Royal Society cautions against censorship of scientific misinformation online

We produce more science now than ever before. If you have 10 scientists and 9 of them produce rubbish it won't take you long to read a paper from the one who doesn't.

If you have 10,000,000 and 9,000,000 produce rubbish you can spend a thousands lifetimes reading nothing but rubbish.

They really haven't. They continued to double-down on ivermectin and other COVID era flim flam as it came to light more studies were dodgy or outright fraudulent. It's pretend science theatre from a former small university lecturer who managed a couple research papers in 20 years. For example, telling the audience with a straight face that it doesn't matter if the studies going into a metaanalysis are biased because the errors will cancel out.

Yes, but that's useful too.

COVID-19 lockdowns largely kicked off in the west due to an epidemiological model from Imperial College London, written over a period of many years by the now notorious Professor Neil Ferguson.

When his team uploaded a pre-print of his paper and started sending it to government ministers, the code for his model wasn't available. They spent months fighting FOIA requests by claiming they were about to release it, but just had to tidy things up a bit first. When the code was finally uploaded to GitHub the world discovered the reason for delay: the model was a 15,000 lines of C trash fire of race conditions, floating point accumulation errors and memory corruptions in which basically every variable was in global scope and with a single letter name. It was a textbook case of how not to write software. In fact it was a textbook case of why variable names matter because one of the memory corruptions was caused by the author apparently losing track of what a variable named 'k' was meant to contain at that point in the code.

Not surprisingly, the model didn't work. Although it had command line flags to set the PRNG seeds these flags were useless: the model generated totally different output every time you ran it, even with fixed seeds. In fact hospital bed demand prediction changed from run to run by more than the size of the entire NHS Nightingale crash hospital building programme, purely due to bugs.

And as we now know the model was entirely wrong and the results were catastrophic. Lockdowns had no impact on mortality. There are many people who looked at the data and saw this but here's just the latest meta-analysis of published studies showing that to be true [1]. They destroyed the NHS which now has a cancer treatment backlog and pool of 'missing' patients so large that it cannot possibly catch up, meaning people will die waiting for treatment from the system they supported with their taxes for their entire lives. They destroyed the tax base, leaving the government with an unpayable debt that can be eliminated only via inflation meaning they will have soon destroyed people's savings too. It's just a catastrophe of incompetence and hubris.

The incorrectness of the model wasn't due only to programming bugs. The underlying biological assumptions were roughly GCSE level or a bit lower (GCSE is the exams you take at 15/16 in the UK), and it's quite evident that high school germ theory is woefully incomplete. In particular it has nothing to say on the topic of aerosol vs droplet transmission, which appears to be a critical error in the way these models are constructed.

Nonetheless, even if the assumptions were correct such a model should never have been used. Anyone outside the team who had access to the original code would have seen this problem immediately and could have sounded the alarm, but:

1. Nobody did have access.

2. ICL lied to the press by claiming the code had been published and peer reviewed years earlier (so where was it?)

3. Then when it was revealed the model wasn't reproducible, they teamed up with another academic at Cambridge and lied again by publishing a report+press release claiming it actually was reproducible and claims otherwise were misinformation.

4. And then the journal Nature and the BBC repeated these false claims of reproducibility.

All whilst anyone who looked at the GitHub issues list could see it filling up with determinism bugs. If you want citations for all these claims look here, at a summary report I wrote for a friendly MP [2].

So. It's good that the Royal Society is telling people not to engage in censorship, but their justifications for taking that stance reveal they're still living in lala land. By far the deadliest and most dangerous scientific misinformation throughout COVID has come from the formal institutions of science themselves. You could sum all the Substacks together and they would amount to 1% of the misinformation that has been published by government-backed "scientists", zero of whom have been banned from anything or received any penalty whatsoever. For as long as the scientific institutions are in denial about how utterly dishonest their culture has become we will continue to see a weird inversion in which random outsiders point out basic errors in their work and they respond by yelling "disinformation".

[1] https://sites.krieger.jhu.edu/iae/files/2022/01/A-Literature...

[2] https://plan99.net/~mike/epidemiology.pdf

However, even if it delivers to you true information about abuses and lies being told to you, you should still be aware that the system is doing this so that you will be outraged, rather than that you will be informed.