I recently watched Dhruv Rathee's video on civic sense and empathy. He makes a strong case for how systems, infrastructure, and empathy all play into why some countries feel more "civilised" than others. I agree with most of it.

But I keep coming back to something that gets glossed over in almost every conversation about civic sense in India. There's a layer underneath systems and empathy that's way more fundamental. What happens when people can't even receive the instruction?

India's official literacy rate is about 81%. Sounds decent, right? Here's the catch: the census doesn't actually test anyone. An enumerator visits households and asks the head of the family whether each member can read and write. That's it. Self-reported, never verified. A PRB (Population Reference Bureau) study found that when you actually test people classified as "literate" by this method, only about a quarter could read in practice. The gap between official and functional literacy in India is probably 15-30 percentage points depending on how you define "functional".

ASER's 2024 data is even more telling. About 55% of Class 5 students in rural India can't read a paragraph meant for Class 2. These are kids who've been in school for five years and still can't process basic written text. This is the population that signage, rules, labels, and government notices are meant to communicate with.

And then there's the script thing. India doesn't just have many languages, it has 12+ completely distinct writing systems. A person who can read Tamil cannot read Devanagari. Someone literate in Bengali can't process Gujarati script. These aren't dialects of each other, they're entirely separate visual systems. Compare that to the EU, where 24 languages mostly share the Latin script, or Switzerland, where four languages all use Latin. Even Singapore (top of my mind given apparent travel) has three scripts (Latin, Chinese, Tamil) but English functions as a shared working language across everything. India doesn't have that luxury at the same scale.

Think about this. A truck driver from Tamil Nadu picks up a load in Gujarat. They can read Tamil. Maybe. The road signs are in Gujarati, Hindi, and English. None of those scripts mean anything to them visually. Anything textual like lane markings, speed limits and warning notices -- beyond basic traffic signs -- are all invisible. They are navigating a state where the entire written communication layer might as well not exist.

India has roughly 60 million interstate migrant workers (Economic Survey 2017 estimate, probably higher now). Sixty million people routinely living and working in states where they can't read the local script. This is on top of the functional literacy gap within each language.

Every developed country that people point to as a model for civic sense designed their rules for populations that can read. Lane markings work because people can read "SLOW". Safety signs work because people can parse a sentence. Metro instructions work because text comprehension is assumed. That assumption falls apart in India at a scale that's hard to overstate.

I know this reads like a list of reasons India can't improve. The opposite. The point is that India's problem requires fundamentally different tools, and probably more design effort than developed countries ever needed, because the communication challenge is that much harder. You can't copy-paste a Swiss road safety programme and expect it to work in a country where a third of the population can't functionally read and even the ones who can might not read the script on the sign in front of them.

FinalMile, a Mumbai-based behavioural design consultancy that's now integrated into Fractal, ran a project on railway tracks that I think about a lot. They identified that people were dying between stations, on the tracks, because they could see the train coming and were overconfident about crossing in time. Classic risk-perception failure.

Their fix was simple. They painted railway ties in alternating yellow stripes. As a train approaches, the stripes create a visual perception that makes the train look faster than it is. They also changed the warning whistle pattern from one long blast to two short bursts separated by silence, because the brain responds more to sounds punctuated by silence than continuous noise. No signs. No text. No language dependency.

Result: 80-90% reduction in fatalities at the 22 locations where they deployed this. The cost? Buckets of paint and a whistle instruction change.

Similarly, 3D zebra crossings in Ahmedabad, where painted optical illusions make drivers perceive a raised barrier, reportedly led to zero pedestrian accidents at four of the city's worst spots in the six months after deployment.

India has a NITI Aayog Behavioural Insights Unit since 2019, set up with the Gates Foundation. But it's basically an advisory team embedded within NITI Aayog, running training workshops and small campaigns. Compare that to the UK's Behavioural Insights Team with 200+ staff running large-scale experiments. India has a nudge unit for 1.4 billion people that operates at a fraction of the capacity. If FinalMile's paint-and-whistle intervention gets an 80-90% fatality reduction and costs almost nothing, why hasn't Indian Railways scaled it across more stations? State-level fragmentation?

Here's where it gets interesting. India has close to a billion internet users and roughly 490 million are on YouTube. The country's non-literate populations haven't been excluded from information, they've bypassed text entirely. YouTube, WhatsApp voice notes, video-first consumption in vernacular languages is the actual information layer for hundreds of millions of Indians.

And the infrastructure is starting to catch up. Bhashini, a government platform, provides AI-powered translation across all 22 scheduled languages. Jugalbandi, built by AI4Bharat (IIT Madras), Microsoft Research, and backed by Nandan Nilekani, is a WhatsApp chatbot that gives people information about government schemes in their local language through voice. UPI 123PAY lets people make digital payments using voice prompts on basic feature phones, no smartphone or internet needed. Hello! UPI takes it further with conversational voice payments in regional languages.

The irony is wild. The government still communicates through gazette notifications, printed forms, and written circulars. Meanwhile, the population has moved to audio and video. There's a format mismatch between how India's government talks and how India's people listen.

Civic rule communication could leapfrog signage entirely. Voice-first, video-first, vernacular-first delivery. The tech exists. Some of it is government-built.

Rwanda does something fascinating. Umuganda is mandatory national community service, last Saturday of every month, for every able-bodied citizen. You physically show up and do community work: building roads, cleaning neighbourhoods, repairing schools. Local leaders give instructions verbally, tasks are demonstrated physically. Civic participation is encoded in practice, not text.

You don't need to read a rule if you physically do it every month.

Different mechanism entirely, but the same underlying principle: design for the population you actually have, not the one you wish you had.

Rathee's framework, caste attitudes, systems, and empathy, is right. But all three assume the message reaches the person. Before empathy kicks in, before systems can shape behaviour, there's communication. And communication in India is a fundamentally harder problem than most people acknowledge. Multiple scripts, massive functional literacy gaps, hundreds of millions of migrants navigating in languages they can't read.

The question is whether India is willing to invest in solutions designed for this specific reality or whether we keep importing templates from countries where everyone can read the sign.

Because the tools exist. Behavioural design that bypasses text. Digital platforms that speak every language. Community participation models that work through action, not instruction. The bottleneck is scaling them. And maybe taking the literacy barrier as seriously as we take the "empathy" conversation.