Four parts of scorekeeping earn the right to one part of advice. And there is somewhere for it to go: the successor exercise is already running. NITI Aayog was tasked in 2023 with consolidating ten sectoral thematic visions into Viksit Bharat @2047, an approach paper went to the Governing Council in July 2024, and several states have published their own @2047 documents.
Our own text argued that the next visioning exercise should begin before the current one expires. On that count the system has done the right thing without being told. The successor is not overdue — it is open, which is a far better position, because a vision under preparation can still absorb a correction. This is the moment to hand one over.
Five things, in the order I would rank them.
1. Say which numbers are forecasts and which are demands
This is the defect that runs through everything in the previous four parts.
Some of our figures were extrapolations: milk at 230.5 MT, life expectancy at 72. Others were statements of desire: distribution losses under 3 per cent, maternal mortality at 15, life expectancy at 80. The first kind was overtaken early; the second kind will be missed by a wide margin. Both were printed in the same typeface, in the same launch, and on life expectancy the two kinds contradicted each other inside a single exercise.
The vision document says plainly in its opening that it is not a prediction, not a foresight exercise and not a projection, but a statement of where we would like to be. That caveat deserved to be attached to each number rather than left in the preface. Label every figure. A reader cannot audit what they cannot classify.
2. Never a coverage target without a service target
Taps connected, capacity installed, tonnes produced, kilometres laid — these are the units of construction. Water reliably delivered, electricity actually generated, nutrition actually improved, children who can actually read — these are the units of delivery. We wrote our prerogatives in the second language and the system has been measured in the first.
Print them side by side, in the same box, and the gap becomes a management problem instead of a rhetorical one.
3. Give the cross-cutting technologies an owner
Twelve sectoral panels had no home for artificial intelligence, so it became a bullet in twelve different lists and a transversal in none. Viksit Bharat @2047 is being assembled by consolidating ten sectoral thematic visions. The structure is the same, and so is the risk.
A standing cross-cutting group, with a mandate to argue with every sectoral panel and one question to ask of each candidate technology — if this works, how many of the other roadmaps change? — would have caught in 2015 what we missed.
4. Publish the baselines
The most valuable thing the Technoscape did, and the reason this audit was possible at all, is that the roadmaps printed their own starting values alongside their targets. Milk at 137.7 MT in 2012-13. Maternal mortality at 190 in 2013. Fossil fuels at 80 per cent of electricity supply.
Without those, an auditor has to assemble baselines from elsewhere, and the exercise becomes arguable. With them, anyone can check. It also disciplines the drafting: a target looks different when the baseline sits next to it and the reader can do the division.
Where the baselines are contestable — ours put maternal mortality at 190 against the official series' 167 — publish the source too.
5. Design the audit before the launch
We built no mechanism for reviewing TV 2035 against itself. There was no schedule, no custodian, no obligation on anyone to report progress against the twelve prerogatives, and no place a citizen could go to see how the country was doing on a document written in their name. A decade on, the audit is being done by one of its authors on a personal blog. That is better than nothing. It is not a system.
A vision exercise should be published with its review dates fixed, the responsible body named, and the indicator list machine-readable from day one. That is what Foundation 2035 was meant to enable, and why I would rather publish a scorecard with three red bands on it than not publish one at all.
What I would not change
Having spent four parts on what went wrong, let me end on what I would keep.
The decision to build the document around Indians rather than India — around a peoplescape rather than a technology landscape — was right, and it is why the twelve prerogatives still read well a decade later. They have not aged. The technologies listed against them have aged badly in both directions, but the statement of what every Indian should be able to expect by 2035 needs no revision at all.
And the decision to be specific was right, even though specificity is what makes this series possible. A vision that had said "improved health outcomes" instead of "maternal mortality better than 15 per 100,000" would have been unfalsifiable, and therefore useless, and I would have had nothing to write.
The purpose of a vision exercise was never to be right. It was to be specific enough that people could check, in public, and act on what they found. On that narrower test — and I think it is the test that matters — Technology Vision 2035 has held up rather well.
Prabhat Ranjan was Executive Director of TIFAC through the preparation and launch of Technology Vision 2035. This is the last of five parts.
Sources across the series: Technology Vision 2035 and the Technoscape roadmap summaries (TIFAC, 2015); Sample Registration System bulletins; National Health Accounts 2022-23; NFHS-4 and NFHS-5; Basic Animal Husbandry Statistics 2025; Ministry of Agriculture; Ministry of Jal Shakti; MNRE and CEA; Power Finance Corporation; PLFS 2023-24; India State of Forest Report; Economic Survey 2025-26; NITI Aayog.