And what did we expect then?

The Simple Shopper strikes again

It was dubbed ‘Air Jenrick’ – a Royal Air Force plane left waiting in a Turkish airport for a vital shipment of medical gowns.

The UK’s severe shortage of PPE had descended into farce, with officials scrambling to secure equipment needed to keep frontline NHS workers safe.

Today the Telegraph can reveal that the mission ended in disaster.

Every one of the 400,000 gowns brought back from Turkey last month has been impounded in a warehouse outside Heathrow Airport after inspectors found the gear was “useless” and fell short of UK standards, senior sources said

Of course, it could be that we’ve got anal retentives as inspectors too……

27 thoughts on “And what did we expect then?”

  1. For the idiotic public on the whole it is more important that something is seen being done – like this disaster – than someone explaining to those same idiots beforehand that the quality and standard of the equipment is not fit for purpose so no need to bother.

    And don’t forget to clap to your masters at 8pm tonight.

  2. On a brighter note, I heard last night through the grapevine about a police man who’d been playing golf locally.

    I was just amazed he’d been given the time off, or had wanted to eschew all that overtime.

  3. Bloke in North Dorset

    OT

    I’ve just posted this on one of yesterday’s threads but for those who don’t go back, here’s a damming code review of Ferguson’s model.

    Conclusions. All papers based on this code should be retracted immediately. Imperial’s modelling efforts should be reset with a new team that isn’t under Professor Ferguson, and which has a commitment to replicable results with published code from day one.

    On a personal level, I’d go further and suggest that all academic epidemiology be defunded. This sort of work is best done by the insurance sector. Insurers employ modellers and data scientists, but also employ managers whose job is to decide whether a model is accurate enough for real world usage and professional software engineers to ensure model software is properly tested, understandable and so on. Academic efforts don’t have these people, and the results speak for themselves.

  4. As I understand it, Ferguson had form from foot and mouth to avian flu, now he can add chinese bat flu to his cv.

  5. The Meissen Bison

    BiND/Timmy

    That link doesn’t and the thread has been vanished. Jolly interesting article btw – I followed the link yesterday.

  6. BiND,

    There’s no specification, the code isn’t properly split into functions with return values, the functions (as they are) are enormous, there’s 6 levels of if/for nesting in there (you really shouldn’t go beyond 3 or 4 at a push). Variables are poorly named, non-descriptive. The data for 15000 lines of code is mostly global.

    It’s amateur level stuff. Like what a kid writing his first game with no discipline in the subject would do. The code uses threading which really isn’t for amateurs. Multi-threading can trip up quite experienced developers.

    I don’t even mind that Ferguson wrote this. If it’s academic processing, fine. But should we have paid for this and accepted it as a deliverable? No. Should we have used it in decision making? Absolutely not.

  7. Meanwhile the Ppeexchange.co.uk website, studiously ignored by the bbc, currently has in excess of 320 million items of PPE located in the U.K. and ready for immediate delivery.

    Just how difficult is this NHS procurement lark?

  8. “Of course, it could be that we’ve got anal retentives as inspectors too……”

    Given that we’re not told what was actually wrong with them, we’ll not find out until Later™…

    But a gown is simply a minor part of your PPE when it comes to CoVid. It just needs to be a spatter-barrier that can be stripped and burnt easily. A bin-bag will do in a pinch.
    So I do suspect some sphincteriffic puckering is involved in this one.

  9. ‘inspectors found the gear was “useless” and fell short of UK standards, senior sources said’

    No gear is better than off spec gear.

  10. Difficult to take seriously anyone who says “On a personal level, I’d go further and suggest that all academic epidemiology be defunded.” Not all epidemiology is about forecasting or actuarial-type work.

    The idea of having proper software engineers working on that kind of coding project is sensible, particularly if it is going to be used to make big decisions. But actually I would be more worried about the modelling itself than the software engineering, plenty of ugly badly-written code out there that basically does what was intended. John Carmack reckoned there were no killer bugs in it.

    https://twitter.com/ID_AA_Carmack/status/1254872370260652033

    (see the thread but also more recent tweets by same author)

  11. Perhaps the equipment actually is to specification but is just missing the CE mark and certificate? One of the UK notified bodies could be engaged to check and certify?

  12. Isn’t someone letting the best be the enemy of the good? We’re continually told that there’s a desperate shortage of PPE, to the point where staff who really need it are having to do without or stop working. Surely the Turkish stuff is better than nothing? Or are the shortages a figment of someone’s imagination?

  13. “John Carmack reckoned there were no killer bugs in it.”

    Define “killer”.

    What John Carmack seems to have done is to run code analysis tools on it. Which will spot really stupid syntax errors, but doesn’t tell you if it does even a half-arsed job of producing a result.

    As he also says:-

    “I can’t vouch for the actual algorithms, but the software engineering seems fine.”

    https://twitter.com/ID_AA_Carmack/status/1254878975459045381

    So, it’s a book with good spelling but maybe the facts are wrong, or a cab driver who takes you somewhere safely, but maybe not to the right place.

    It doesn’t have an accompanying specification. It doesn’t have any automated tests. The code is hard to understand because of how it’s written. It’s garbage as a living piece of software.

  14. It seems the evil Tories are only denying PPE to Labour Party activists, all part of some nefarious scheme to wipe out the opposition no doubt.

  15. @BoM4

    I’m not saying the code quality isn’t a worry especially for something that so much depended upon (though universities are going to need a bit more funding if they’re going to hire in appreciable numbers of professional software engineers from industry to support this kind of research), just that if there is a problem it’s rather more likely to be in the conceptual side of the modelling itself, rather than because of bugs in the code. You must have seen in your time some atrocious-looking production code which nevertheless did the job. Will be interesting to see if the folk trying to replicate the model in Python succeed, for example. I suspect they will given the comments from many of those looking at the code and the fact Ferguson’s results were pretty much in line with what other research groups were forecasting, but I’m pretty sure the critics are going to immediately move on to a position of “well that doesn’t prove anything, if two people implement the same wrong model they’ll just both get the same wrong results”.

  16. I heard that the UK planes freigting PPE from china are going out laden with bullion to pay for it.

  17. MBE,

    ” just that if there is a problem it’s rather more likely to be in the conceptual side of the modelling itself, rather than because of bugs in the code. You must have seen in your time some atrocious-looking production code which nevertheless did the job”

    I think you’re probably right.

    I’m only commenting where I am because that’s my field of expertise, and I don’t know much about pandemic modelling.

    This in the Spectator is rather damning.

    https://www.spectator.co.uk/article/sweden-tames-its-r-number-without-lockdown

    Sweden didn’t lockdown and never hit the predicted r number, nor were deaths even 1/10th of the Imperial prediction. And this, from Jonah Norberg might be the killer line about the Imperial model:-

    “In other words, the model could only handle two scenarios: an enforced national lockdown or zero change in behavior. It had no way of computing Swedes who decided to socially distance voluntarily. But we did.”

    I hope at the end of this that there isn’t just some big day of celebration with medals handed out to nurses but that we actually do a serious review of what happened. I’m starting to think it has been a very expensive mistake.

  18. Did you see the Banksy NHS nurse hero artwork, as my wife commented there’s usually an uproar when you depict a nurse in such old fashioned uniform (a la Barbara Windsor in Carry on Nurse for those of us of a certain age)

  19. @BOM4

    Understand your view. One of the advantages of getting proper software people to write the actual code is that it would have forced there to be a proper written specification to test against, which would have been very much more reassuring in terms of whether the code does what it’s suppose to, and allowing other groups to replicate and potentially improve or extend it. The focus of attention would then largely be on the specification and whether that really achieves its purported aim – plenty of the sceptical code-review comments on twitter are basically snarky, stylistic and largely superficial (which isn’t to say they’re inaccurate, just that in terms of “does it work as an epidemiological model” they’re missing what’s going on at the core).

    The way people in Sweden changed the way they live is important. I reckon a lot of people in the UK would have shifted quite dramatically too, particularly if more prodding had been given – lots of people were not working from home who could have been, for example, because until lockdown day the government didn’t explicitly say to do so where possible and many employers were holding out until such a signal was given. Incorporating behavioural responses to disease into the epidemic model is something the eggheads have clearly been thinking about in recent years (here’s a paper from 2015), but there are obvious reasons why it’s tricky.

    Funk S, Bansal S, Bauch CT, Eames KT, Edmunds WJ, Galvani AP, Klepac P. Nine challenges in incorporating the dynamics of behaviour in infectious diseases models. Epidemics. 2015 Mar 1;10:21-5.

    The interrelationship between the spread of an infectious disease and the behaviour towards it is subject to a number of dynamic feedbacks. Specifically, an outbreak of an infectious disease can trigger behavioural responses, which in turn can affect the course of the epidemic. Mathematical models provide an invaluable tool to study such feedbacks. Yet, behavioural dynamics have, until recently, rarely been incorporated in models of infectious disease dynamics. Taking into account individual behavioural heterogeneities and shifts in such models can be important because (1) predictions may be unreliable if they fail to take into account behavioural dynamics and (2) most policies target individual-level behaviour and not macro-scale dynamics.

    To formulate models in which infectious disease dynamics and behaviour are interdependent, we need to understand the mechanisms behind any mutual influence. To what extent do people themselves, their social “networks”, media opinion leaders, or health care providers affect individual behaviour? And how are the perceptions that determine behaviour influenced by properties of an infection, such as its prevalence or severity? There are often several ways of interpreting the same influence; in the case of disease prevalence, for example, people could respond to current prevalence, recent prevalence, or historical prevalence. Disease severity also affects behaviour (Sadique et al., 2013), but the relationship is not necessarily straight-forward: different responses will be prompted by a disease that infects 50% of a population and kills 1% of those infected versus an infection that infects only 0.5% but kills them all.

  20. @John

    Useful, thanks

    @Grikath

    +1 A bin bag is more than adequate

    Shame there aren’t hundreds of millions of cheap supermarket bags available for shoe covers, hats etc

    @Near the coast

    Yep. Or words in Turkish only, but CE etc numbers all there

  21. Bloke in North Dorset

    It’s nice to think that a fully specified and professionally written model could have been used but I’ll bet there isn’t one anywhere in the country. Nobody will have paid for on in advance, that isn’t how it works.

    Im guessing this model started out as an academic project, in the real meaning of academic, and then got applied to a real,world scenario and tweaked a bit. Each time a model was required Ferguson will have been the go-to guy and each time his team will have tweaked and added a bit more. A bit of SARS here, now some MERS, throw in a good dose of foot & mouth.

    By the time everyone in govt realised the needed a model he was the only guy they thought of, time being of the essence.

    I haven’t read the operation Cygnus stuff But I’ll bet nobody said what we really need is a good, well written, well documented epidemiology model.

  22. @BIND

    I reckon it would make more sense to leave academics to do the cutting/bleeding edge stuff, and if they show a novel modelling technique is useful then for it to get added to the spec of “live” disease and healthcare demand forecasting software owned in-house by PHE/NHS or similar. The way university research grant money works at present, “hire a software engineer to tidy up our decade-old code” is unlikely to be something academics are going to be able to invoice to anyone. Since new diseases do seem to come along every few years (already since 2000 there’s been SARS, MERS, H1N1 swine flu, H7N9 avian flu…) and you’d want our health services to be keeping an eye on how bad they could get, they seem to be the logical place to manage this for the long-term.

    As I understand it the reverse happens, and they actually subcontract out a lot of their modelling work in general to academia – I guess the £££ helps sustain a bigger ecosystem of academic modellers, which has benefits in terms of maintaining a pipeline for new ideas and training new researchers, but I’m not sure it’s the safest way to do things. Not just the difficulty it creates QAing code etc, also the “bus problem” or risk an academic research group gets swiped by a better-funded US university or has a terrible falling-out. Some stuff just seems too important to outsource.

  23. @MBE

    …lots of people were not working from home who could have been, for example, because until lockdown day the government didn’t explicitly say to do so where possible and many employers were holding out until such a signal was given.

    Depends what you mean by lockdown day; if you mean 26 Mar, when the Public Health (Coronavirus, Restrictions)(Nation) 2020 came into force, that’s just wrong. The PM explicitly said people should work from home where possible on 16 Mar. My chief exec was on holiday on the day, so I was the one who had to check that was the official advice after hearing a headline about the announcement and therefore to enact our contingency plan for homeworking.

  24. Bloke in North Dorset

    MBE,

    Yes, it’s a difficult one, trying to find a course free market innovation versus stultifying central planning. Perhaps a DARPA style completion every couple of years?

  25. @TomJ

    Cheers, I think you’re right / I worded it poorly. I definitely know people who complained they could and should have been WFH earlier but their bosses felt the first HMG guidance wasn’t compelling enough to. It’s the gap between people who already had a WFH infrastructure and maybe did so a few days per week versus those where nothing was in place that seemed to be the main issue. The first rumblings of government advice on the issue were quite late and quite mild, and companies in the “we don’t do WFH” camp (even if they could, they had culturally or managerially no interest in implementing a system for one) didn’t seem to think it applied to them. Then hurriedly put one in place when lockdown came in officially!

    @BiND

    Interesting point and I believe Gates funds occasional competitions or at least projects where two rival teams get set the same task to see how closely their conclusions agree.

Leave a Reply

Your email address will not be published. Required fields are marked *