What a property data licence actually lets you republish

The interesting part of an open data licence is rarely the licence. All three of the property registers this network uses sit under base terms that permit commercial reuse with attribution. What differs, and what actually constrains a publisher, is the layer of sector-specific conditions on top, and those are easy to miss because they are not in the licence document.

This is a practical summary, not legal advice. Read the primary texts, which are linked below, before building anything on them.

How similar are the base licences?

The Open Government Licence, the Singapore Open Data Licence and Licence Ouverte all permit copying, adapting, combining and redistributing the data, including commercially, provided the source is attributed. None requires share-alike. None charges. For a publisher, the practical differences at this level come down to how attribution must be worded and whether a source date is required.

That is genuinely permissive, and it is why independent analysis of property markets is possible at all in these three countries and largely impossible in countries whose registers are paywalled or closed.

Where do the real constraints live?

What is the French condition, and why is it unique?

The second French condition is unusual and worth understanding because it constrains an architecture rather than a use. It does not say you may not publish statistics derived from the register. It says reuse must not allow the data to be indexed by external search engines.

In practice that separates two things a publisher would otherwise treat as one: the derived figures on a page, which may be indexed, and the underlying records behind an endpoint, which may not. A site that exposes a downloadable transaction file to a crawler is in a different position from one that publishes a commune median. It is a condition about plumbing, and a publisher who thought only about content would miss it entirely.

What happens when the decisive document is one nobody has read?

Singapore’s private register illustrates a general trap. The published terms defer to an open licence that reads permissively, but access requires a key, and the key comes with a registration agreement that is not published. A reader can therefore establish that the visible terms are permissive and still not know what was actually agreed.

The conservative position, and the one this network takes, is to publish derived statistics and not redistribute the records, on the basis that the asymmetry is severe: a modest gain against a dispute with the agency the whole market depends on. The cheap way out is not more reading, it is a question in writing to the publisher.

What should a reuser actually do?

• Find the sector conditions, which are usually on the dataset page rather than in the licence.
• Attribute in the prescribed words, not a paraphrase, where the licence prescribes them.
• Separate the data surface from the content surface, so a condition about indexing can be honoured with a header rather than a rewrite.
• Set a sample floor, because re-identification conditions make thin published cells a licence question and not only a statistical one.
• Write the position down, so the next person does not have to re-derive it.

Which registers publish transactions at all is the companion directory. The three positions described here are worked out in public on RealScout, IndexProp SG and IndexProp FR, each of which states its licence on its own methodology page.