Silicon Crossroads – Own Your Every Decision

Play the game

Play Silicon Crossroads on itch.io

I’d love for you to play it before reading the rest of this post. It takes about 10 minutes, and knowing how it works ahead of time would spoil some of the surprises. Every playthrough can turn out differently, so feel free to play it more than once.

Content notes: academic pressure, academic dishonesty, burnout, financial stress and loneliness.

Overview

Silicon Crossroads is a short, branching interactive story about a first-term Stanford student. You’re here on a scholarship that needs a B average, you work shifts at Coupa Café to pay rent, and you’ve been thinking about a career in social media content creation with your dog, Mochi. Over one ten-week term, AI, your phone, a smartwatch and a shot at going viral all compete for your time, and every choice costs you something.

The game follows three things about you, shown in words rather than numbers: Knowledge (what you’ve really learned), Wellbeing (how you’re holding up) and Network (the people in your corner). Some risky choices come down to a die roll, with odds shaped by those scores. At the end of the term, you find out whether you kept your scholarship, then choose your future at the Crossroads. There are nine possible endings.

How the idea came about. Being a first year Stanford graduate student exposed me to the realities of balancing  graduate life. My starting point was a simple question: what happens to a student who uses AI for their assignments compared with one who does the work? In my first version, the student who used AI would reach a job interview and fail it, while the one who did the work would get the job. I also wanted the game to show that life isn’t black and white: a student who skips the studious path could still get lucky, and the uncommon choice doesn’t always make you the loser. Through playtesting, the game grew from “AI versus no AI” into something broader: balancing student life when technology as a Stanford student. Deadlines pulling at you from every direction.

Branching choice map

Game Branching Map

The biggest branch happens in Chapter 1. How you handle the essay sends you to one of three different Chapter 2s: writing it yourself, using AI as a tutor or letting AI polish your draft leads to The Midnight Scroll; copying with AI leads to The Dreaded Email, where your professor asks you to explain your essay; and filming a video with Mochi instead is a die roll that either goes viral (Going Viral) or flops (back to The Midnight Scroll).

Chapter 3 has two versions. Most players reach Midterm Week, but anyone caught bluffing, faking their drafts or cheating with a smartwatch is summoned to The Hearing instead. Everyone meets again on The Last Saturday before finals, where several good things compete for one free day.

At The Crossroads, you learn whether you kept your scholarship, then choose your future. Everyone can apply to TechNova or enter the pitch competition. Two more options only appear if you earned them: going full-time as a creator if you built a following, and joining a friend’s startup if your network grew enough. Your scores, your record and a final die roll decide which of the nine endings you reach.

History of versions

If you’d like to see how the game evolved in more detail, I documented each stage on the class blog:

• Silicon Crossroads: Playtest Observations (October 19): the first paper prototype and Rachel’s playtest.
• P2: Refining Silicon Crossroads and Playtesting Outside Class (October 21): Rupert’s playtest and the move into Twine.
• P2: Silicon Crossroads, Playtesting and Refining, Part II (October 26): Game’s and Maia’s playtests, and the changes they led to.

All playable old game versions 1 – 9 can be found here. Below is a summary of every version

Version 1: paper prototype (October 19). Four chapters with three options each, and a score card where players added up Knowledge, Wellbeing and Network. The point values were visible on every card, and the endings were a job interview, a lottery or streaming. Played by Rachel.

Version 2: refined paper prototype (October 20). Refined from Rachel’s feedback and played by Rupert outside class.

Version 3: first Twine version (October 21 to 23). Rupert’s feedback reshaped the game, so I moved it into Twine to handle the bookkeeping. Consequences were hidden until after you chose, and how you handled the essay now sent you to one of three different Chapter 2s, with a hearing for anyone caught cheating. I added outlandish options decided by dice, rewinds with a cost, options you unlock through earlier choices, a backstory that gives every option a reason, and scenes that open by remembering how you felt about your last choice. Played by Game and Maia.

Version 4. Game and Maia didn’t understand what their scores meant or what the goal was. I replaced the numbers with words (Knowledge from Shaky to Sharp, Wellbeing from Burnt out to Thriving, Network from On your own to Well connected), put the grade weight in every scene, made dice odds depend on your scores, gave the Coupa shift a guaranteed reward, replaced the lottery with a campus pitch competition, let players type their name, and changed the subtitle to “Own your every decision” to match the second-person writing.

Version 5. On my own playthrough, some options still made no sense, so I checked every option against one test: would a real student in that moment consider this? Confusing options were replaced (for example, a bet with friends became handing your phone to your roommate until after the exam), and I simplified the writing and the look. Played by Jeffery on October 26.

Version 6. Jeffery’s good path felt too easy (“Wow. I wish life was this easy”), and he cheated safely because he still had two rewinds. Every choice now has a drawback, outcomes show the effort and what you missed, and Rewind became “Change your mind,” which costs more after a risky choice and can’t undo getting caught. Played by Emma, Jeffery and Yuina on October 28.

Version 7. Long scenes now reveal a beat at a time as you click, the missed movie night is set up before you choose, the dog was renamed from Pixel to Mochi to avoid confusion with Google’s phone, the daily channel now costs wellbeing, every roll says which score affects the odds, and the startup odds are tougher.

Versions 8 and 9: current version. Following Lucas’s feedback, the opening screen gives a playtime instead of a linear chapter diagram, the dice and locked options explain themselves when they appear, and the Instagram option shows the temptation growing rather than joking about it. Most importantly, the game now has an overarching goal: keeping a scholarship that needs a B average. Your scholarship status is shown at all times, every chapter raises the stakes for it, and the Crossroads reveals whether you kept it.

Why I’m convinced the game is better now. The clearest sign is in the endings. My early playtesters mostly landed in the same place, because the best choice was easy to spot: Rachel picked option B in every chapter and walked straight into the job. By the last round, every session ended differently. Emma took a risk on the pitch and ended up “Rich in Other Ways.” Yuina’s startup was turned down. Jeffery became a full-time creator in one run and lost the pitch in another.

People also wanted to keep playing. Jeffery played twice on Monday and three times on Wednesday, just to see where different choices would take him. And the choices feel heavier now. Jeffery said there was always “some kind of drawback one way or another,” and Yuina told me the game felt “too real, too similar to my regular life.” The one thing I haven’t tested yet is the scholarship goal, which I added after Maia and Lucas pointed out that the game had no overall objective.

Playtests

I ran eight playtests with seven players between October 19 and 28. The full records for every session, including videos, demographics, detailed feedback and timestamps, are in my Playtest Records folder. Here’s the short version, with one player per session:

• Rachel, Oct 19 (female, LDT and game design classmate; paper). Picked option B in every chapter, showing the best choice was too obvious. What changed: options now conflict, and consequences are hidden until you choose.
• Rupert, Oct 20 (male, LDT ’27, outside the game design class; paper). Asked “Can I do five minutes of each?” and wanted branching, outlandish options and a way to undo. What changed: the move to Twine, with branching chapters, dice and rewinds.
• Game, Oct 21 (male, LDT and game design classmate; Twine). Couldn’t tell what his scores meant, and felt a lottery didn’t fit Stanford. What changed: scores became words, and the lottery became a pitch competition.
• Maia, Oct 23 (female, LDT classmate; Twine). Didn’t understand the goal and played to win. What changed: the goal is stated up front, every scene shows its stakes, and later the scholarship became the overarching goal.
• Jeffery, Oct 26 (male, game design classmate; version 5). Found the good path too easy, and cheated safely because he had rewinds left (7:21). What changed: every choice has a cost, and Rewind became “Change your mind.”
• Emma, Oct 28 (female, LDT classmate; version 6). Liked that every choice has a cost, but missed a movie night the scene never mentioned (6:57). What changed: trade-offs are set up before you choose.
• Jeffery, Oct 28 (returning playtester; version 6). Said the game felt harder and more balanced (6:02), and suggested click-through beats (11:53) and renaming Pixel. What changed: scenes reveal a beat at a time, and the dog became Mochi.
• Yuina, Oct 28 (female, game design classmate; version 6). Called it “too real” (5:52), but felt the startup odds were too generous (10:48). What changed: every roll explains its odds, and the startup is harder to win.

Reflection

What I Did: I took a simple idea, a student choosing between using AI and doing the work, and turned it into a branching interactive story about student life. I started on paper, moved to Twine, and went through nine versions and eight playtests with seven players in about two weeks.

What I Learned: Players will game any system they can see. Rachel picked the obvious best option every time, and Maia played to maximize her score, so hiding the numbers and giving every choice a cost mattered more than adding content. An option only works if a real student would consider it in that moment, and the scene has to show why.

Small details carry a lot of weight: a dog called Pixel made one player think the game was sponsored, and a B-minus that was meant as praise read as a bad grade. I also learned to let the theme change.

The game started as “AI versus no AI,” but my players kept telling me it was really about balance, and the game is stronger for listening. Finally, playing with the same person twice was some of the most useful testing I did, because Jeffery could tell me directly whether my fixes worked.

Lessons for Next Time: I’d define the goal and the central conflict before writing a single scene, instead of adding the scholarship intricacy later on. I’d write each scene’s stakes and time limit first and the options second. I’d record every playtest from the start, including the early paper ones. And I’d test each round of changes with fresh players before stacking more features on top, rather than building several versions between playtests and then simplifying.