I am proud to have launched Frequency on the last day of 2018. It is the first game I created from start to finish as project lead. I’ve worked in games for eight years now, but always on someone else’s project or as part of a larger team. Frequency was the first time I was in charge of the production, and I am immensely proud — and rather surprised — that it was ever finished.
Frequency started as a simple project intended to take a few months work, but it got away from me, ballooned out of proportion and sat dormant for over six months.
When I started the project in early 2018 with the help of Braydon Beaulieu, under the title “Natural Language”, we wanted to make a short, contained ARG with a distinct set of restraints. We had a clear idea of the game flow and production rollout. But as so often happens with personal projects, I jumped into the details of the work that I enjoyed most, creating complex puzzles and a detailed visual aesthetic before we even had a basic gameplay loop tested.
As I developed the puzzles, the list of features needed for the game’s central website increased, as did the playtime and difficulty level. What began as an ARG contained within a single website and email with a 1-3 hour playtime became a website, two auto-response emails, plus audio and video content with a playtime of 10-15 hours. Continuing to add features and content felt like progress, since I was working with no time constraints and had no sense of my skills (or lack thereof) in website development.
In April, life caught up to the project, and suddenly I had no free time on my hands. The game, which was perhaps a third finished based on the new elaborate plans, sat untouched for months. I would occasionally consider finishing it, but now in the abstract, unspecified future. In late November 2018, I was looking back on my year, and looking ahead to future projects, and I decided to finish Natural Language before the end of the year. That way, I could start the new year fresh.
When I first sat down to reread all the content, I was expecting to find scattered scraps, barely held together with a barebones narrative arc. Instead, I found most of the puzzles complete, drafts of every single web page written and nearly all visual content ready and waiting. About 80% of the game was finished, with the remaining sections simply marked with “need something here” or “figure this out” notes. Taking the time to reread all the content and reacquaint myself with the puzzles, I realized that the game could be quickly and easily finished up, simply with a little editing and removal of all the extraneous components I’d added on mid-production.
The first thing I cut were the email accounts. For the sake of simplicity, I removed all external communication and had the player focus exclusively on the website, receiving and solving puzzles in the same place. A convoluted gameplay loop with multiple possibilities for things to go wrong was dropped in favour of a single delivery method for everything.
I then reviewed the puzzle rollout and the difficulty ramp. My original plans for a 1-3 hour game involved about eight puzzles, ranging from very easy to mildly difficult. The working draft had 15 semi-complete puzzles starting simply but quickly ramping up to an intensely difficult plateau. I cut all the hardest immediately and looked to combine elements of others together to form a more gradual difficulty grade. Most importantly, I reintegrated all puzzles within the context of the narrative, removing anything that was there for its own sake. In the end, there are nine puzzles in the final game.
The story itself needed to be simplified as well, along with the means in which it was delivered. Once again, I removed all external content and focused on the website text and images alone (except one small excerpt that had to be external for narrative reasons). I simplified the plot to fit the more linear medium and removed secondary characters to focus on the relationships between the two AIs and the player. The final product was less grandiose than I had initially intended, but that was probably for the best. Braydon did a final pass on all the written content, and then it was ready to integrate into the site.
The final hurdle was the true interactive elements between the player and the AI. While the simple answer delivery system worked well for the majority of puzzles, there were certain story points which required a more interactive method. Wix, which I used to build the website, had limited options for chatbots or other user text input, and I know next to nothing about website coding, so I turned to the one bit of interactive software I knew well – Twine. I wrote mini Twine files with password boxes that displayed custom messages based on the input, and hosted them in nestled windows within the main larger webpage of the game. It worked like a charm. And with that, I was ready to put the pieces together and launch the game, on December 31st, 2018.
I learned a lot from this project, more than I had imagined I would. Most importantly, I learned the importance of sticking to my own constraints and project guidelines, treating them as though they were imposed by an external force. It is far too easy to continue expanding a game and adding new features and content without first solidifying the basics, and this can get out of hand quickly. The second thing I came away with was a understanding of the importance of distance. When I was deep into production, I lost track of the size of the project and it seemed infinite in scope. Therefore, when I had to stop working on the game, I still had it in my head that the project was huge, and therefore returning to work on it would require monumental time commitment. It wasn’t until I made myself return to the project that the smaller scale and ease of completion became apparent. The entire process has me invigorated and ready to start my next project based on the feedback I receive from the game. Please let me know what you thought of Frequency. You can can reach me here.
Thanks for playing!
All images from Frequency, made using Pixel Sorting tools by Diego F. Goberna