Know who is on the ballot before you are standing in the booth
Turnout at municipal elections is low, and one reason people give is simply not knowing who is standing. The information exists, but it is spread across party sites of wildly varying quality, PDFs on municipal servers and printed notices. Meanwhile the commercial 'voting advice' sites that fill the gap are opaque about their funding and enthusiastic about tracking. An election information tool that harvests visitor behaviour is a bad answer to a real problem.
Candidates are rendered in the order and grouping of the official ballot, not sorted by relevance or by anything a ranking algorithm decided. The point is recognition: the layout you studied is the layout you see in the booth. This also removes the editorial judgement call of what to put first, which is exactly the kind of judgement an election tool should not be making.
Each election is a set of Markdown and structured data files in a repository. Every change is a commit with an author and a timestamp, so the edit history of what a candidate page said is public by construction. There is no admin panel to compromise and no database to leak.
The site builds to plain HTML. On election day, when traffic is concentrated into a few hours, there is no application server to fall over — only files on a CDN. It also means the whole site can be archived after the election as a straightforward snapshot.
No analytics, no cookies, no embedded fonts or third-party scripts. Which candidate a visitor spent time reading about is data that should not exist in the first place; the most reliable way to protect it is not to collect it.
The concept is documented and the data model is drafted; no public prototype is deployed yet. This page will carry the demo link once one exists.
This is not electronic voting. Nothing is cast, counted or transmitted here — the ballot stays paper, in a booth, exactly as it is. Digital Ballot only addresses the step before that: knowing who the names belong to. Conflating the two is how well-meaning election projects end up arguing about verifiability when the actual problem was that nobody could find out who was standing.