I have now sat through eleven consultations on digitising land records in three states, and the minutes of nine of them record somebody invoking the Permanent Settlement inside the first ten minutes. In one it was the second speaker. The line is always some version of: the British made records in order to extract revenue, and we should be careful we are not doing the same thing.
I am a historian of that period and I want to say plainly that the analogy, as it is used in those rooms, is doing almost none of the work its evidence supports — and that it is displacing a better question that the same rooms are perfectly capable of answering.
Be precise about what the Permanent Settlement was, because the room's version is a compression. In 1793 the Company fixed the revenue demand on a class of intermediaries in Bengal in perpetuity, converting a fluctuating claim on agricultural produce into a fixed money obligation, and thereby converting collectors into something close to proprietors. The consequences were enormous, largely bad, and are still legible in eastern Indian agrarian structure two centuries on. None of that is in dispute among people who study it.
Now notice what has to be true for it to bear on a state government putting cadastral maps into a database in 2026.
The claimed mechanism is that recording is extraction — that writing down who holds what serves whoever collects. But the damage in 1793 did not come from recording. It came from fixing the demand in perpetuity, and from vesting proprietorship in a party that had previously been a collector. Those are two specific, separable decisions, and neither is a property of records. The Company also produced surveys that were, by the standards of the period, quite good, and those surveys are the reason we can say anything at all about who held what in Bengal before it.
So the analogy transfers a conclusion without transferring the mechanism that produced it. That is the general form of the thing I object to, and it is why I have become tiresome on the subject of analogy: the stated reason for reaching for one is that it illuminates, and what it usually does is relieve the speaker of having to name a mechanism at all.
Now the objection, which is not small and which I will not be able to dismiss.
Demanding that non-specialists reason without analogy is impossible and slightly contemptible. Analogy is how anybody approaches an unfamiliar thing, historians very much included. And the room's instinct is not baseless: there is a genuine pattern in which administrative legibility precedes extraction, it has a serious literature behind it, and somebody saying 1793 is usually gesturing at that pattern rather than making a claim about revenue settlement in particular.
I have to go further than conceding it, because I have done it myself and the honest version of this essay says so.
In 2011 I wrote a paper arguing that a proposed tenancy registration scheme would reproduce the dynamic of the nineteenth-century survey settlements — that registering tenants would make them visible and therefore evictable. It was a good analogy. It had a shape. I did not name a mechanism, and when the scheme ran in two districts the effect was close to the opposite of what I had predicted: registration made a category of tenant harder to evict, because for the first time an eviction had to contend with a document. My analogy was not refuted by an argument. It was refuted by four years of district data that I had not thought to ask for, because the shape had felt like enough. A test I only apply when I dislike the conclusion is not a test, and in 2011 I liked my conclusion.
So the rule I want is narrow rather than prohibitive, and it is the same shape as the one I have proposed about textbook revisions, because I think it is the same problem: make the stated reason checkable. An analogy must name the mechanism it claims transfers. Not the outcome — the mechanism. If you say digitisation risks repeating 1793, say which decision is being repeated. Perpetuity? Vesting? The exclusion of subordinate rights from what the record can hold? Each is a different objection with a different remedy, and each can be checked against the district's own material by somebody in an afternoon.
Do that, and the third possibility turns out to be the live one, and it is exactly what the analogy is crowding out.
Here is what actually goes wrong when a record is digitised in an Indian district. The paper record has a principal holder, and in the margins, in a hand nobody had to justify, a great deal else: a tenant of long standing, a share arrangement, a path, a well used by four households, a widow's residence right that no document confers and every neighbour acknowledges. The digital form has fields. The fields were built around the principal holder, because the principal holder is the party the state transacts with. Everything in the margins goes into a free-text remarks box that no query ever reads, or it does not go in at all.
Nothing has been extracted. Nothing has been fixed in perpetuity. What has happened is that a body of claims which survived for a century by being written informally where a village-level officer could see them has been sorted into what the database knows and what it does not, and the second category has been, in the most literal sense, forgotten by the state.
That is a serious harm, it falls on precisely the people the room is worried about, and it is a completely different harm from 1793 — different route, different remedy, and the remedy is a schema question argued with the people who built the form.
I have made this point in perhaps six of the eleven consultations. It lands less well than the analogy does and I understand why. 1793 is a story with villains; mine is a conversation about fields in a form. The analogy is also unfalsifiable, which a room experiences as strength, and my version can be checked, which a room experiences as a delay.
In the one consultation where it did land, an officer from the revenue department told me afterwards that the remarks box in their system has a character limit of two hundred and fifty, that it was inherited from the vendor's earlier product, and that nobody at any point had chosen the number.
