During the 2026 Tamil Nadu Assembly election campaign, I watched an interview of a candidate from one of the parties that had risen to prominence in this cycle. She was introduced everywhere with the prefix “Dr.” and spoke at length about medicine, including the idea that astrological consultation should precede medical treatment. I remember feeling deeply uneasy listening to it. A doctor privately believing in astrology is one thing. A public figure, contesting elections, advocating astrology before medicine while carrying the authority of the title “Dr.” is something else entirely.
Only later did I realise that her doctorate was not in medicine.
The title had done the work it was meant to do. The ambiguity was never corrected because it did not need to be. Most viewers would assume what I assumed. What unsettled me most was not that misinformation existed. It was that even someone actively trying to fact-check what they consume could still get caught by the architecture of presentation itself. The system had worked exactly as intended.
The same election produced another post I have not stopped thinking about. A campaign graphic for a constituency candidate prominently displayed the words “first female candidate ever” alongside her name and “rising political force.” The constituency name and the party were visually subordinate, almost incidental. Every literal claim on the image was true. She was, in fact, the first female candidate fielded by that party from that constituency. But the impression generated was that the party had broken some broader feminist barrier, that this was a historic first in some larger sense was nowhere stated and so nowhere refutable. The image had been engineered to do its work in the gap between what was said and what was implied. A fact-checker reading the image carefully would find nothing to flag. A voter scrolling past would carry away an impression that no single sentence had earned.
This is the form modern political misinformation increasingly takes. Not lies, exactly. Selective truths. Compositions. The careful arrangement of the literally accurate to produce the substantively misleading. It is harder to catch than outright falsehood, and it is becoming the default register of sophisticated political communication.
Around the same time, I was watching something else happen around me. Political opinion among people my age was increasingly being formed through Instagram reels, meme pages, fan edits, WhatsApp forwards, and short-form political content. Older voters and even some politically aware people within my own generation responded to this with ridicule. Tharkuri, they called them. Ignorant. Gullible.
I never liked that response. Mocking people you disagree with has never persuaded anyone into thinking differently. But I also could not ignore what I was seeing being normalised online. Romanticised infidelity existed beside aggressive slut-shaming. Factually incorrect claims circulated beside emotionally charged propaganda. At one point, people were confidently repeating that welfare schemes would be funded from the Chief Minister’s personal income rather than through state expenditure. None of this always looked obviously false. The most effective misinformation rarely does. It survives precisely because it feels plausible enough to scroll past without questioning.
And that is what began bothering me.
When did we start forming political opinions from fifteen-second reels?
When did asking questions begin to feel more exhausting than simply believing what appeared on our feed?
And how did we slowly become comfortable consuming political content without demanding the most basic thing citizens in a democracy should demand i.e. evidence?
A few years ago, people my age used to make fun of our parents for believing everything they saw on WhatsApp. “WhatsApp University,” we called it, with a kind of generational confidence. We believed we were different. Smarter. More aware. Better at checking facts. Better at recognising manipulation.
Looking back, I think we misunderstood the problem completely.
We were never immune to misinformation. We were only immune to one format of misinformation.
The reel replaced the forward. The meme replaced the badly edited JPEG. The algorithm became more sophisticated. So did the presentation. And slowly, without really noticing it, our threshold for scrutiny weakened alongside it.
The danger of platforms is not simply that they contain falsehoods. Falsehoods have always existed in politics. The danger is that these platforms are designed to reward emotional immediacy and punish friction. Asking questions takes longer than scrolling. Verifying claims takes more effort than reacting to them. Over time, the medium trains us away from scrutiny itself. And once scrutiny disappears, everything else becomes easier like polarisation, propaganda, manufactured outrage, communal mobilisation, personality cults, and the gradual erosion of any shared informational reality.
This problem is structural. It exists everywhere platforms exist. The form it takes in any given place reflects the cleavages of that place. In Tamil Nadu, the 2026 election was won to a large extent through digital infrastructure like fan networks, reels, meme accounts, influencer ecosystems. The campaign did not end with the result. The same infrastructure was redirected, in the weeks after, toward defaming journalists, attacking opposition figures, and laying ground for new political mobilisations. The Chennai Press Club publicly flagged the proliferation of fake accounts impersonating media outlets. State machinery has since been used to take down opposition-aligned online accounts. None of this is unusual. It is the new normal of how political power is contested and consolidated in India, and increasingly in many democracies.
I have been thinking, alongside many others, about what could be done.
The intuitive answer : government-led fact-checking which has already been tried and struck down. In Kunal Kamra v. Union of India, the Bombay High Court ruled in September 2024 that the executive cannot constitute itself as the arbiter of political truth about its own conduct. This was the right judgement. A government deciding what is true about the government is the architecture of censorship, regardless of which party holds power. Karnataka’s recent attempts to draft a misinformation bill under an opposition-led government show that the failure is not partisan. It is structural. State-led fact-checking does not work in democracies, because the state has a conflict of interest in everything political.
The platform-led alternatives are not better. Meta exempts politicians from its existing fact-checking systems. The shift to Community Notes places the burden on users without addressing the underlying asymmetry. The Election Commission has done meaningful work on deepfakes and AI labelling, but its authority is bounded by election windows, and the most consequential damage now happens between them.
What is left, after all of these foreclosures, is a narrow space. Inside it, I have been trying to articulate a proposal.
I call it Know Your Citation.
The idea is borrowed, in spirit, from the KYC norms we have already accepted in other parts of public life. When we transact with a bank, we accept that the bank must verify who we are. We do not call this censorship. We call it a basic procedural threshold, because the bank handles something of public consequence. Political claims at mass digital scale handle something of even greater public consequence. The same logic should apply.
Know Your Citation does not adjudicate whether political claims are true. It only requires that the parties making those claims attach a source which is a publicly verifiable URL, a specified location within that source and that the source be visible to viewers. Posts that meet the standard carry a verification label. Posts that do not carry a “no source attested” label. Nothing is taken down. Nothing is hidden. What changes is that the gap between assertion and evidence becomes visible in the same frame as the assertion itself.
I have built a working prototype of this, alongside a longer policy document. The prototype, on real captions from real Indian political accounts, has already done something modestly diagnostic, it caught a recent post that claimed “the NDA secured 293 seats in 2024” and matched it against the party’s own cited source, an official government bulletin, which states 292. A one-seat inflation, surfaced not by a fact-checker’s judgement but by the speaker’s own evidence. That is the work the policy is designed to do. Not adjudicate truth. Make assertion and evidence visible together.
It is a narrow proposal. It does not solve hate speech. It does not solve deepfakes. It does not solve the algorithmic amplification that sits upstream of misinformation. It will not, on its own, catch the Virudhunagar campaign graphic I described earlier, because the misleading work in that image happens in the gap between literal accuracy and composed impression which is a gap a citation cannot close. Each of those problems requires its own intervention. What Know Your Citation does and the only thing it does is restore a procedural minimum to political communication at scale.
The reason I am writing about this now, rather than waiting until the proposal is fully developed, is that the conditions under which scrutiny becomes possible do not sustain themselves automatically. They have to be protected deliberately. And if the architecture of platforms helped erode those habits, then institutions, platforms, and citizens alike have a responsibility to begin rebuilding them.
But the heart of it is what I have just described. Political claims made on platforms that now reach more voters than any other public channel should carry the same procedural standard we already accept in every other domain of public life where claims have consequences. We have lost the muscle for scrutiny because the medium has trained us away from it. We can begin to rebuild that muscle, slowly, by building it back into the infrastructure itself.
Slowly. Deliberately. Before the erosion becomes irreversible.
Thanks for reading Praveena’s Substack! Subscribe for free to receive new posts and support my work.