The NHS is watching Mr Streeting speedrun the dictator’s dilemma, powered by Palantir to go ever faster.
Just recently, Palantir rolled out a worldwide advertising campaign using the slogan “making Americans irreplaceable” – irreplaceable because they use Palantir – suggesting Mr Streeting and his Department of Health in England are “making Americans [and Palantir] irreplaceable” to the NHS, and beyond as the “operating system for government”.
(Meanwhile, the Blair government-in-exile is trying to make Oracle irreplaceable. The proposals in the Ellison-backed Blair Institute’s ‘Preparing the NHS for the AI Era’ effectively amount to, “We like what Palantir is doing in the NHS, but it should be using Oracle instead”.)
Unintentionally, new decision makers
A recent presentation to the Bank of England asked, “Can a 22-year-old engineer close all the banks?”
If there was a big red button that could close all of the banks in England, it would be very heavily guarded – and extremely robust, precise processes would have to be followed before it was ever pressed. While no such button exists, a random junior programmer is able to make a typo which shuts down a few airlines, banks, and various other pieces of rather important infrastructure worldwide for days, and can do so within minutes. (Palantir calls its equivalent mechanism “Apollo”.)
To change the cancer treatment pathways across the NHS has traditionally required significant clinical input, a lot of meetings, and in some cases a degree of campaigning – especially when the suffering of particular groups of people are being ignored. That’s the old way; the legacy system, and Palantir doesn’t like old ways of doing things.
With Palantir’s Cancer360 monopoly being imposed on NHS cancer treatments by NHS England, another ‘new’ way to introduce a cancer pathway would be to persuade a single Palantir staffer (who can decide to work on Cancer360 next week) that they could make a real difference to their own Grandma’s cancer. Such a change could be rolled out quickly, especially if it were to be mislabeled as an “option for those who want to use it”. Such options can be prioritised, or not – but those are Palantir decisions for Palantir products, not NHS decisions for NHS care pathways.
NHS England has, in effect, given Palantir engineers a monopoly on deciding how the NHS can work.
Palantir can and does refuse requests from its public body customers for things that are necessary for those bodies to implement democratically-mandated decisions. If a feature works as Palantir wants it to work, but not how the NHS needs it to work, then NHS staff could find themselves waiting on hold until Palantir staff deign to change their minds. And such waits can take a very, very long time when bugs benefit Palantir’s worldview. If Palantir doesn’t want a feature to be supported, it doesn’t necessarily have to block it – the company can simply not give any support to it, and use their intelligence network and spies (sorry, FDSEs) to undermine those who want it at every turn. (As they’ve tried before…)
The recent biography of Palantir’s CEO makes it clear that Palantir’s culture is to have opinions, and that they aren’t shy about deciding that their opinions are right and should be heard by others.
In much the same way as Mr Streeting wants control over the entire NHS to be centralised in Mr Streeting’s hands, so Palantir wants the decision makers to be Palantir – because, as they said in their S-1 Registration Statement:
“Our society [NHS] has effectively outsourced the building of software that makes our world possible to a small group of engineers in an isolated corner of the [foreign] country. The question is whether we also want to outsource the adjudication of some of the most consequential moral and philosophical questions of our time. The engineering elite of Silicon Valley may know more than most about building software. But they do not know more about how society should be organized or what justice requires.”
Why might Mr Streeting want to make Palantir irreplaceable?
Politicians desire “the seductive lie of control”, which Palantir will sell to them for a hefty fee. However, Palantir’s claim to be “making Americans irreplaceable” only works when you care solely about your US audience – the NHS and UK mean that little to Palantir’s high command.
Palantir talks about its security credentials by pointing at its other customers. In the same way, Google says they’re the people who run gmail (of course, they also ran Google+) and it was probably only a single product manager at YouTube that undermined all of Google’s security model (thread).
Developing discharge and developing the Single Patient Record
NHS England is very behind on publishing its “FDP check and challenge” minutes which should come out “within 2 months”. At the time of writing, in late November, NHSE’s most recent published minutes are from May.
We expect the September minutes to confirm that NHS England has decided to roll the Single Patient Record (SPR) into the FDP, and that NHS England and Chelsea & Westminster Hospital have been caught experimenting on patients with AI models that have not gone through the proper MHRA assessments.
NHS England officials will deny this, but will not write down that they have denied this; it’ll be very interesting to see what the published minutes say.
As outlined above, if you want new features added, there are several ways to get them. But what about research and public health?
Research and public health
Because DH/E have spent all of their tech budget on Palantir, they are currently decommissioning all their other data systems, including those within the UK Health Security Agency, formerly Public Health England (4.1 in 23/10 AGD).
In future public health incidents and emergencies – and for future decisions – data informing those decisions will be managed using Palantir systems. Palantir at the same time may have to make complex decisions about how they are willing to have their product used.
It is possible Palantir will decide it is entirely happy with Foundry being used to end lives, but not used to inform better public health interventions that will save them. Palantir’s embedded consultants that work for the NHS or UKHSA on smoking prevention this week can next week be working for tobacco companies, or anyone else who buys Palantir’s services.
Devolution means choices
The NHS in Manchester knows what it’s like to get slapped around by NHS London and NHS England, both of whom care little for what Manchester needs and care even less what Manchester does about it. (One could of course replace Manchester with Liverpool, North Yorkshire, or other locations.)
Normally, the choices of NHS England are limited to England. With the imminent announcement of the UK-wide “Health Data Research Service” (HDRS), however, it is NHS England’s position that Scotland, Wales and Northern Ireland will be required to put all of their data into Palantir in order to participate in that new Service. (Whether this is also the position of HDRS is subject to much disagreement…)
Local NHS decision makers can reject nationally created monopolies
There are many areas and many NHS bodies which do not believe NHS England’s promises about the Utopia of Palantir; they don’t believe ‘everything will be wonderful’ if they just abandon everything they are currently doing, and do whatever the soon-to-be-abolished NHS England demands they do this week. Experience shows they may be correct in their belief.
Such Trusts can implicitly reject the NHSE-imposed Palantir monopoly by choosing and stating they will use one of the alternate suppliers for the functionality provided by Cancer 360, OPTICA, etc. If necessary, they should build those functions in-house to avoid being dependent solely upon Palantir’s products.
The core Foundry platform may be a commodity platform with a well funded lobbying arm, but NHS England has allowed Palantir to build Palantir-only functions on top of it.
One question that remains to be answered is whether Trusts will be forced to use the Cancer 360 and OPTICA products which NHSE has given Palantir a monopoly on. Trusts can choose to be trusted, and Manchester in particular is in a good place to force an alternative to Palantir’s cancer pathway tools. How the NHS treats cancer patients should not require the approval of Palantir staffers – because if Palantir doesn’t want something to be possible in Cancer 360, it won’t be possible.
To use an analogy from GP land, it is as if NHSE found itself with a duopoly of suppliers who largely disregard what they want, then decided it wanted to take the TPP/EMIS duopoly away, and to create a Palantir monopoly instead.
NHSE’s “solution exchange” for the FDP claims to be a “multi-vendor landscape” – except “Currently, the Solution Exchange does not support third party products”.
This is a place where Trusts can make sure NHS England keep their promises and not hand a long-term monopoly on cancer treatment pathways or discharge to Palantir. If the failure to allow multiple suppliers onto FDP is because NHSE has been busy spending all of its internal budget on Palantir tools, then Trusts and ICBs with existing platforms are perfectly placed to wait until NHSE delivers and approves competing tools before they make their assessment what is best for the Trust. Such actions also enforce a competitive market that NHS England would (temporarily) prefer remains a monopoly (the same way a monopoly is less work for NHS England in GP IT Services than the existing duopoly which previously seemed better than what came before…)
The lack of support for third party products could also be another bait and switch by a large US actor making outlandish guesses at savings, who does so knowing that by the time they are not delivered it will be too late. That company may see this as another business opportunity, but Palantir has become the “small group of engineers” inside the NHS – potentially working against the UK’s public interest, and against the NHS.
==
In addition to our annual-ish newsletter, you can also join our free substack to get emailed whenever we post some news or commentary.
[Since this may get read by some Palantir and US tech types; if Trump’s Government can shred Anthropic, Palantir can follow…]