A Hackathon Is a Demo Contest, Not a Coding Contest

A hackathon sounds like a race to write the most code in 48 hours. It is not. The judges watch a two-minute demo, and they remember what they saw working — not how many lines you wrote, and not how clever the code was underneath. The team that wins usually wrote less code than the team that lost. They just finished something a stranger can understand in two minutes.

That changes everything about how you spend the weekend. You are not proving you can code for two days straight. You are proving you can pick one small problem, build the smallest thing that solves it, and show it working. Keep that in mind and a first hackathon stops being scary — it is the fastest way to learn what building software feels like.

Friday Night: Pick an Idea You Can Demo

Most first teams lose the hackathon on Friday night, before writing any code. They pick the biggest idea in the room — an app that does everything — and spend Saturday discovering it does nothing. The fix is three questions, asked honestly, before you commit: does it solve a real problem someone actually has? Can a small team demo it by Sunday afternoon? Is it fun to watch for two minutes?

"An AI that writes your essays" fails all three: the problem is fake, a weekend cannot build it, and watching it is boring. "A canteen queue timer that tells you the shortest lunch line" passes: every student has waited in that line, one page with a timer is buildable, and the demo is a phone counting down in front of the judges. Ideas that demo well beat exciting ideas that never run.

Score every idea 1 to 5 on each question and add them up. The winner is rarely the idea you love most — it is the highest total. That is the point: the score protects you from your own enthusiasm.

Try It: Score Your Ideas Before You Build Any of Them

Here is the scorer. Paste it into the browser editor and run it — then replace the three ideas with your own and run it again.

# hackathon_ideas.py — score each idea 1-5 on three questions
# 1. Does it solve a real problem someone has?
# 2. Can we demo it in one weekend?
# 3. Is it fun to watch for two minutes? def score_idea(name, real_problem, demoable, fun_to_watch): total = real_problem + demoable + fun_to_watch return name, total ideas = [ score_idea("Study-buddy matcher", 5, 4, 3), score_idea("Canteen queue timer", 4, 5, 4), score_idea("AI that writes your essays", 2, 2, 2),
] best = max(ideas, key=lambda idea: idea[1])
print(f"Build this: {best[0]} — {best[1]}/15")
for name, total in ideas: print(f"{name}: {total}/15")

The canteen timer wins with 13 out of 15: a real problem (4), demoable in a weekend (5), and fun to watch (4). The study-buddy matcher scores 12 — good, but harder to demo live. The essay-writing AI scores 6 and should be dropped now, on Friday night, when dropping it costs nothing. The numbers turned an argument about whose idea is cooler into a decision you can all see.

Keep the scorer. When a teammate suggests a new feature on Saturday, score it the same way — if it cannot beat the demo you already have, it waits.

Saturday: Build the Smallest Demo That Works

Pick the one thing the judges must see, and build only that. For the queue timer, it is a page that shows a live countdown for each lunch line. Everything else — logins, profiles, a settings page — is decoration, and decoration is what kills demos. Fake what you must: hard-code the queue lengths for the demo. Judges judge what they see, not how real the data behind it is — and you can say so out loud in the demo.

Work in the smallest steps you can, running after each one, the way the plan-first habit teaches: one step, one run, one saved version. When something breaks at 2 a.m. — and something will — you want to know it was the last step, not somewhere in fifty lines you wrote without stopping. Save a version every time the demo works again, so a broken step costs one restore instead of one rewrite.

Split the work the way pair programming splits it: one person drives the code while the other watches for mistakes, checks the demo script, and keeps the list of what still needs doing. Swap every 25 minutes so nobody goes stale. A team of two to four works best — small enough that everyone knows the whole demo, big enough that one person can rehearse while another fixes the last bug.

Sunday: Rehearse the Demo, Not the Code

Stop coding two hours before judging. That sounds early, and that is the point: the last two hours belong to the demo, not to one more feature. Write a two-minute script with three beats: the problem in one sentence, the demo working in front of them, and the one thing you would build next. Then run it with a timer, out loud, three times — the first run will take four minutes, so cut words until it takes two.

Rehearse the failure too. Decide now what you will do if the Wi-Fi dies or the demo crashes: keep a phone video of it working, or screenshots in order, and keep talking. Judges forgive a crash they were warned about; they do not forgive a team that stares at a broken screen in silence. And sleep — tired presenters forget their own script.

After the Weekend: the Demo Becomes Your Portfolio

The hackathon ends on Sunday, but the demo keeps working for you. Save the working version as a snippet so the code does not rot in a forgotten folder — it is now the strongest piece in your portfolio, because it is a real thing you shipped under a deadline, with a story: the problem, the score, the demo, the result. That story is exactly what interviewers ask for, and it is what your GitHub portfolio should show.

Take the same idea further in Lab: the Build Products with AI course walks the same arc — find a real problem, specify it, build it, launch it — so the weekend project can grow into something real users try. And the next time someone asks what you built, you will not describe a tutorial you followed. You will show them a demo.

When the Free Plan Runs Out Mid-Weekend

A hackathon weekend burns executions fast: run after every step, all weekend, across a team of three or four — and Free gives each of you 20 executions a day, 10 AI requests a day, 10 saved snippets, 3 private snippets, and 3 collaborators per snippet. That fits a careful weekend. But if your team keeps hitting the daily limit on Saturday afternoon, that is the upgrade moment: Pro is $9/month or $69/year — unlimited saved snippets and executions, 100 AI requests a day, 10 collaborators per snippet, unlimited private snippets, and GitHub integration, which Free does not include. Team ($19/user/month, or $15/user/month billed annually) covers a class or club entering together — see the pricing page.

Your next step is tonight, before any hackathon: run the scorer on three ideas of your own, save the winning idea as a snippet called "hackathon idea", and send it to one classmate you would want on your team. Then plan the demo the way you would plan any program — one sentence, small steps — and build the flashcard quiz next as practice: a whole project in one sitting, finished and demoable.