Dominic Cummings recently appeared before MPs at a Commons Joint Select Committee investigating the government’s response to the coronavirus pandemic. Cummings argued throughout his appearance that the government’s original plan for a pandemic — allowing it to spread throughout the population in order to achieve herd immunity — was terrible. It was eventually averted, in his telling, as he and others within government managed to convince the Prime Minister that this plan would be disastrous (the NHS would collapse and thousands would die), so we locked down instead.
Cummings himself has since expanded upon two major points of discussion at the meeting, namely the performance of Secretary of State for Health Matt Hancock (abysmal, apparently) and the issue of procurements i.e. how the government decides who to contract for stuff (it’s also abysmal).
But there was another strand to Cummings’ testimony to MPs that relates more directly to the afore-mentioned original plan: openness.
Cummings described on several occasions how openness would have revealed the issues with the original plan earlier, and so bought the government more time to come up with a new one (transcript):
If this process had been opened up to outside scientific scrutiny, and other smart people’s scrutiny—remember, a lot of people who figured out the answer to this were quantitative people who were not epidemiologists or disease specialists; they were people like Hassabis, people like Gowers, people like Marc Warner—in January and we had put all of these assumptions out on the table, we’d have figured out at least six weeks earlier that there was an alternative plan, and this whole thing of, “Well, you’ve only got a choice between that peak, or delay and then that peak” is complete garbage. In fact, the plan that we put to the PM on the 14th, we could have that six weeks earlier, about supress [sic] and then build, and crash programmes for drugs and vaccines and everything else. So I failed, and I apologise for that.”
One thing that I did say… is that there ought to be an absolutely thorough total review of all such risk register programmes. There ought to be an assumption of making this whole process open. It should be open by default and only closed for specific things… Bear in mind that, on this, as soon as people like Marc Warner, Demis Hassabis and Tim Gowers looked at what was being planned, they could say straightaway, “This is wrong; your logic is wrong here; your logic is wrong there.” I am absolutely sure that if you opened up this kind of process—not to people like me, who frankly wouldn’t have good questions to ask—there is definitely a way in which that process could be improved.”
[If you want to read these passages in full for yourself, you can find them on pp.23; 33-34; 36; 72; 115 of that document.]
To summarise Cummings, openness as a default practice for government means plans for things like pandemics would be published so that external individuals can scrutinise them properly.
To be clear, these plans exist. Cummings also pointed out in his statements to MPs that they exist for anthrax, for example. However, it’s difficult to say how good these plans are without seeing them. The plan for a pandemic proved to be quite bad, as we all saw (i.e. herd immunity, which was abandoned in favour of rolling lockdowns), and Cummings contended that this would have been clear to outsiders soon after the virus emerged. Opening up government would allow the plans to be evaluated prior to the catastrophe actually happening.
This idea is similar to that of ‘red teams‘ i.e. a group that evaluates something from the perspective of the enemy, and then reports back their findings. They are useful because there is usually a limit to both expertise and critical thinking with an organisation — whether because of bad incentives, poor personnel, or groupthink — so it’s better to let smart outsiders comment on plans and help to improve them.
This is fully in line with what people like Philip Tetlock have written about in recent years (and Dietrich Dörner decades ago), and unsurprisingly Cummings is a student of this kind of work. See the links in the ‘further reading’ section here, where he points out that feedback loops in politics are too slow for improvements to happen.
Opening up government would also help address a problem I discussed in a previous post. More than ever, the internet highlights and surfaces smart individual thinkers:
But these individuals can’t be put in charge, writes Scott Alexander, because politics and power would change the calculations they’d need to make; they’d have to try to hold onto their job instead of focusing on being right.
Better ways of governing can’t just be about getting ‘the right people’ into power, then. Alexander makes his peace with it in that post, suggesting that power-conscious officials who are mediocre are perhaps the best we can hope for.”
One way to utilise these smart outsiders is to make plans public, as Cummings suggested, and invite their comments. Making this process public would allow the press to ask smarter questions and push the government to justify its plans.
Openness is pretty unobjectionable to me, but let’s consider at least one reason why this might be unwise: national security. Obviously, publishing UK government plans on how to handle terrorism or the invasion of a NATO ally could give terrorists or competing states a competitive advantage in such a scenario.
Cummings addressed this in his testimony:
The SAGE stuff should have been published—scientific advice should be published, unless there are very specific national security reasons for it not to be. You can imagine some crises where it would be different from covid—”
This is where the ‘default’ part comes in — plans should automatically be published unless there is a compelling reason not to. Openness is the default.
Perhaps by this point in the post you still think this is ridiculous and could never work. But did you know that the Bank of England’s Monetary Policy Committee publishes its minutes eight times a year? I don’t know of any studies on whether this improves their performance or not, but it shows that in principle such an approach is possible, at least for this fairly apolitical institution.
And remember how poor the government’s performance during the pandemic has been at points — it’s worth trying something new. Imagine a Ministry for Existential Risk, apolitical like the Bank of England, which regularly publishes its plans for solar flares, superintelligent AI, and other potential catastrophes. Smart people can comment on the plans, which then get updated periodically.
Obviously, openness won’t solve all the government’s problems in an instant, and may be more or less applicable in different areas. I’m not sure whether this method will be entirely successful in overcoming the need of bureaucrats to play internal politics and hold onto their job. But at least if this was all happening in the public gaze, there would be some more incentive to do the right thing, instead of the bureaucratically-expedient thing. When the plan is eventually enacted, there’ll be a public record showing who pushed for what, and when.
Other changes to government seem sensible too — Cummings also pushed for openness in the context of civil service hiring (it’s difficult to bring outsiders in), and changing the incentives of government. He’s also been tweeting praise of Fast Grants, and it’s worth thinking about how government could be sped up in future. We need, as Ben Thompson has written, new defaults for government during the internet age:
First, it should be the default that free speech is a good thing, that more information is better than less information, and that the solution to misinformation is improving our ability to tell the difference, not futilely trying to be China-lite without any of the upside.
Second, it should be the default that the status quo is a bad thing; instead of justifying why something should be done, the burden of proof should rest on those who believe things should remain the same. This sounds radical, but given the fact that the world is undergoing profound changes driven by the Internet, it is the attempt to preserve the unsustainable that is radical.
Third, it should be the default to move fast, and value experimentation over perfection. The other opportunity cost of decisions not made is lessons not learned; given the speed with which information is disseminated, this cost is higher than ever.”
Combining principles like this with quadratic funding/voting (which Cummings himself appears open to) could allow for a radically different form of government. But more work needs to be done to help us get there, including work to raise the status of these topics in British political discussion.
Thanks to Saunderson for feedback on a draft of this.
If you want me to write more things like this, you should subscribe to my weekly email, which includes links to what I write and a few extra interesting things.
Comments welcome below.