The victory claims were audacious, even amusing.
“I’ve created a powerful algorithm and it is operating at 99.1%… Let this email message serve as my claim to the prize for translating the Indus Valley Script,” read one of the many messages that flooded Rajesh P N Rao’s email in the months after the Tamil Nadu government’s January 2025 announcement of a $1 million reward for anyone who deciphered the Indus script.
With no mechanism in place yet for claimants to register their ‘discovery’, many have been writing to Rao, a Seattle-based computer scientist who is arguably among the most public faces in a race to solve one of mankind’s most enduring mysteries – decoding the Indus Valley script.
“The first two months were intense. I must have received some 200 emails, claiming all kinds of fascinating and delusive decipherments of the Indus script,” says the 56-year-old Rao, who has been trying to decode the script for the last two decades.
Among those who wrote to Rao were a mother-daughter duo in South America, who argued that the Indus script shared a common origin with the writing system of an Indigenous Latin American community. Another claimed the signs represented a celestial “address” — a coordinate-based naming system rooted in the astronomical traditions of the third millennium BCE. Others ventured even further into the improbable, proposing that the script encoded altered states of consciousness or mystical knowledge.
Since the announcement of the discovery of the Indus Civilisation in 1924, the script has remained one of history’s greatest uncracked codes, inspiring generations of linguists, archaeologists, computer scientists and enthusiastic amateurs.
Then in January 2025, when Tamil Nadu Chief Minister M K Stalin announced the $1 million prize for decoding the script, the interest got renewed. The announcement followed a study by the state’s Archaeology Department that claimed similarities between signs on Indus artefacts and graffiti marks recently unearthed at archaeological sites in Tamil Nadu, bringing the century-old puzzle back into academic and political debates.
With that, the code crackers went back to work.
The interest in the script has evolved over the course of an entire century. On the one hand are those such as Finnish indologist Asko Parpola and Canadian epigrapher Bryan Wells, who have dedicated their lives and careers to decoding the script; on the other are those who have recently turned to artificial intelligence to crack the problem that has eluded human effort for so long.
Archaeologist Pallavee Gokhale, who has been studying Indus artefacts and the scholarship around the script, says attempts to decipher the Indus script have always been shaped by the intellectual and historical context of the time. When the script was first discovered in 1924, decipherment efforts drew heavily on prevailing archaeological theories. “As the Aryan migration theory gained prominence, those working on the script became increasingly interested in identifying the language of what they considered the indigenous people of India,” Gokhale says.
She believes that unlike many other problems of ancient history, the Indus script is in the form of data that is conducive to computing and statistical models. “Today, with artificial intelligence, machine learning and data science at the forefront, many believe these technologies can finally help solve the problem, ” says Gokhale. “However, the real test lies in truly integrating the archaeological context and explaining the same, interpreting the signs,” she adds.
Sundar Ganesan, director of the Roja Muthiah Research Library (RMRL), which hosts the Indus Research Centre that’s backed by the Tamil Nadu government, says, “We have received at least 60 decipherment claims in the last year and a half, and we tell them to approach the government since we do not have any locus standi on this yet,” he says.
The Indus Research Centre is creating tools to help people examine the script scientifically.
⠀
Long before AI became the loaded buzzword that it is today, Rao had in the early 2000s turned to computational methods to understand the corpus of symbols that would throw light on “who we are as Indians, and who our ancestors were”.
A school-time fascination with the Indus Civilization stayed with him even as he pursued his undergraduate study in engineering in the US and went on to do his doctorate in computer science from the University of Rochester.
Later, during a visit to India, he got to know that a group of scientists at the Tata Institute of Fundamental Research (TIFR) in Mumbai were trying to decipher the ancient Indus script using tools such as statistical analysis, machine learning, and information theory — methods that would be alien to historians of ancient India, but were most familiar to him. For Rao, it was the best opportunity to use his expertise in unraveling the world of ancient India, one that had had him hooked since he was a child.
Nisha Yadav, Science Communication & Publication Officer at TIFR
Among the team working at TIFR was Nisha Yadav of the Department of Astronomy and Astrophysics. She studied Physics and Computer Science, and was part of a research group at TIFR that was studying astronomy in ancient India, working on the instrumentation of a satellite telescope, when she stumbled upon the various theories about the Harappan script. And then, much like Rao, a fascination that had stayed in the recess of her mind since Class 5 became a full-blown obsession.
“These theories about the script lacked an objective approach,” she says in an interview to The Indian Express. Computers, she believed, could interpret the script, without giving in to human biases.
She and her colleague Mayank Vahiya decided that instead of finding meanings in the signs, it might be better to find patterns and sequences. “The idea was to find out what is the probability of a particular sign following or preceding another sign,” says Yadav, who is currently Science Communication & Publication Officer at TIFR.
Later, a few other independent researchers, including Rao, Hrishikesh Joglekar, and Ronojoy Adhikari, joined the TIFR team to take the project ahead.
Once the computer had identified the statistical patterns in the sequence of Indus signs — measured using a concept from information theory known as ‘entropy’ — the researchers compared these patterns with those found in known languages, including Sanskrit, Tamil and several tribal languages, as well as in non-linguistic systems such as DNA sequences and Western classical music. They found that the statistical structure of the Indus script more closely resembled that of natural languages than non-linguistic sequences, suggesting that it likely encoded language, even if the language itself remains unknown.
“We can write in the Indus script, but we cannot read it,” Yadav explains.
Bahata Ansumali Mukhopadhyay, a Bengaluru-based computer scientist.
Sometime in 2014, Bengaluru-based computer scientist, Bahata Ansumali Mukhopadhyay met Adhikari, who was part of the TIFR team working on the script. Mukhopadhyay was then desperately looking to find an intellectual cure to her corporate boredom. The Indus script was the perfect outlet.
“There is a romance in solving a problem from ancient history,” she says.
She soon moved on to research the script independently, all the while continuing with her day job. As she immersed herself in the world of Indus signs — in the process, losing out on sleep and social life, and even postponing key medical treatments — Mukhopadhyay used skills and tools from archaeology, phonetics, linguistics, and mathematics and statistics, as and when whatever was required. “I had the right questions, and I did whatever I needed to prove the answers,” says Mukhopadhyay, who recently concluded that the script does not phonetically spell words of any language.
While Mukhopadhyay’s conclusion has been published in multiple scientific journals, and she has received recognition from celebrated archaeologists and linguists, she has not bothered to find out about the process to claim the prize money. She is, however, glad that the announcement has made many more people aware of the script.
Louie Merriam, a high-school student based in Washington DC, and Berlin-based geodesist Andreas Fuls.
The youngest to jump onto the bandwagon of Indus script decipherers is 17-year-old Louie Merriam, a high-school student based in Washington DC who grew up alongside a lot of “desi” friends, and is fond of watching Bollywood movies. He found himself drawn towards Indian culture and history and consequently found a passion project in the Indus script. He says, “As a student, I have time and can spend time on leisurely activities like the Indus script.”
Currently, he is assisting Rao in the decipherment project. “We are using AI to investigate how each sign behaves — whether it is a noun or verb, for example,” Merriam explains.
He concedes that AI could be influenced by the human biases fed into it, but he tries to make the program as blind as possible. “I don’t discuss much about the script with AI. Instead, I just provide it with the data I have collected and ask it to point out the statistical patterns and anomalies,” he says.
Merriam says that the announcement of the million-dollar prize had prompted many of his friends in America to embark on deciphering the script. He says most of them were first- and second-generation Indian migrants who are adept at computers and STEM subjects, but have very little interest in their own history. “After a few excited hours, they quickly realised that the script was hard work, and moved on,” he says.
Merriam stuck on. “I would much rather use my computer science skills on the Indus script, than on finance spreadsheets.”
⠀
Quantitative and statistical methods, it seems, guided even an earlier generation of scholars — whether cultural historians, epigraphists, or linguists — in their attempts to unravel the mystery of the Indus script.
Many first encountered the script while working on other ancient writing systems, only to end up dedicating their entire lives to the study of Bronze Age India.
Such was the case with Bryan Wells, a 75-year-old Canadian epigrapher who first encountered the Indus script in 1988 while working on Maya hieroglyphs as an undergraduate student of archaeology at the University of Calgary. The ancient script of the Mayan civilisation in Latin America, however, was overcrowded with researchers, prompting Wells to turn towards the ancient Indian script instead.
There was very little work done on the Indus script at the time, except for the corpus of signs that had been built by epigraphist and civil servant Iravatham Mahadevan and Asko Parpola. Wells says he soon realised that the corpus had several defects. “Until we had a proper computer-based database, it was pointless to try to decode the script,” he says.
Computers, he says, were absolutely necessary for decoding the Indus script simply because of the exorbitant amount of detail in the signs — there are about 5,500 texts, more than half of which are on seals, the small, square steatite objects found from Indus sites. “Every artefact is talking about a different subject matter,” he says.
At the time, it was rare for those working on ancient scripts to make use of computers. Wells says that even with Mayan hieroglyphs, which had attracted huge monetary investments, a computer-based database was created only as recently as 2005. “There was no proper computational software either that could hold the database of an ancient script,” he says.
Wells spent the first few years documenting every tiny scratch of Indus signs, from site reports to photographs, and every online mention. He needed a software to record the signs and, in 2006, he reached out to Andreas Fuls, a Berlin-based geodesist (scientists who measure the geometry, gravity and spatial orientation of Earth), who at that time was researching Maya archaeoastronomy. Fuls was building a software for recording ancient undeciphered scripts.
Since 2008, the two have built one of the largest databases of the Indus script, comprising 30 tables and 1.4 million data points.
Over the last 30 years, during which he devoted himself to the script, Wells lost his eyesight to diabetes, but he kept working anyway.
Despite giving it his all, Wells believes one can never decipher the script 100 per cent. “That is because there are at least 300 signs in the corpus that occur only once,” he says.
About the recent attempts to decipher the script using AI, he says it would be a futile attempt unless one knows the language in which it is written.
A script, he says, is considered deciphered only if an entire community of archaeologists, philologists, and other specialists believe it has been deciphered. “Now, what is the probability of a hyper nationalist from north India agreeing that the script is Dravidian?” he asks. “There will always be dissent, and the Tamil Nadu government would never have to really cough up the million dollars.”
⠀
Shortly after Stalin made the announcement of the prize money, Bharat Rao, a Texas-based computer programmer in his mid-fifties, claimed to have cracked the script. He first reached out to CM Stalin on X, then wrote an email to the minister in charge of science and technology in the then DMK government, but received no response.
“They announced a prize, but no mechanism or evaluation method has been put in place all these months through which one can submit a claim,” he says.
Rao was locked down in Puerto Rico during the Covid-19 lockdown, when he read about the century-old efforts to crack the script. Thereafter, he found himself sleepless across several nights, obsessing about finding the truth behind the script. “It was as if our ancestors were speaking to me,” he says in a phone interview from Texas.
Using a mathematical model called cryptography, Bharat Rao, in a matter of six months, reached the conclusion that the script encoded Vedic Sanskrit. The seals, he claims, carry religious significance and spell the name of Rudra and other deities frequently. There are also stories from the Puranas inscribed on the artefacts.
Bharat Rao’s claim is yet to receive scholarly approval, which he believes is unsurprising. “It is natural for scholars to react with shock and disbelief, since most of them are pursuing their own methods,” he says. He argues that he has the proof of “mathematical correctness”.
Despite this and other claims, scholars working on the Indus Civilisation say all decipherment attempts are an exercise in futility unless backed by archaeology.
UK-based Indus archaeologist Adam Green argues that the Indus script can be understood only in the backdrop of the archaeological context in which it appears. “For instance, in Mohenjodaro, we find a lot of script on copper tablets, which is not the case in other Indus sites where the script may only appear on seals, or still others, where pottery graffiti is more common,” says Green.
Rajesh P N Rao agrees, saying there needs to be genuine collaboration among epigraphers, historians and those using AI and quantitative methods.
“Right now, everyone is just working on their own, with their pet theories, refusing to talk to others. Most are after some personal glory, or perhaps a million dollars,” he laughs.
Reporting for this article was conducted by Adrija Roychowdhury during her fellowship as a Journalist-in-Residence at the Max Planck Institute for the History of Science, Berlin.
Louie Merriam; Andreas Fuls; Nisha Yadav; Bahata Ansumali Mukhopadhyay. 
Seattle-based computer scientist Rajesh P N Rao
⠀
Canadian epigrapher Bryan Wells
⠀
Nisha Yadav, Science Communication & Publication Officer at TIFR
Bahata Ansumali Mukhopadhyay, a 
Louie Merriam, a high-school student based in Washington DC, and Berlin-based geodesist Andreas Fuls.
⠀
⠀
Share
