🎯 Desired Outcome
Two things, by July 24:
The Innovation Team is visibly seen as thinking outside of the box.
Most interns in the building know who you are. Most staff have heard about what we've built. The game is the proof that this team ships things people see as useful and actually enjoy.
By late July, walk into any room at IMB and say "I built this with AI."
Use Chipp.ai, Copilot, Claude, Claude Code, or whatever you have access to. We aren't talking about company data, so use whatever you want. If you need help using the AI tools for code, design, testing, debugging, etc., feel free to ask me. The game is the textbook — you'll have a real artifact to show.
Everything else in this brief serves those two outcomes. If something we plan doesn't, drop it.
📊 The KPIs I'll Judge This On
Track them. Show them on your admin dashboard. Bring them to our Friday standups.
1
Engagement
% of the other interns who play, by Week 4. Target: 80%.
2
Spread
Does it reach staff? How many non-intern players by Week 6? Do staff talk about it?
3
Visibility
At any point in the summer, can someone unaffiliated look at the Innovation Team and see something genuinely cool happening? Photos, stories, conversations in the hallway — all count.
4
Durability
Did the game stay alive and fresh through Week 8, or did it stall by Week 4? Most office games die in Week 3. Don't let yours.
5
Reflection
What did you each learn about (a) engagement design and (b) building real software with AI? You'll present both at the end of the summer. Bring the prompts, the architecture decisions, the surprises — the process, not just the product.
🦆 The Setup
You're going to design and run a game this summer for the IMB office. The Innovation Team is paying for the ducks, the website, and the prizes. You own everything else.
There are 72 tiny 3D-printed ducks — 24 yellow, 24 red, 24 green. They get hidden around the office. People find them. Points get scored. A leaderboard fills up. Something happens at the end. The shape of the game between those ducks exist and something happens at the end is what I'm handing to you.
The other ~15 interns will be your first audience. The ~200 staff in the building are the second. The site lives at duckhunt.donbarger.com.
🔒 What's Fixed (Don't Negotiate These)
LaunchMonday, June 1. Whatever you have on that morning is what launches. Ship rough; iterate weekly.
EndFriday, July 24. Finale event that afternoon.
Materials72 ducks to start — 24 yellow, 24 red, 24 green. Numbered Y-01..24, R-01..24, G-01..24. If you want more, make a case; I'll print more.
AudienceStarts with the three of you. Pulls in the ~15 other interns at the time of your choosing. Spreads to staff at the pace you set.
BrandingThis is an Innovation Team project. Your names and faces are on it. Don't be coy about who built it.
Budget$100 total for coffee, prizes, and swag. Spend it wisely — that's part of the design challenge. Run anything over $25 by me first.
Domainduckhunt.donbarger.com
Stack recommendation, not requirementFlask + SQLite. Matches everything else I run on the droplet. If you want to argue for something else, bring me the reason.
PhotosNo laptop screens, no whiteboards with names, no badges. Use judgment.
Where ducks goAt Monument or ILC. Try to stay in places where they'll be seen — shelves, monitor tops, break rooms, plant pots, the edge of a meeting-room whiteboard.
Where ducks can't goBags, jackets, lunches, drawers, locked offices, server rooms, mechanical spaces, restricted areas.
🧭 Design Principles (Use These When You're Stuck)
Photos beat stats. The home page is a window into the office, not a dashboard. Numbers are secondary.
The reveal is the dopamine. A find without a moment is just a form submission. Build the moment.
Visibility is part of the product. If the other interns can't see you having fun, you haven't shipped the game — only half of it.
Ship rough by June 1. Don't over-engineer Week 1. Week 6's version will be radically different. Plan to iterate.
Watch what people do, not what they say. Your admin dashboard will tell you more than any survey.
One rule that makes the whole thing run. Find the single rule that creates self-sustaining behavior. Don't bury it in fifteen sub-rules.
🎨 The Scope of Your Assignment
What you own (the design)
Game mechanics
Who scores what, how, when. How finds are verified. All scoring is individual — no teams. Side quests, bounties, weekly themes — your call.
The recruitment arc
When and how the other interns get pulled in. How spread to staff happens, if it happens. What the on-ramp looks like for someone who finds a duck and has no idea what's going on.
The story
Theme, lore, named ducks, or none of that. Whether the game has a personality or is just a leaderboard.
The artifacts
What you leave behind. Documentation, a video, a Hall of Fame — whatever makes this real after Week 8.
What we build together
The website experience. Home page, registration flow, photo display, what makes someone refresh the tab. You bring the design direction; we wireframe and build it side-by-side.
What I'll help with
Code when you need it. Bugs when you're stuck. Anything else — just ask.
How you work together
- No more than 1 hour per week on this outside our teaching sessions. Keep it bounded.
- All three of you contribute from your specific areas of expertise.
- Work as a team — not three solo projects in a trench coat.
❓ Five Questions to Wrestle With This Week
Bring me your answers Friday. I won't accept "still thinking about it."
How do you make the rest of IMB see you having a blast?
What's the visible artifact they encounter — photo feed, weekly recap, something else? Be specific.
How does the game spread without us emailing everyone?
Organic spread is the entire growth mechanism. Design it.
What keeps it fresh in Week 6 when Week 1's novelty has worn off?
Most office games die at Week 3. Why won't yours?
What does "winning" look like for a player? For your team running it?
If you don't define this clearly, the game has no shape.
How will you use AI tools in this build?
Don't default to "we'll ask Claude when stuck." Be intentional. Which decisions go to AI, which stay with you, what your prompts look like, what you'll learn from working this way. This is half of why I'm having you build it.
📅 Timeline
| When | What you owe me |
|---|---|
| Tue 5/26 | 30-min walkthrough of your initial direction. Not polished. Just your gut. |
| Thu 5/28 | Tech architecture decided. Backend stood up. A screen, however rough. |
| Fri 5/29 | Soft alpha — the three of you can register a fake find end-to-end. Everything else can break. |
| Mon 6/1 | Launch. Ducks hidden. Site live. Game on. |
| Each Fri after | 15-min standup: stats, what worked, what's next. |
| Late June (TBD) | Mid-summer review (end of Week 4). We assess KPIs together. Decide what to change. |
| Fri 7/24 | Week-8 finale event. Awards. Photos. Story. |
| Mon 7/27 | Your final presentation: what you built, what you learned, what's next. |
🛠️ Resources You Have
Me. Exec sponsor, build partner, and bug fixer. 15-minute Friday standups all summer. Final yes/no on anything that could embarrass us. Push me when you need a decision; I'll make it fast.
AI tools. Chipp.ai, Copilot, Claude, Claude Code — use whatever works for you. No company data is involved here, so don't worry about policy questions. Ask me if you want help getting set up or thinking through your prompts. At the end of summer I want to see the prompts you used — I'm building a library.
$100 budget. Coffee, prizes, swag. Run anything over $25 by me first; approvals are fast.
Each other. Three of you, one product. Decide who does what — based on your specific areas of expertise.
🚫 What I Will Not Do
- Design the game for you
- Pick your theme
- Recruit players for you
- Settle disagreements you can resolve yourselves
What I will do: build the website with you, help you write code, help you debug, make sure you have what you need, push back when I see something off, celebrate specifically when you ship something good, and give you the room to make this yours.
💬 One Last Thing
You're at the IMB this summer because the mission matters to you. This game isn't about missions — but how you do it says something. People watching you will see how you treat each other, how you handle setbacks, whether you tell the truth when something didn't work, whether you make space for the quiet person in the office to be part of something. That's the part of this you can't fake.
Build something we'll be talking about in October.
Let's go.
— Don
P.S. The previous prototype version of this page had cascade-scoring (1st finder gets the most points, descending) and a "you don't know your rank until after you submit" reveal mechanic. Both are keepers if you want them. But everything is up for redesign. If you can do better, do better.