On the evening of 20 July 2026, the Rapid Action Force cleared unarmed student protesters off the road near Jantar Mantar with pellet guns, tear gas and spike-studded batons. The march had been heading to Parliament to demand the Education Minister’s resignation over the NEET-UG paper leak. Within an hour of the crackdown the photographs were everywhere. Within a week they would begin to disappear, as placards were thrown away and images were taken down from timelines under pressure, and the visual record shrank one screenshot at a time.
Democracy by Design started that afternoon with a phone call between Pallavi Nayak and me. Pallavi runs Arts and Culture within the All India Professional’s Congress, and she had been looking at the same photographs I had from a different angle. She saw the joy in the record: the sarcasm written on cardboard, the Hindi slogans shaped like memes, and the women who turned out in numbers nobody had predicted. She also saw that this record was about to be lost.
The website went live in about two hours.
What a protest archive has to refuse
The design constraint was simple. Most of the work went into deciding what the archive would refuse to do.
We refused to identify protesters. The Jantar Mantar photographs contained faces the state had every reason to want, and a public archive that kept those faces without consent would be doing part of the state’s work for it. The pipeline obscures faces automatically before anything is published, and manual review catches the edge cases. Anyone who finds their own face in a photograph and wants it removed can use the site’s documented takedown process.
We refused to become a news outlet. Journalists were already covering the violence. What was disappearing was the art, meaning the placards and the photographs of them. So we kept the scope narrow: posters, placards and photography. The archive does not carry video, commentary or the news of the day.
We refused to filter for ideology. The archive accepts non-violent placards criticising the government from anyone who sends them, including from political quarters we would not usually publish. A collection of posters that only agree with the curator would be a slideshow and could not serve as an archive. The only filter is content that incites violence, and everything else goes in.
We refused to build a narrative. The essays that accompany the visual record explain each placard’s social and political context in the plainest language we can manage. They do not celebrate or try to rouse anyone. They describe what is on the placard and leave the reader to look at it.
The two-hour part
The technology was nearly the smallest part of the work.
We registered the domain (democracybydesign.in) in the first fifteen minutes. The backend went up in the following hour. By the end of the second hour most of the site was in place: the submission form, the anonymisation pipeline, the metadata verification, the cross-reference against public sources, and the takedown request process. The editorial guidelines took longer. We wrote them together in a shared document over the following days, and they govern everything the site does.
We say “two hours” because it is true, and because it pushes back against the habit of treating every civic tool as a six-month project. Some tools need to ship as fast as the news cycle that created the need for them. If Democracy by Design had launched in November, the placards from 20 July would already have been forgotten.
What verification can and cannot tell you
A person checks every submission. The system then runs EXIF and metadata checks, looks for signs that the image has been altered, and searches for versions published elsewhere. After that a moderator makes the decision.
None of this is foolproof. A determined bad actor can fake metadata, and a photograph that appears nowhere else can still be a convincing fake. The verification chain makes it expensive enough to get a fake through that most attempts fail early. It cannot make the archive certain. What it shows is that the archive is looked after with care.
This matters because a public archive that claims certainty can be turned against the people it shows. An archive that states openly what it can and cannot verify, and puts the safety of contributors ahead of the completeness of the record, behaves like a witness and does not pretend to be a court.
What the archive is about
Pallavi said it in her interview with The Hindu: the archive is about “what we choose to remember”.
The 20 July placards would not survive on their own, because nobody was going to keep them. The cardboard would go out in the rain. The word “Resign” spray-painted on a rain-soaked street in Kolkata would have been walked over by Monday morning, and the photograph of it being sprayed is the last version of it that will exist.
An archive that collects these is not neutral about the future. It is a bet that the memory of how the protest looked will matter to somebody, three decades from now, who was not there: the joy alongside the fear, the humour alongside the violence, the handwriting of the person who wrote koi bhi sarkar ho, jawabdehi honi chahiye. If that person is a fourteen-year-old in 2056 trying to understand what her grandmother did in her twenties, the archive is for her.
If you were at Jantar Mantar, Nagpur, Kolkata or any of the other marches this summer, and you have a photograph on your phone that is not on the site yet, please submit it. The verification is real and so is the takedown process, and the archive keeps material that nobody else is going to keep.