Litter Wars: turn Basel into a live team-vs-team playing field for real cleanups

What game would make someone pick up litter they'd otherwise walk straight past — and then go out and do it again next weekend?

Cleancentive already handles the boring half. You photograph what you picked up, a vision model names it, estimates its weight and puts it on a public map with your team and your cleanup attached. What the platform cannot do is make you care on day three. Impact statistics are a scoreboard nobody plays.

So build the play layer. Not a badge bolted onto a form — a game: a loop, an opponent, stakes, and a reason to be outside. Steal shamelessly from what already works. Pokémon Go's reason to walk to one specific corner. An MMO's guilds, seasons and progression. Valorant's round tension and a scoreboard you can read in half a second. Strava's segment rivalry.

Four constraints that make this interesting rather than easy:

• The map is real. Hotspots, streets and cleanup locations come from actual Basel data, not a generated grid.
  The score is real.
• Every point has to trace back to a real photo of real litter, stored in our database, visible on the public map.
• It must survive a cheater* Any rule strong enough to motivate is strong enough to game. Tell us how you'd cheat, then show us what stops you.
• It runs in the street. Cold hands, gloves, one thumb, patchy signal, late October in Basel.

Why now
Cleancentive was born at this hackathon. At BaselHack 2025 Johannes posted a challenge called "Cleanup Application", because he was picking up litter while jogging and had hit the ceiling of a no-code Bubble prototype. Team Delitter built the thing. Twelve months later it is a real platform: a PostGIS litter map, an offline-capable PWA, an ML detection pipeline, teams, recurring cleanups, a documented REST API, AGPL-3.0 on GitHub, live at cleancentive.org with real users and real cleanups in the Basel region.
And it has one problem that more backend code will not solve. People try it, log a handful of picks, and stop. The word "incentive" is in the product's name and it is the single feature that has never been built — no points, no streaks, no leaderboards, nothing. This needs you!
Why us
We're bringing back to BaselHack the thing BaselHack built, one year on, with working infrastructure, a year of real data and an honest gap in the middle of it. You will not be prototyping against a mock API and a Figma file. Matthias Keys-Cullmann, and Johannes MasterMind-Bohren are on site with you the entire weekend to code, ship new API endpoints, chase litter and win the BaselHack. ;-)

A person holding a phone in one hand and a litter grabber in the other, standing on a Basel street.

Concretely:

• Ploggers and runners: who already pick things up, and currently get a spreadsheet's worth of thanks for it.
• Cleanup regulars — Pfadi groups, neighbourhood associations, school classes, the people who show up to organised clean-up days.
• People who would never sign up for a cleanup, but would absolutely sign up to beat another team. Teenagers, students, gamers, competitive runners. This is who the game is actually for. Everyone else is already convinced.
• Companies and clubs who want a team day that isn't a trust fall.
• Cleanup organisers, who need something that makes people show up — and then show up again.

Downstream: municipalities, schools and partner organisations who use the resulting map and the data it produces.

1. A standalone client — web, mobile, whatever you like — that a stranger can open and understand within thirty seconds. Your repo, your stack. You are building an MVP on top of Cleancentive, not rebuilding Cleancentive.
2. Team versus team, with a live scoreboard** that visibly moves when a pick lands. If it isn't competitive and it isn't live, it isn't this challenge.
3. It talks to the real API. Picks made in your game are real records in Cleancentive and appear on the real map. Reading our data is not enough — the game has to contribute.
4. Proof outdoors. At least one genuine litter pick in Basel during the weekend, played through your solution, shown in your Sunday demo. Real litter, real coordinates, real data in our database. Go outside.
5. A scoring model with a stated threat model. Say out loud how someone would cheat your game, and show what you did about it.
6. It works in the street. Phone-first, usable one-handed with gloves on, and it doesn't fall apart when the signal does.