The Bill Gates connection is the laugh line here although not mentioned.
I did my own projections on the back of an envelope.
It ain’t that hard.
Well, who would have thought an ‘epidemiologist’ would have a Crackerjack mathematical biology diploma...../s
Follow the Science, they said
Trump should let the science experts manage the national response, they said
Fauci and Birx are the smartest scientists in the US, they said
Funny, they are still saying this
The author is bloviating. He may be right on a few details, but he misses the point all together.
It doesnt matter how many lines of code there are. Its the basic algorithm that drives the “answer”.
This part he gets right. The problem, as I have stated many times now, is that trending, fitting historical data, whatever you want to call it, does not lead to good prediction. The only way to do that is by having a model of the complete process that has been proven out with real data. There are no good models that have all the details of both viral growth/decay in the real world (vice a lab) AND the infinite impact of human behaviors and interactions. All you can do is postulate.
Reminds me of the Obamacare website which cost something like a billion dollars and basically could not function as a website.
Fraud? or Incompetence? Sometimes the lines seems very thin.
I appreciate that the author does not get caught up in which language the program was written in. He does get worked up about the program wanting one core. Adding cores allows more speed and complexity, but does not increase accuracy (except that you have more time to run more simulations and scenarios).
Hope he never looks at the models for climate change. They might make this one look accurate.
In the most unprofessional manner imaginable, the Imperial College code does not even have a regression-test structure. They apparently attempted to but the extent of the random behavior caused by bugs in the code to prevent that check? On April 4th, 2020, Imperial College noted:
However, we havent had the time to work out a scalable and maintainable way of running the regression test in a way that allows a small amount of variation, but doesnt let the figures drift over time.
This Ferguson Model is such a joke it is either an outright fraud, or it is the most inept piece of programming I may have ever seen in my life
However...For example one model tried to predict number of deaths yet the 95% confidence range was across 3 orders of decimal magnitude. Since the actual number of deaths was in this range the authors declared success. This model was cited by a number of others.
Another model tried to deduce infectiousness of asymptomatic carriers using phone tracing data from Tencent. Once again there was a glowing conclusion in the abstract and plenty of cites, but the supplement revealed number of deaths 95% confidence range of 10x from high to low. Plus the authors admitted a wide range of values for number of asymptomatic carriers and infectiousness were actually possible, they settled on the values published after repeated iteration - specifically choosing values that kept the model numerically stable in a MatLab solver library.
I cringed and stopped reviewing models.
Complicated code is bad code. Period. If you need someone to explain what they’re doing with the code, they are not good coders.
As St. Exupery famously said, “Perfection is attained, not when there is nothing left to add, but when there is nothing left to take away.”
Lets just say Ferguson’s mind was on other things — like banging some man’s wife in the midst of a pandemic.
This code and a code of conduct are thrown by the wayside. Both are in ruins and so is his reputation.
as we from the 70’s used to joke, “this sounds like a big Kludge.”
Gates indeed. some friends of mine used to joke back in the day that they needed to keep “Bill” away from the code.
however, no joking here, unlike Gates and DOS/Windows, this software is killing people.
It’s very clear we don’t have to understand or even look at the details of the code in order to know the whole thing was a sham.
to many GOTO statements ?
More than 5 million Americans will be infected with coronavirus and 290,000 will die by the end of July if social distancing isn’t adhered to, according to COVID-19 model
https://www.dailymail.co.uk/news/article-8340325/5million-Americans-infected-COVID-19-July-model-shows.html
What if Ferguson screwed up on purpose...
Do NOT cite Martin Armstrong to denigrate Ferguson's code.
I had saved this post's citation for later research and discovered the following (in a mere 60 seconds) of some background:
"In September 1999, the Securities and Exchange Commission and the Commodity Futures Trading Commission prosecuted Armstrong for fraud. He was imprisoned for over seven years for civil contempt of court, one of the longest-running cases of civil contempt in American legal history. In August 2006, Armstrong pleaded guilty to one count of conspiracy to commit fraud and began a five-year sentence. Armstrong appeared in a 2015 documentary titled The Forecaster regarding his mathematical model and prosecution. [4],[5], [11]"
https://www.desmogblog.com/martin-armstrong
This NYT article makes the following citation:
"All told, by the time Mr. Armstrong is released, he will have served 12 years for orchestrating what prosecutors called a $3 billion Ponzi scheme through his investment fund, Princeton Economics International."
There are allegations that Armstrong was an insider with knowledge of gold market manipulation, but without hard evidence such allegations must reside in the same folder as The Clinton Body Count and, IMHO, citations of Armstrong pertaining to his "analysis" of Ferguson's computer code should be avoided.
In fact, there are other, potentially-valid sources of citation for analysis of Ferguson's code. Unfortunately, they are behind paywalls:
Neil Ferguson's Imperial model could be the most devastating software mistake of all time
Corona modeling was "worst software bug ever"
Perhaps a lurking FReeper can shed some light on the content via a friendly source with subscription access???
In closing, I cite the only valid analysis I could source for Imperial College's code (of which the source code is still not released):
Code Review of Fergusons Model
Author credentials:
"I have been writing software for 30 years. I worked at Google between 2006 and 2014, where I was a senior software engineer working on Maps, Gmail and account security. I spent the last five years at a US/UK firm where I designed the companys database product, amongst other jobs and projects. I was also an independent consultant for a couple of years."
Unfortunately, the author (perhaps righteously) prefers to remain anonymous.
Citation:
"Conclusions. All papers based on this code should be retracted immediately. Imperials modelling efforts should be reset with a new team that isnt under Professor Ferguson, and which has a commitment to replicable results with published code from day one.
On a personal level, Id 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 dont have these people, and the results speak for themselves."
Follow-up:
Second Analysis of Fergusons Model
In summary, the code is SO defective that it cannot reproduce its own results and produces different results when run on multi-processor systems.
"For standards to improve academics must lose the mentality that the rules dont apply to them. In a formal petition to ICL to retract papers based on the model you can see comments explaining that scientists dont need to unit test their code, that criticising them will just cause them to avoid peer review in future, and other entirely unacceptable positions. Eventually a modeller from the private sector gives them a reality check. In particular academics shouldnt have to be convinced to open their code to scrutiny; it should be a mandatory part of grant funding."
Do tell. It is a travesty that truthsayers need to stay anonymous to protect themselves from punitive persecution for daring to come forward...
Note: "ICL" = Imperial College London.