Below is some further evidence including a note I sent on 26 April regarding how we could shift to Plan B with a serious testing system.
It helps people understand what an incredible mess testing was and why care homes were neglected. Hancock had failed terribly. The Cabinet Office did not have the people it needed to solve the problem. Many were screaming at me that Hancock was failing to act on care homes and spinning nonsense to the Cabinet table while thousands were dying in care homes.
There are clearly errors in my note but the fact that *I* had to write it tells you a lot about how the system had collapsed. As you can see it is a draft for a document that needed to exist but didn’t because Hancock had not done his job properly and was absorbed in planning for his press conference at the end of April, not care homes and a serious plan for test-trace.
Note how I referred then to suspicions that we were NOT testing people in care homes as we’d been assured by Hancock would happen: e.g ‘We agreed 10 days ago with MH that we would test everyone going into homes regularly but I don’t see this happening. Shouldn’t we be testing asymptomatic workers?’
DW = David Williams, then senior in DHSC. It appears below exactly as sent, I’ve left errors/typos unchanged (some formating is a bit different, bold is as in original). I hope this sort of raw note helps people understand what was really going on. Such things are often more useful in figuring out ‘what really happened, what were they really thinking’ than formal notes written ‘for the record’.
To: [Various in Private Office / covid team]
Date: 26 April
Subject: Crucial numbers and actions -- I will copy PM into this to help him understand the complexity of it all -- no idea what CABOFF unit doing but SOMEONE has to get these numbers and coordinate across system ... pls copy to Malcolm... there are clear errors in the DW plan!
Few thoughts for PO re action on crucial numbers...
We urgently need two bits of paper, short but very carefully done TOMORROW morning for the PM:
a) set of data on T&T,
b) data on the overall testing program under different assumptions including T&T assumptions.
Each must also set out critical milestones so we can actually judge day to day whether we are making progress. And in general unless we are absolutely confident in the people running projects and their power to get things done (and almost nothing fits these criteria now), Shinner must have visibility of the detailed plan so we can judge a) whether it’s any good and b) if they are making progress. E.g for T&T: how many people hired and trained, what are the steps for integrating with app and testing structure and so on.
Crucial people to look at these 2 notes: Vallance, CMO, D Williams, S Stevens, Aldridge, XXX…?
We must look at some overall numbers and how they fit together to guide policy. Now all meetings are happening in a largely data-free way so neither officials nor ministers can really get to grips with the choices, implications, demands in terms of staff and resources, policy implications etc.
Also there are huge interdependencies and feedback between for example:
Level of infection and R when we start lifting;
which elements of lockdown are lifted how fast, where and in what order;
parameters for T&T, how fast can thousands of people be recruited and tested etc;
overall testing capacity and trajectory;
how the public changes behaviour before and after lockdown is lifted...
Integrating these judgements and figuring out all the policy and operational implications for things like testing capacity and NHS operations is hard and we are far off the pace.
Malcolm, TomS, Imran, BenW and others need to look at the below and turn it into a proper note and get it into the system tomorrow... We must start this process and get Stephen Aldridge and others on this.
CURRENT GOVERNMENT ASSUMPTIONS
The ONS are running community testing at ?10k random sample per week shortly, then 40k per week, and will detect the true level of infection and R within these parameters XYZ with this confidence XXX.
Assume actual infection rate is between A and B and the R is XX by 7 May.
Assume the number of patients who test positive per day by 7 May is between A and B. (~1% of these will die so accepting X cases means accepting roughly 1% of X deaths.)
Assume we pursue 30 of the closest contacts of each person who tests positive per day. If daily new positive tests is 4k per day (per Germany’s plan), then there are 120k per day to T&T (and implied 40 future deaths per day). (Another complexity: the same person will be a contact of multiple cases with different dates of last contact. What are the activities for the low-risk contacts? Are we advising them to self-monitor, for symptoms, remind them of hygiene etiquette and remind them to self-isolate immediately should symptoms develop? This could just be done by text but if we do this, we should add resources here. ECDC estimate that a person's total contacts are 90, of which 1/3 will be high risk (close contacts). That still leaves ~60 people with low risk contact with a case.)
Assume we quarantine X% of contacts within N hours and test X% of them. (SAGE says N must be <48 hours for tracing. Quarantine must happen fast, testing can be later. What is the plan for MONITORING, this is different to T&T, will we send texts, compulsory wrist bands, police random checks etc?)
The cascading effects of this are that per month we will be testing X, T&T X, quarantining X, releasing X back to work within [48 hours?] and so on. (Flows from decision to test all contacts or only symptomatics?)
This has the following implications for the volume of tests we need.
This has the following implications for the level of T&T we need, workforce recruitment & training, etc.
This has the following implications for the NHSX app.
This has the following implications for logistics of quarantine -- we will need to pay XX people YY per day to stay at home and ZZ for free food. The enforcement mechanism will be XX and will cost XX and require the following decisions by ministers XXX.
This has the following implications for the NHS operationally. [SS is writing this]
For comparison, here is a A3 chart of the Taiwan and South Korea test-track-trace infrastructure with timelines attached. Here are the Taiwan, SK and Singapore numbers: how many tested per day, T&T, speed etc. Here is our suggested model on an A3 chart. [These should be ready for PM on Monday.]
Connected there are many policy questions concerning this that we need rapid answers, e.g:
Do we enforce electronic wristbands connected to phones on quarantined people as per HK?
Do we mandate response to texts as per Singapore (to reveal location)?
Like Taiwan and Singapore do we link medical and health insurance records, travel history to contact tracing efforts? Do we link credit card and other data? It obviously improves efficiency and could be done in a way that protects privacy reasonably but is another big data challenge to do over 4 weeks.
How do we know if our track and trace program is working?
If T&T begins to fail what is our plan in this scenario?
T&T will be very different in autumn when flu symptoms return and we need to plan for this now.
We need to have a plan to ensure that we are protecting places where isolation is difficult i.e care homes. We agreed 10 days ago with MH that we would test everyone going into homes regularly but I don’t see this happening. Shouldn’t we be testing asymptomatic workers?
The CABSEC / D Williams plan
David Williams’s current plan assumes:
Tracing 10 contacts (but the official plan they presented in CAB room this week says ‘30’) of 25k new cases per day, so 250k T&T per day.
18k people will be able to handle this from ‘early May’.
Williams says ‘the app is additional’ capacity on top of what is building, but later in the note says ‘the app will need to pick up some element of this’.
SAGE assumes only 90% will be actually traced, so 25k index cases actually requires 225k (not 250k) T&T. (This may be slightly wrong?)
SAGE says that to be effective >80% of these non-household contacts should be traced within 48 hours.
It seems he is distinguishing between 10k ‘index cases’ and ‘tested positive’ -- he says 10k I.C ‘broadly equates to’ 4k T.P.
‘We can manage 10k index cases per day with 20-30 contacts … or 25k at 10-15 contacts.’
10k per week tests used for ONS surveillance and 65k per day for patients/key workers so 25k free for T&T. (This is wrong -- a/ ONS is planning to contact 14k households per week and assumes 10k will agree, which implies 20-25k tests; b/ there are other demands like care homes.) We will have 200k tests p/d by end May and have ‘line of sight’ to 300k p/d by ‘end of June’ but ‘50% likelihood’. So there should be 125k tests for T&T by end of May.
We will not test contacts unless they are symptomatic. (NB. I thought we want to test asymptomatic people and places like SK do this? I am very concerned this is not thought through.)
NB. the current plan assumes a certain steady state -- infections per day, contacts per day etc -- but the critical number is R. So we need to calculate things like: how sensitive are these numbers to fluctuations in R over time? We start with a level of positive tests per day etc but in a dynamic environment where the number will be shrinking if R remains <1, but will start growing fast if R>1… R at the time will impact the number of people, per 1 index case, that were close contacts, developed symptoms and tested positive. These people will become an index case, and need their own round of contact tracing. If they have successfully self-isolated when they were informed they were a contact, this list should be much shorter, as they should have far fewer contacts (but still would have had some in the 48 hour T*T period of the original index case.)
We need to see the detailed plan for T&T with timelines. When and how will it be piloted? There will be bugs that need fixing. There will be glitches at best of integration between parts of the system. We will need to refine to drive the cycle faster. It’s pointless if it takes 5 days. And we still haven’t seen a plan for the app testing (but we may this afternoon when we talk to M Gould).
PHE currently use a system where people are asked to enter their contact details themselves, with about a 66% participation rate in this. Can we think of other measures to speed this up (aside from the app)? For example, when we relax measures and people do go to garden centres etc, can they be asked to provide their contact details (time stamped) to facilitate contact tracing? They are rolling this out in Singapore, anyone attending a market/shopping centre has to provide their contact details at the door.
If we have this active contact detail collection measure in place, then we can actually relax other sectors of the economy that we otherwise might struggle to do (e.g hair dressers and other 1:1 services that are all appointment-based).
Summary: we need crucial assumptions clarified, a real plan needs actual numbers that make sense from a SAGE/health perspective and translate to the operational plans, these need to be checked across SAGE/DHSC/CABOFF/No10/HMT, we need to see the operational plan for T&T and ensure it is resourced and supported, and this all connects to the ‘observatory’ function.
I have no confidence PHE can organise the observatory and the CABOFF can ONLY do it with a lot of expert help -- the CABOFF could not do this sort of data science operation by itself pre-March. If we leave the system to its default, we will not have the observatory we ought to have already and will urgently need when we want to undo lockdown.
I forwarded this note to the PM who was still in Chequers with this message:
FYI -- if you skim through this you will have good idea re some of the problems/complexities...
As usual, my team of irregulars is having to do this cos the centre cannot do it.
All this should have been done weeks ago and should not need me to do it.
We can't go on like this.
The PM replied on 26/4:
Thanks totally agree
The whole track and trace thing feels like whistling in the dark
Legions of imaginary clouseaus and no plan to hire them
Apps that don’t yet work
And above all no idea how to get new cases down to a manageable level or how long it will take
By which time uk may have secured double distinction of being the European country w the most fatalities and the biggest economic hit
So your email is bang on
We GOTTA turn it round
I also forwarded it to Vallance and Whitty with this message:
FYI -- Ive sent this because a) I do not see CABOFF providing this crucial integration function, b) I am extremely worried that neither CABOFF or DHSC has an operational grip on T&T plans and how they connect.
I will try to push this this week.
Obviously please send any thoughts...
I had returned to work on 13 April.
After a couple of weeks of meetings with Hancock, many people concluded that:
He had told us he was organising care home testing including asymptomatic.
There was no serious plan, operational delivery was terrible.
He was not telling us the truth around the Cabinet table.
We were therefore killing people we claimed to be shielding.
On 3 May I sent this WhatsApp to a core No10 group:
As is clear from this message, I and others thought government ‘NEGLIGENCE’ was killing the most vulnerable people and I was trying to get to the truth. The Cabinet Secretary had also told me that he’d lost confidence in Hancock’s honesty and operational grip. As you can see from the reply, the PM saw this — he cannot claim ‘nobody told me’. And this message was repeated to him many times.
The meetings over the next few days that resulted from the above message convinced me and others in No10 that our suspicions (1-4 above) were true.
This led to another insistence from me and others that Hancock be removed. On 7 May I told the PM that Hancock was ‘unfit for this job’ and him staying in place was ‘killing god knows how many’.
The PM agreed that Hancock’s failures were a catastrophe but refused to fire him.
The failures, the lies and the PM’s refusal to remove someone who had failed so badly are not just about Boris Johnson and Hancock.
The media and MPs inevitably focus on individuals but the real covid story is a story of system failure — a system that successive PMs, Chancellors and Cabinet Secretaries did not seriously analyse because they did not have the intellectual tools or relevant experience to do so. Heywood in particular — assuring everybody that Whitehall had ‘nothing to learn’ from the private sector on project management, procurement and data — was a great fixer but a lousy manager. The Cabinet Office that collapsed was mainly his creation.
We can see from history the ‘unrecognised simplicities’ of high performance entities and the sort of characters, with certain skills, capable of it.
Our political institutions systematically exclude such people, systematically promote Hancocks, and systematically block learning from high performance, which is seen as a dangerous menace by normal bureaucracies.
Thus we get Hancocked.
As Tolstoy put it wonderfully in War and Peace, ‘we get Macked’. The cynical diplomat, Bilibin, is explaining the latest disaster against Napoleon, a tragicomic story in which the Austrians accidentally gave away the Tabor bridge to the French because of a general, systemic dysfunction in General Mack’s army.
“‘It’s not treason, or dastardliness, or stupidity: it’s the same as at Ulm… it is…’ – he seemed to be trying to find a suitable expression. ‘It’s …
c’est du Mack.
We’ve been Macked,’ he concluded, feeling that he had coined a word, a new word that would be repeated.’ (p. 186, Edmonds translation.)
Britain gets ‘Hancocked’ every week. Few MPs have any serious understanding of management and neither do almost all political journalists so the crucial issues are never discussed, punditry is all about surface phenomena rather than things like deep incentives and operational performance.
In a world of ‘launch on warning’ nuclear protocols, biological weapons, genetic engineering, the spread of AI/ML and autonomous robots and so on, we cannot go on like this.
We need regime change. A lot of Whitehall needs to be closed, everyone removed from the building, and new institutions created. If we don’t, this sort of disaster will inevitably repeat. Remember, as Buffett and Munger say, turning around hopeless institutions is so hard they never try — this is why in the economy we have startups. We need the equivalent to replace some of our old political institutions, like a Cabinet room that in March 2020 looked practically identical to July 1914 and collapsed in the crisis just as I wrote in 2019 that it would do. It would be much easier to close the Cabinet Office and reopen something 10X better than ‘reform’ it.
Please SHARE, tell your MP to support an immediate open inquiry, and SUBSCRIBE — I hope to hire some help to a/ push for an inquiry, b/ to develop specific plans for regime change…