Gold Machine

Please take a moment to fill out this brief questionnaire.

Pardon Me, but Have You Heard of the Interactive Fiction Competition?

My readers come from varied backgrounds and and have different reasons for following Gold Machine. Some are digital humanists, interested in the history of interactive media. Others are new media types, who enjoy interpreting games games along multiple axes. A lot of people are here for nostalgia’s sake. Some participate in a small DIY arts scene often referred to as the “interactive fiction community.” There are many overlaps among the groups. I write from all of these perspectives, and many readers likely read from them, too. These aren’t camps or sides, they are lenses or points of view.

The interactive fiction community, or scene, makes and plays games that it collectively (with some detractors) considers “interactive fiction.” Defining the term is a bit of a time waster. People generally agree that interactive fiction involves interaction and text, but there are so many edge cases that going further can provoke fights. In reality, despite this lack of clarity, it is very rare for any specific work to provoke a disagreement.

A lot of what goes on is based on vibes, and I mean that in a good way. A community is, among other things, a vibe or series of vibes.

There’s a lot that goes on to maintain a scene that exists to make and play games. There are people who maintain the systems authors use to produce their work. I work with the programming language Inform, for instance, and a number of people do hard work to keep that going. Still others–and some of the same people, too–maintain systems for executing game files. Discussion spaces must be maintained and operated.

It takes a lot to keep this interactive fiction thing rolling. There are also competition and jam events held the year round, where authors can showcase their works and sometimes compete for clout or prizes. These events all have their own special character. Spring Thing, for instance, is known for welcoming experimental work from varied and diverse perspectives. Ectocomp is the “spooky”-themed event held every October.

The largest and oldest of these events is the Annual Interactive Fiction Competition (IF Comp). Founded in 1995, IF Comp has its own character too: it is big and important. This importance is borne out, materially, in terms of entry counts, review counts, and rating counts. So many more people rate and review IF Comp games than they do works released elsewhere that it might be fair to say that authors pay an “IF Comp tax” for releasing their works in other events.

People who presumably do not play or talk about interactive fiction for most of the year return, a sort of migratory species, for IF Comp. The event sometimes garners attention from creators and media types who do not otherwise engage with interactive fiction. For some people, all that they know of interactive fiction is what they see during the competition.

IF Comp is run by the Interactive Fiction Technology Foundation (IFTF), a non-profit organization responsible for, most visibly, the Narrascope conference. There is a committee responsible for running the competition, which features ten members. Additionally, persons without committee status perform various volunteer functions. It is a big operation. I’m grateful for the services of all who contribute to the operation of these events.

For the past two years, works created using large language model technologies have been accepted as entrants in the competition. There is, naturally, an affirming normalization to their inclusion. After all, this is the big, important event, so important that it might be a person’s only window into the state of the community. The controversy surrounding generative AI robs IF Comp of much of its conversational oxygen. LLM-generated content has adversely affected outside perceptions of the competition. It is my position that its presence disrespects authors, players, and reviewers. It disrespects the competition itself.

This year’s competition has ended, and a post-event survey is available. I am told that, unless a sufficient number of people comment on LLM content, well, anything could happen. We could have an all-LLM event, including LLM judges and reviewers. The committee might even ask LLMs to complete next year’s survey. Even though it’s a bit like telling someone to quit smoking cigarettes, I suppose there is nothing else for it: please take a moment to complete this brief questionnaire. After all, your opinion is important to us.

Luddites vs. the Environment-Destroying Plagarism Bot

Look, it’s a waste of time to relitigate what LLM content is and why it’s bad. My experience is that people who advocate for LLM technology already know what’s wrong with it; they just aren’t moved by those concerns. Let’s speedrun through the bullet points.

• LLM technology is bad for the environment. It just is, OK? LLM-housing, water-and-electricity-guzzling data centers are so large as to defy imagination. MIT, which sees potential in the technologies involved, recognizes the concerns. They have become a Mortonion hyperobject, something like “all the styrofoam in the world.” A common argument against the environmental case is that people who commute to work use more energy as individuals than they might using LLMs. Yes, working to eat and live is exactly as important as using LLM technology to extrude bad cover art for a middling, fantasy-themed text game.

• LLM is a plagarism machine that runs on the work of uncompensated and often uncredited human beings. This is how the technology works. LLMs do not have “ideas.” They are not creative. Instead, they are the world’s largest sausage grinders, transforming human endeavor into extruded substance. This is what it does. Displacing humanity is its defining feature. The persons who feed it receive no material rewards, despite the billions of dollars that have gone into creating the technology. To value what is generated is to devalue what is made.

• LLM is not a fun toy for you to make games with, it’s a billionaire-backed tool that will obsolesce knowledge work and enforce ideological consistency in the content it generates.

Let’s Be Honest; Ethics Aside, This Stuff Is not Worth Looking at Anyway

I guess it takes all kinds, but I’ve never seen generative AI produce good craft prose. I’ve never seen it produce a good image. Everything is middling at best, and some of it is far worse than that. This has been discussed at length. I’ve sympathized with critics who have tried to engage with these works in good faith, detailing what they perceive as weaknesses and strengths, but I think it’s time to take another tack. I do not want to read these reviews any more than I want to play the games. Even if we ignore the ethical issues–it must be easier than it looks–we are ignoring them for the sake of mediocre sludge that wastes the time of players, organizers, and critics.

Since player attention ultimately is a zero-sum game–there are only so many hours in the day, after all–these works deprive human creators and their works the attention of critics and players. In fact, the mere presence of LLM-generated work in a space can completely take over the discourse, forcing supernaturally polite persons to repeat the same valid concerns regarding generative AI again and again while enlightened centrists are, as the expression goes, “just asking questions.” Why are we still doing this? It’s a waste of time and transforms the nature of the event in negative ways.

Consider the situation on the Interactive Community Forum, a small space within the IF community. In the midst of the Annual Interactive Fiction Completion–the big, important one–the busiest and active thread isn’t about games. It doesn’t contain reviews. It’s about generative AI in the competition. The technology is a spoiler in so many senses. It is genuinely off-putting to see the characteristically unremarkable LLM-generated artworks listed among the competition entries. It is off-putting to see the discourse commandeered by the subject. It’s exhausting to explain again and again to people who don’t care.

I know I’m breaking my own rule here, and I hopefully this won’t again be needed. But please stop reviewing LLM content in events. Stop playing them. Stop making threads about them.

Stop entering your own work in events that platform LLM technology. Stop playing games in events that platform LLM technology. Stop reviewing games that share a platform with LLM technology.

Just stop. All of it.

Please Enter Your Comments in this Field, and Leave Your Survey in this Survey Collection Box

I know, I know. A lot of these convesations have happened at the Interactive Community Forum. It’s maintained by the IFTF, just like IFComp is. Some of the Competition Committee members are regulars there. Those conversations will not be considered regarding future LLM-related decisions. I’m not sure if forum members on the committee must recuse themselves from the decision process, or how that’s supposed to work, but the overwhelming amount of public acrimony toward LLM-generated content has no bearing on whether or not the public feels acrimony toward LLM-generated content. You must leave your comments here, in this form.

Don’t forget! Your comments must go on this form! This one, here!

You may think you’ve talked this thing to death, but it’s time to slug it out one last time.

Your Opinion Is Important to Us

It is down to us! Only we have the power, should we choose to use it, to respect authors, players, reviewers, and events. Why it is down to us is an interesting philosophical question that lies beyond the scope of this helpful reminder.

Please complete this brief survey.

Next

Moonmist.