Thread by @Gilesyb

Determined to be even-handed, I have forced myself to read this to see how much can be easily refutable. Tl;dr - most of it, easily
1/

https://twitter.com/toadmeister/status/1347441658200260610

First, this bit about things being no worse than a normal winter. Clearly wrong, as shown by @jburnmurdoch in this tweet stream

https://twitter.com/jburnmurdoch/status/1347200855376875523

But also, there are cancelled urgent operations in the NHS. Does he think they do this for fun? 2/

Second, this bit that argues "well the lockdown won't make any difference". Again bizarre, because a. it makes an argument for a tougher lockdown, and b. clearly it does make a difference. Before, the schools were going to open. Now they are not. Contacts are reduced 3/

Third, this bit is sheer model-Luddism "Your model relies on unseeable counterfactuals, so I don't want to accept the results". Tough. We can only run reality once, so we use models. These are not complicated. That life is full of confounding variables =no reason not to try 4/

Fourth: "But it is incontestable that lockdowns cause harm". Compared to letting it run riot, that is not proven. At all. Particularly given the repeated pattern that failing to act early means having to act later, and harsher ...5/

and for someone who *just a few paragraphs earlier* dismissed model-based results, quoting without caveat as a fact the idea that ongoing restrictions will cause 130 million people to STARVE TO DEATH is just insane.

More than the entire famine death toll of the C20? please 6/

Fifth, here is the most ridiculous self-own.

"If the new variant is so much more infectious why did it fall in November?" BECAUSE THERE WAS A LOCKDOWN, WHICH YOU SAID CANNOT WORK.

sorry to shout, but honestly ... 7/

Anyway: I don't want to try for @NeilDotObrien mantle of supreme refuter of Lockdown Sceptics. I don't question their right to publish their stuff. But the reason to respect these rights is that bad arguments should be fully surfaced, so they can be destroyed. /ends