Bayesian analysis
The FT has a good article that might be seen as being tangentially related to my earlier post on lab leak vs. zoonosis:
If a screenwriter were to come up with a storyline in which a tech tycoon drowns when his luxury yacht is hit by a freak storm just two days after his co-defendant in a multibillion dollar fraud trial — for which both men were recently acquitted — is fatally hit by a car in another set of ostensibly unsuspicious circumstances, they might very well be told this was rather too implausible for viewers to buy.
And yet this was the tragic real-life series of events over the past week or so. . . .
It didn’t take long for the conspiracy theories to start. Pro-Russia personality Chay Bowes posted on X a clip of himself speaking on the Russian state-owned RT channel in which he pointed out the low probability of being acquitted in a federal criminal trial in the US — about 0.4 per cent, according to Pew. “How could two of the statistically most charmed men alive both meet tragic ends within days of each other in the most improbable ways?” asked Bowes.
The article also mentioned that back in 2009, two identical numbers came up back to back in Bulgaria’s national lottery, at odds of four million to one.
PS. The name of Mike Lynch’s yacht? Bayesian.
No, I’m not making that up.
Tags:
25. August 2024 at 16:19
It was revealed to me in a dream that the Chinese government deliberately created COVID (also, that Shi Zhengli is dead) -after all, the Chinese government seemed a little too well prepared for the pandemic.
Now, to be fair, it is not safe to trust my dreams.
25. August 2024 at 17:34
Only 0.4% of federal defendants were acquitted at trial in 2004, but it is highly misleading statistic. One should cite either 8.6% who were acquitted or for whom the charges were dismissed. Or 0.4/2.3 = 16% who went all the way to the verdict and were acquitted.
To the statistician, first, do not mislead with numbers
25. August 2024 at 20:14
At the core, it is the failure to account for randomness and the dominant role it plays to the outcomes of even the most deliberate actions.
In the article, she mentioned
“Sometimes, though, all we will find is that truth can be stranger than fiction. ”
I would say the more appropriate word should be “Always” or “Often” instead of “Sometimes”.
Another related case is regarding Lucy Letby:
https://www.msn.com/en-gb/news/uknews/lucy-letby-serial-killer-or-a-miscarriage-of-justice/ar-BB1pF6Wm?ocid=BingNewsVerp
26. August 2024 at 08:18
I feel bad for the Lynch family and everyone else involved, but when I read that the vessel was called Bayesian, I couldn’t help but think that Lynch’s hubris had brought down the gods’ wrath upon them.
Off topic, I had a small heart attack going through your latest posts, Scott, when I saw you were closing down the blog! I’m not sure when I first began commenting here regularly but it was certainly a good few years ago. Glad to hear you will continue the blogging, just at a new location. I’ll rather miss this format, however. It’s a nice throwback in an era when so many people are on Substack.
26. August 2024 at 10:12
Everyone, The mysterious missing comments have now been restored to “A final few links”.
Harding, No, they were horribly unprepared. Have you forgotten? If prepared, it never would have left Wuhan.
D.O. Thanks, I’m not surprised.
Mark, And today’s Econlog has another case:
https://www.econlib.org/ignoring-probability-theory-is-dangerous/
Tacticus, Thanks. Get 1000 other readers together and bribe me to stay here. 🙂
27. August 2024 at 11:19
“If prepared, it never would have left Wuhan.”
But for the most part, within China, it didn’t leave Wuhan over the course of 2020.