Palantir is not made of magic. It is just some people with some software.

Palantir Foundry, the software that currently runs the NHS Federated Data Platform, is a rebranded copy of Apache Spark – open-source software you can download and run yourself today. What makes Foundry expensive isn’t innovation; it’s packaging, lobbying, and the phalanx of consultants in branded polo shirts feeding intelligence back to the mothership.

There’s nothing secret or particularly unique about how it works. Like the Snowden revelations in 2013, the secret was and is: there is no secret; it’s just better funded.

Palantir’s consultants are embedded deeply into customer organisations, giving the impression that only Palantir can understand and evolve the system. But everything they do can be documented, reverse-engineered, and replicated with more open, less ideological tools.

Palantir sells itself as transformational. But the truly hard part of any digital transformation is not the code – it’s the institutional work: turning analogue services digital, cleaning up decades-old processes, getting buy-in from stakeholders, designing new ways of working, prosecuting citizens for failings of the system, etc.

Palantir heard “America First” and “has chosen sides” – if the US is no longer a reliable ally to the UK, then Palantir is no longer a reliable supplier to the UK’s NHS. To explain the analogy in Dr Baw’s “Everything Digital Health” youtube series, the NHS “Federated Data Platform” is a “Federation” in the style of 1970s BBC SciFi Blake’s 7, not in the style of Star Trek – no one ever wants to think they’re a cog in the grinding wheel of dystopia.

Replacing Palantir is possible

The NHS doesn’t think twice about swapping from an expensive branded drug to a  biosimilar generic as soon as it becomes available. The same should apply to Foundry. It’s time to treat software platforms like we treat pharmaceuticals: the best version at the best price, without unnecessary monopolies. It can have exactly the same screens and data flows, but without the dependency on a company whose ideology is a driving factor in what they do.

Once the real work is done, once the diagrams of data flow are clear, they can be exported from Palantir/FDP. They can be tested (one hopes the tests have already been written). They can be re-implemented – without the high ongoing costs of the “America First” baggage. Anything Palantir claimed as proprietary can be replaced, cheaply using modern AI, from the documentation the NHS owns and the tests the NHS wrote to make sure it all works. This wasn’t possible when FDP was conceived; it is now.

It’s a lot easier to get off Palantir than to get hooked on it in the first place – but there are interests in pretending otherwise.

More in part 2.

===

In addition to our annual-ish newsletter, you can also join our free substack to get emailed whenever we post some news or commentary.