#, Python, Ruby, Java, Objective-C, or JavaScript game.” This is what Inkle Studios has done. Their Ink platform is the engine behind their 80 Days and Sorcery products, supported by custom game mechanics tailored to the specific game situation. I personally am not a big fan of the Choose-Your-Adventure genre but I’ve played their games and even I can see the creative and monetary impact of their blending of game mechanic and story. Storytron/Deikto/Sappho (the current underpinnings of Siboot 2.0) is a programming language and the past attempts to make it palatable to poet-programmers or algebraic-scribes made it a lot harder to get actual work done and locked the platform into a particular form that might make acceptance difficult. My work on interactive storytelling has lessoned since 2014 but my interest is still high. I’ve enjoyed what I’ve read so far and I’ve even gone so far as to boot up my old copy of SWAT v3048 to review the work I attempted all those years ago. Who knows where that will lead to?
|
|
|
Post by alexvostrov on Nov 3, 2016 20:50:19 GMT -8
I lean toward’s Alex Vostrov’s idea of integrating social interaction of Storytron/Sappho type into existing genres. I second his idea of limiting interactions to a single dynamic. I question his idea of “jettison[ing] as much of the current mechanics as [he] can,” since I believe it is these mechanics that will provide context for the conversations. Imagine a Balance of Power where discussions with your advisors happen in the environment of country and historical selections. Imaging a Global Dilemma: Guns or Butter where your conversations with other rulers incorporates promises about exports of commodities, tariffs, troop movements, and industry and economic goals. Even in Excaliber, Chris Crawford’s 1983 Atari game, social interaction was just one of several mini-games. Game mechanics are part of the world and part of the language. If it's all Vetvel Like Kendra DoublePlus Good then it's going to be quite boring indeed. Over the last year, I've come to agree with this sentiment Bill. I didn't want to get into it, but I guess we're talking about it already. Specifically, I now believe that IS can't be just about people. It has to be about people AND things. When I talk about jettisoning mechanics, I do not mean 100% - just enough so that they don't take center stage. My ideal structure would split people/thing mechanics 50/50 and have some very heavy interleaving. I'm curious how you came to this conclusion yourself. Me, I arrived at it two ways. One - I've spent the last couple of years banging my head against the problem of the pure people game. Every time I'd get a solution idea that was "things", I'd say "not good enough" and toss it. This approach has been very restrictive. Every prototype I've made that has worked has made some concession to "thing" mechanics. The second (and more recent) line of reasoning is this. Why do people have conflict? Well, our desires bang into one another and a lot of the time you can't have it both ways. But is the reason why we have conflict a "thing" reason or a "people" reason? It turns out that most conflicts are driven by the mechanics of the world that we inhabit - things! Some conflicts are baked into us - status and mate selection will always cause people to butt heads. Pretty much everything else is a construct or a limited resource. Worried about the banker taking your house? Economic construct. Nuclear war? Enabled by technology. Geopolitics? Geo is in the name - it's spacial. You want to be King of Scotland? Rules of succession are a social construct. The world refracts our desires and flings them against one another. It turns out that pretty much any topic you will tackle is going to need THING reasons playing along with people reasons. Sometimes the thing mechanics can be very simple. Dream combat in Siboot or fighting in my prototype are like that. It can be just a dice roll. Still, the shape of the non-personal mechanics is CRITICAL! Getting this wrong can wreck the whole storyworld. Imagine if Chris's expectations about the dream combat are wrong. For example, what if the resource-limited rock-paper-scissors doesn't actually make information critical. Kabooom! The whole thing comes down. So to summarize, I think that we should almost approach the problem like MMO and tabletop RPG designers. We're making worlds for several AIs and one player. Oh, and we have to make the AIs behave properly. That's a pretty crazy burden to take on - no wonder we're stuck.
|
|
|
Post by billmaya on Nov 4, 2016 4:34:06 GMT -8
I'm curious how you came to this conclusion yourself. I think my first inklings of it were after I created Teen Talk, a port of Chris Crawford's Gossip, for iOS (it's free, available here for anyone who's interested). It had no replay value and even I got bored with it once the actual coding and release was done. It was a sketch, an experiment, but it got me thinking that this type of social mechanic needed some help because it couldn't stand on its own (no amount of combat or animated faces would have made me want to play it a second time). I did a blog post in 2014 entitled Entire Game on One Screen that sort of hints at combining social and other genre game mechanics towards the end–"As interactive storytelling matures I don’t know what final components will make up the complete experience but I have a feeling that the mechanics implemented in Teen Talk and Rumours will play a big part." So your thoughts made me think back to this blog post. Which made me recall the Inklecasts I've listened to where the authors/programmers discuss some of the same questions you raised and how they answered them in their 80 Days/Sorcery games, combining their Ink platform with a whole lot of Objective C and Java (their success in the marketplace so far validates their approach). Why do people have conflict? Well, our desires bang into one another and a lot of the time you can't have it both ways. But is the reason why we have conflict a "thing" reason or a "people" reason? It turns out that most conflicts are driven by the mechanics of the world that we inhabit - things! Some conflicts are baked into us - status and mate selection will always cause people to butt heads. Pretty much everything else is a construct or a limited resource. Worried about the banker taking your house? Economic construct. Nuclear war? Enabled by technology. Geopolitics? Geo is in the name - it's spacial. You want to be King of Scotland? Rules of succession are a social construct. The world refracts our desires and flings them against one another. It turns out that pretty much any topic you will tackle is going to need THING reasons playing along with people reasons. Sometimes the thing mechanics can be very simple. Dream combat in Siboot or fighting in my prototype are like that. It can be just a dice roll. Still, the shape of the non-personal mechanics is CRITICAL! Getting this wrong can wreck the whole storyworld. So to summarize, I think that we should almost approach the problem like MMO and tabletop RPG designers. We're making worlds for several AIs and one player. Oh, and we have to make the AIs behave properly. That's a pretty crazy burden to take on - no wonder we're stuck. I have to agree with this 100%. Time will tell. Chris will release Siboot someday. I'd also be happy to look at any of your current or future experiments or prototypes.
|
|
|
Post by chriscrawford on Nov 4, 2016 19:02:05 GMT -8
Wow! I forget to visit for a few days and you guys really take off. Maybe my presence has inhibited open discussion... Naw...
Some thoughts. First, the exuberant response to Choice of Games demonstrates the pent-up demand for creative opportunities in interactive storytelling. However, the algorithmic expressiveness of the technology is far too limited to permit much drama. This could mean one of two things:
1. People don't want to tackle serious algorithm design. They just want something easy to understand. 2. Having mastered the simple algorithmic expressivity of Choice of Games, they're ready for something meatier, like my technology.
So, depending on which interpretation you pick, we're either doomed or fated for greatness.
Second, I realize that we have an excellent halfway house for the technology in the Encounter Editor. Here's a greatly simplified form of the technology that people can get into easily. They can learn the basics, which will prepare them for the big jump to the full system.
One other factor to consider: I propose to reincarnate AlgoLizard, a feature in the Erasmatron that could easily be implemented in the current technology. AlgoLizard is a helper for building algorithms for use in scripts. AlgoLizard asked a series of simple questions:
What is the most important motivation for this act? What is the second most important motivation for this act? How much more important is the first motivation than the second motivation? Is each of these two motivations, all by itself, crucial? Or can one's strength compensate for the other's weakness?
I might prefer a different system for AlgoLizard that relies explicitly on the use of Blend, as in the current EncounterTester.
Between the use of the Encounter Editor as an introduction and AlgoLizard as a support, I think we can get people into the technology without much bloodshed.
|
|
|
Post by chriscrawford on Nov 4, 2016 19:29:46 GMT -8
I need to address the major issue of integrating the Storytron technology into the games universe. Indeed, I have to address the role of gamers in the development of interactive storytelling. To explain this, I first need to provide some background. The starting point is the notion of mental modules, which is now central to most concepts of evolutionary psychology. Here's the Wikipedia article on it: en.wikipedia.org/wiki/Modularity_of_mindHere's my own piece on it: www.erasmatazz.com/library/the-mind/history-of-thinking/033d91601d1f41bdbb9d/mental-modules.htmlThere's tons of material on this, and it is crucial to understanding my reasoning. The basic concepts of significance to this discussion are: 1. There's a mental module for social intelligence. It helps you understand how other people feel. 2. Gals are MUCH better at this than guys. 3. In fact, guys really suck at social intelligence. Next, let's look at some of the major mental modules: 1. Linguistic 2. Spatial/Visual reasoning 3. Natural history: understanding cause and effect in the real world. 4. Social reasoning. Next, let's note that guys are really good at #2 and gals are really good at #4. Now look at games. What mental modules do games challenge? Certainly not #1 and #4. Definitely #2, and some #3. What happens if you produce a game emphasizing #4? You're going to turn off the guys. They don't care about social reasoning. In fact, most guys (including some of the participants in this discussion, I'll guess) don't even realize that social reasoning exists. They perceive EVERYTHING in the universe through lenses that don't include #4. Hence, if you try to combine interactive storytelling with games, you're going to run into a collision at a fundamental level. The most avid gamers will be most dedicates to #2, and least interested in #4. They'll reject #4, because they're weak in it or don't even know it exists. And they'll therefore conclude that there's nothing of interest in the game. Mixing interactive storytelling and games is like making a Barbie doll that drives a Tonka truck. It's like a movie that is both action-packed and romantic comedy. Pearl Harbor kinda-sorta pulled that off. But that was, I think, the exception that proves the rule. How many other movies have made that combination click? Sylvester Stallone, Arnold Schwarzenegger, and Jean-Claude van Damme don't work as romantic leads. And how in the world are you going to integrate some testosterone-soaked violence into Pride & Prejudice?
|
|
|
Post by chriscrawford on Nov 4, 2016 19:39:39 GMT -8
I realize that some of you might be a little put off by my suggestions that guys don't get social reasoning. Let me soften the blow by pointing out my own slow learning process. When I was young, I was a social-intelligence idiot. My "social intelligence IQ" was down in the 50s. I still recall with much guilt the early years of my marriage, when I would logically prove that my wife's hurt feelings were not logically justifiable. Right. Thank god she didn't leave me. It took her decades to give me a basic education in social reasoning.
It didn't really start to sink in until I was in my mid-40s. A combination of many factors and personal crises opened my mind to the realities of social reasoning. And the more I learned, the more guilty I felt about how stupid I had been all through my life. I could write a large book entitled "Social Intelligence Bloopers of Chris Crawford".
I know very few men who have any inkling of social intelligence. Sure, the older ones are better at it, but I don't know how many times I observe a guy doing things that I was wont to do in my twenties and thirties, and smile inwardly. He'll figure it out in a few decades, if he gets kicked in the teeth often enough.
|
|
|
Post by alexvostrov on Nov 4, 2016 21:17:18 GMT -8
Mixing interactive storytelling and games is like making a Barbie doll that drives a Tonka truck. It's like a movie that is both action-packed and romantic comedy. Pearl Harbor kinda-sorta pulled that off. But that was, I think, the exception that proves the rule. How many other movies have made that combination click? Sylvester Stallone, Arnold Schwarzenegger, and Jean-Claude van Damme don't work as romantic leads. And how in the world are you going to integrate some testosterone-soaked violence into Pride & Prejudice? You're right to a certain extent, Chris. Yes, the stories that men gravitate to are much less socially and psychologically complex. Still, to say that men "don't get" social reasoning is mind-blowing from an evolutionary perspective. Our ancestors didn't survive by being social dullards - that's how you get your head bashed in with a rock. Look at the success of Game of Thrones - that's a show about political maneuvering and I bet that a good number of guys enjoy it, as well as gals. The difference is that men and women care about separate aspects of social life. Hierarchy and dominance matter more, for example. We can appeal to males by making games about themes that resonate with them. Guess what you need a lot of social intelligence for? Leadership. Negotiation. Building alliances and making friends. Status. We can make games about those topics right now. If you make a romantic comedy game that genuinely captures what romcom is about, my head will explode Chris. The depth of social reasoning there will need actors that are very very detailed. How else will you get someone to agonize over "Will he respect me the next morning?" with a computer on the other end? Here's a question that women answer in any relationship "Is this guy for real?" - i.e. is he just acting to get in my pants? Try to write an AI that can play that role without coming apart at the seams. That stuff is in our future, but I will be shocked if we can tackle it as step 1. Also, how do you see us making games purely about romance? I don't think that's doable - the problem is that you start tugging at the romance thread, you pull on status, respect and friendship, then you get to the sources of status - prowess, power, bearing, etc. Prowess can only exist in relation to concrete world outcomes. Before you know it, you have the entire world in your game. If the entire universe is in a glass of wine, it's certainly within two people falling in love. So far, I've not found a good way to stop the non-people factors from sneaking in. The best I can come up with is "Only simulate those aspects of the world that people already know".
|
|
|
Post by alexvostrov on Nov 4, 2016 21:32:02 GMT -8
One other factor to consider: I propose to reincarnate AlgoLizard, a feature in the Erasmatron that could easily be implemented in the current technology. AlgoLizard is a helper for building algorithms for use in scripts. AlgoLizard asked a series of simple questions: What is the most important motivation for this act? What is the second most important motivation for this act? How much more important is the first motivation than the second motivation? Is each of these two motivations, all by itself, crucial? Or can one's strength compensate for the other's weakness? I might prefer a different system for AlgoLizard that relies explicitly on the use of Blend, as in the current EncounterTester. Between the use of the Encounter Editor as an introduction and AlgoLizard as a support, I think we can get people into the technology without much bloodshed. Assuming that it's powerful enough, I suspect that some variation on this idea is the way to go. Building verbs by hand is the wrong level of abstraction - it's too low. In an ideal world, we'd tell the system what the verb does and it would generate a draft of all the inclination code for us. We don't know how to do that yet, so maybe a wizard is an acceptable kludge. My only worry is that maybe the AlgoLizard won't integrate well into Storytron. Imagine if you normally worked in machine code and you only had a compiler for the UI code. The compiler would spit out machine code that you'd later have to tie in with everything else. Would that work? How would it integrate with all the hand-written stuff? You should think about how everything is going to fit together.
|
|
|
Post by alexvostrov on Nov 5, 2016 9:32:26 GMT -8
|
|
|
Post by Chris Conley on Nov 5, 2016 9:33:02 GMT -8
Also, how do you see us making games purely about romance? I don't think that's doable - the problem is that you start tugging at the romance thread, you pull on status, respect and friendship, then you get to the sources of status - prowess, power, bearing, etc. Prowess can only exist in relation to concrete world outcomes. Before you know it, you have the entire world in your game. If the entire universe is in a glass of wine, it's certainly within two people falling in love. So far, I've not found a good way to stop the non-people factors from sneaking in. The best I can come up with is "Only simulate those aspects of the world that people already know". Sure, but this is perfectly reasonable to me; I don't think you can remove all things, and I'm sure that's not what Chris means when he writes about "games about people, not things". You can't have a "pure people game", any more than there are "pure people novels". No one writes pure talking head stories with no background or setting; the drama always needs some grounding in the physical and hierarchical reality of the world. Even Russian novels have crackling fireplaces, winter, rifles and governments. (N.B.: I've never actually read a Russian novel. I probably should.) Jockeying for position or honor or glory, fighting for the macguffin; they provide flavor and meaning to the story. And the design of Storytron recognized this, providing places and things. And there's no need for a storyworld to cover the entire spectrum of human experience; just limiting to a useful subset, with enough variations possible and clear causality with reactivity, would be enough. My only worry is that maybe the AlgoLizard won't integrate well into Storytron. Imagine if you normally worked in machine code and you only had a compiler for the UI code. The compiler would spit out machine code that you'd later have to tie in with everything else. Would that work? How would it integrate with all the hand-written stuff? You should think about how everything is going to fit together. The way it worked in Erasmatron was it just generated scripts, same as you could build by hand. And you could then edit the resulting script like any other. I think this simple-mode approach is a reasonable way to make development approachable, if the scripts so generated are variable enough to produce complex reactive behavior.
|
|
|
Post by alexvostrov on Nov 5, 2016 10:23:53 GMT -8
Sure, but this is perfectly reasonable to me; I don't think you can remove all things, and I'm sure that's not what Chris means when he writes about "games about people, not things". You can't have a "pure people game", any more than there are "pure people novels". No one writes pure talking head stories with no background or setting; the drama always needs some grounding in the physical and hierarchical reality of the world. Even Russian novels have crackling fireplaces, winter, rifles and governments. (N.B.: I've never actually read a Russian novel. I probably should.) Jockeying for position or honor or glory, fighting for the macguffin; they provide flavor and meaning to the story. And the design of Storytron recognized this, providing places and things. And there's no need for a storyworld to cover the entire spectrum of human experience; just limiting to a useful subset, with enough variations possible and clear causality with reactivity, would be enough. I was responding to Chris Crawford's analogy of a Barbie riding a Tonka truck. I think that we're in agreement that all stories are ultimately grounded in some sort of context. Take that out and you get Waiting for Godot. We can shrink the Tonka truck and pick what colour it is, but it's still going to be there. My point is that saying "How are you going to put romance into Rambo" is a red herring. Romance will not have rocket launchers, but it might have very light resource/economic dynamics if you want to depict a noble family fighting off creditors via marriage. That's not the dreaded resource management monster taking over IS. It's a necessary component.
|
|