On this page
Few education experiments have a better origin story. In 1999 a computer was set into a wall in New Delhi, children with no prior experience gathered around it, and they learned to use it on their own. The Hole in the Wall project made Sugata Mitra famous. It also set off a debate about how much children can learn without teachers that is still useful for students and parents today.
What happened in Kalkaji?
According to the project's own history, Mitra, then chief scientist at the IT training company NIIT, had been toying with the idea of unsupervised learning with computers as early as 1982. On 26 January 1999 his team "carved a 'hole in the wall'" between the NIIT premises and an adjoining slum in Kalkaji, New Delhi, and put a freely accessible computer in it (Hole-in-the-Wall Education Ltd, Beginnings). Children learned to use it on their own. Similar results followed in Shivpuri, Madhya Pradesh, and Madantusi, Uttar Pradesh, and NIIT and the International Finance Corporation set up a company to expand the work, with more than 30 learning stations set up in India and abroad.
Mitra proposed that basic computing skills could be picked up through "incidental learning" given access, motivating content and "some minimal (human) guidance". Note that last phrase; it comes back later.
What did the published studies find?
First month (2001): Mitra and Vivek Rana reported on slum children in New Delhi given internet access in their settlement. In the first month of unsupervised use, the children seemed "to understand and use the technology fluently", and language and formal education did not seem to make a significant difference (Mitra and Rana, 2001).
Seventeen sites (2005): groups of children in 17 rural Indian locations were tested on computer literacy for nine months. The authors concluded that groups of children could learn to use computers and the internet on their own and that such group self-instruction was as effective as classroom instruction (Mitra and colleagues, 2005).
Molecular biology in Kalikuppam (2010): 10 to 14-year-old Tamil-speaking children in a remote village tried to learn basic molecular biology from a Hole-in-the-Wall computer. With the computer alone, they reached test scores comparable with a nearby state school. With a mediator who did not know the subject, they matched children at a high-performing private school in New Delhi (Mitra and Dangwal, 2010).
What did critics find?
In the same journal issue as the Kalikuppam paper, Payal Arora studied two Hole-in-the-Wall sites in Almora and Hawalbagh in the Central Himalaya, and set out to explain why they "failed to take root". She argued for separating Hole-in-the-Wall "as an idea and as an institution" and for examining what unsupervised, public, self-guided access can really do (Arora, 2010).
A broader line of criticism comes from cognitive science. In a 2006 review, Paul Kirschner, John Sweller and Richard Clark argued that evidence from the past half-century "consistently" indicates minimally guided instruction is "less effective and less efficient" than instruction with a strong emphasis on guidance, with the advantage of guidance receding only when learners already have high prior knowledge (Kirschner, Sweller and Clark, 2006). Our piece on John Sweller's cognitive load theory explains why novices benefit from worked examples.
The two views are less opposed than they look. Children can teach themselves a tool such as a computer, especially in groups and when curious. Learning a structured subject from scratch, with nobody checking, is much harder, and Mitra's own subject-learning results improved when a mediator joined.
How to use the idea well at home
Here is a session plan that keeps what worked and adds the guidance the research says novices need.
One big question. For example, "Why does a steel ship float when a steel nail sinks?" Pick something your child is curious about.
Small group, one screen. Three or four students sharing a device must talk to each other, as the children at the wall did.
A mediator, not a teacher. A parent or older sibling asks "How do you know?" and "What did you find?" and encourages effort, without giving answers. Like the Kalikuppam mediator, they do not need to know the subject.
Check against a reliable source. After 30 to 40 minutes, compare the group's answer with the NCERT chapter or a teacher's explanation, and fix errors.
Test without notes the next day. A few questions show what stuck.
For exam preparation itself, mix this sort of exploration with guided practice: curiosity first, then worked examples and feedback. Our guides on encouraging curiosity at home and running a study group that helps give more ideas.
When a student exploring alone gets stuck, Fermi's AI tutor is built to notice where they are stuck and decide whether to explain, ask or show, so curiosity gets some guidance rather than none.
Try one curiosity session this weekend, and finish it with step 4.
