I recently started developing a game in my spare time, and discovered Ink, which has tremendously simplified development of conversations and managing conversational state.
I've found Ink to be quite flexible, If someone wanted to, they could manage their entire game state with Ink.
Live recompilation of my ink script in Unity will be very helpful!
I wrote my "portfolio" page[1] in a choose-your-own-adventure style with Ink. Documentation would be pretty cool, might experiment with that at some point.
Because ink is a type of markup language, using it to write narrative is very easy. You're writing the script for your game and ink will handle how parts of it connect to eachother. It is also very portable owing to being plaintext with relatively simple syntax. There are currently ports of it in both Unity and Unreal, and many developers have used it in some capacity for html/css/javascript games. Procedural generation can be done in ink, however you may need to use the frontend language to save them for later; ink does procedural text very well but has a weak concept of objects.
Inform7 is a declarative programming language with english-like syntax. It's quite good compared to ink at simulating objects with a lot of metadata. However, the runtime isn't quite as portable as ink; as a result, Inform7 has no easily located Unity or Unreal packages. Ordinarily, I would recommend Inform7 where you need to process lots of content and randomly generate things. But it seems in addition to having an arbitrary runtime entity limit of 1000 things, those things must exist before you randomize them. Although counterintuitive, procedural generation is more challenging in Inform7 than ink!
Use Inform7 when you need a language that can reason "because Alice's relation to Bob is at -50, Alice hates Bob. Because she hates Bob and he is walking nearby, she can trip him. If she does, he will take a random amount of damage and Bob's relation to Alice is reduced by 20."
Use ink when you need a highly portable language that can script /other/ frontend frameworks or languages. Ink has been used in 2d games, 3d games (notably Heaven's Vault), on the web, and on three different game engines (Unity, Unreal, Godot). Ink is also very good at making variations on text and tracking how often parts of the story have been visited.
They're both excellent tools for interactive fiction, but they excel at wildly different things.
Wow! Thanks for that response. I will check out Ink. Back in the 1980’s, I wrote a choose your adventure system based on plain text. I rewrote the system a number of times to learn new languages. Probably been 20 years since I touched my system. Maybe I will try using Ink with some of my narratives. Ink is obviously servers like orders of magnitude more powerful than anything I ever built.
> A race is a kind of person. Dwarf, elf, and half-elf are kinds of race.
> A town has numbers called friendliness, services, comfort, and population. Leadership relates a person to a town. The verb to lead implies the leadership relation. "F [friendliness] / S [services] / C [comfort] / P [population]".
> Rivalry relates various towns to each other. The verb to be rivals with implies the rivalry relation. The verb to be rivals of implies the rivalry relation.
Then you can make a town like this:
> Bremen is a town in Ten-Towns. "Founded by dwarf prospectors, the sleepy town of Bremen sits on the west bank of Maer Dualdon, at the mouth of the Shaengarne River.". It has friendliness 3. It has services 1. It has comfort 2. It has population 150.
> Dorbulgruf Shalescar is a dwarf. Bremen is led by Dorbulgruf Shalescar.
And now inform7 can be used to query or act on all this metadata:
> Every turn when the player is in a town:
repeat with town running through towns:
say "Rivals of [town]: [list of towns that are rivals with town][line break]";
This is the kind of thing you have to work hard to represent in ink because you only have functions, variables, and lists (a hashmap<variable, boolean>). As you add more metadata to inform7, it can make EXTREMELY sophisticated queries or conditions like "list of weapons that were in hidden rooms which are gilded". So for very little work, you can have inform7 reason about whether a player can 'wear' a specific item by checking what they're wearing, how the item affixes to things, and what attachment points you've defined (which can be as broad or granular as you want). And as a bonus, it's also very little code to comment on the player wearing a strange thing (like pants on their head) or wearing colours a specific character thinks clash with eachother.
This kind of query is one of ink's notable weaknesses (and requires quite a lot of lists or very clever lookup functions to achieve). But, ink can leverage its frontend (often a general purpose programming language like C# or javascript) to provide that functionality.
Ink isn't a programming language, it's a markup for scripts that branch a lot or have text substitution rules. The FFI means that if a project hits ink's limits, you can go beyond them. Despite being a markup, ink is very close to turing complete. XML is another type of markup that relies much more heavily on external languages. So I think ink is quite powerful. (at least enough for most interactive fiction!)
I haven't played with either of them much, but they seem to take two fundamentally different narrative approaches.
Inform is primarily simulation driven: You describe a world model. Every user input translates into some action that affects the world, and the world changing produces relevant output.
Ink is primarily narrative driven: You write the text as a branching story, with a set of explicit options for each branching point. Variables can be read and changed arbitrarily, but have no meaning other than how the narrative uses them.
I am familiar with inform and zork (which auto-corrects to dork ;-) but not with how that is different than branching narrative software. How are they different or similar?
Branching narratives are like Choose Your Own Adventure books. You can get 80% of the way to what Ink does with a carefully programmed static website. But this is an environment designed to author that, and to publish to game engines as well.
The main competition is Twine, which is more focused on a self-contained deployment. Twine also got a huge update recently with its "Harlowe" theme supporting storylets - scenes that trigger from a pool or "deck" based on logic you program.
zork and other "parser" games would say something "you are in a room with a table and a door to the north" and there would be a cursor.
Twine and "choice" games would say something like "you are in a room with a table and a door to the north. Do you want to look at the table or go to the door?"
On the software side, ink is open source and has C# and JavaScript compilers, a Unity plugin and a separate IDE called inky.
Inform7 is not open source yet, although they had plans in 2019. As far as I can tell, it's a single app that outputs its own format that can be uploaded to https://www.ifarchive.org/
Disclaimer: I've never used either, ink just seemed more legible.
To add some detail, Inform 7 compiles to reasonably well-documented portable binary formats that have been standardized for decades, with the earliest versions dating back to Infocom days. There are a variety of emulators, including web-based. This is why games uploaded to ifarchive many years ago still work. Even if an old game’s source were available, they were sometimes written in obscure genre-specific programming languages that nobody uses anymore.
The binary formats are specific to the genre, which is pretty niche. Casual users are unlikely to have downloaded an emulator even if it’s easy to do, so a game will be bundled with one when it’s published to an app store.
But as formats go, it seems like a pretty remarkable success story for the long-term preservation of games, and it would be neat if there were a similar file format that were more general-purpose and more popular, much like we have for images, music, and video.
Very cool. I’m working on an interactive fiction engine in my spare time, and it’s interesting to see how someone else chose to manage the complexity that exponential branching starts to cause. Their concept of knots makes a lot of sense for a video game with narrative “hubs”. Definitely like the text based approach, the flow graphs that other engines use look like they get really messy really fast.
or kotobee or Twine which is from the Interactive Fiction Technology Foundation. .. and used to make this cool interactive browser experience burnt matches. more here"
I think that means we always have the choice in life to just open the door and look around and chose to stay home and not start anything. There should be a snow with a slight glow to the BG colour.
I absolutely love the ink team, for releasing this. They have a fairly decent Unity which is free and is more than good enough for creating basic narrative games. It would still probably be a good idea, to have a good grasp of unity before integrating it though
Is anyone doing something similar , or using this as-is, with comics and graphic novels? Is this a different form of narration when it is applied to video games? I'm curious if there are discussions about how this is different or the same.
Very cool. I wonder if there is an avenue for using this with something like GPT3; be interesting perhaps to see what it might come up with, or how it might add speed/scale to the process in some cases?
That sounds amazing! Where would be a good place to start digging into chatbot tech? Specifically, I'm wondering how to dynamically rewrite scripted dialogue and it is a deceptively deep topic.
There are two types of chatbots -- unreliable freeform ones (e.g. GPT-3) and reliable script-based ones (e.g. Alexa). I believe that what you're looking to do (dynamic rewriting) is moreso an active research topic!
The script-based chatbots are very similar to those in the 80s, with one major addition -- the options at each choice point are now "intents" instead of exact text (e.g. "the user intends to order a taxi"). Machine learning is then used to classify whatever the user says/writes as an "intent," then the script is followed to the next choice point.
Maybe I'm just old at this point, but I honestly thought this was a reference to Inkscape which had it's v1.0 about a year ago, and I became very confused.
Just wanted to quickly add that I am in no way trying to steal the thunder from this news, just referencing where my brain had gone.
All that aside, this looks like an incredible language to use for some of my own projects. I have always been a fan of text based adventures, and have played MUD's off and on for at least 25 years at this point. Maybe I can put this language to good use. :)
Ink – Scripting language for interactive stories - https://news.ycombinator.com/item?id=26147329 - Feb 2021 (15 comments)
The Intercept (2012) - https://news.ycombinator.com/item?id=23244812 - May 2020 (4 comments)
Writing web-based interactive fiction with Ink - https://news.ycombinator.com/item?id=21794981 - Dec 2019 (9 comments)
The Intercept - https://news.ycombinator.com/item?id=21538414 - Nov 2019 (1 comment)
Ink – inkle's narrative scripting language - https://news.ycombinator.com/item?id=16925588 - April 2018 (22 comments)
Ink: the scripting language behind 80 Days and Sorcery - https://news.ycombinator.com/item?id=11269782 - March 2016 (18 comments)