[cupofcoffee]

Odds of death from AstraZenaca are 1 in 1 million. Let's take odds of the event of having serious complications to be 1/100,000 just to be more skeptical.

The odds that 5 people in a row have such serious side effect is 1/10^25. That is equivalent to tossing 83 coins in a row and getting all heads.

Which is more likelier. 1/10^25 or that you have false information regarding this. : )

[Apes]

I'm pro vaccine, and am convinced that the benefits of the vaccine absolutely eclipse the risks, but your math is so horrendously wrong here. Delete this.

Edit, here's some corrections:

This isn't 5 people in a row who had issues, it's 5 people within 4 nodes of distance in their social network. 4 nodes of distance is a lot of people, probably around 200,000 - 2,000,000.

The odds of having severe complications from the vaccine are about 1 in 100,000.

Therefore, everyone should expect to have 2-20 people within 4 nodes of their social network to have severe complications from the vaccine.

The mistake is not realizing the absolutely massive amount of people within 4 nodes of distance in your social network.

The other massive issue I see here is that there should also be somewhere around 500 covid deaths in the same pool of people that produced these 5 people with vaccine complications. The fact that they are focused on the 5 and not the 500 speaks heavily to their biases.

Keep in mind, all of this also assumes that what they're saying is 100% accurate, and these complications were definitely caused by the vaccine, and were not a coincidence. In truth, for every 1 person that has complications with the vaccine, 10-100x had the unlucky coincidence of something bad occurring that would have happened even if they hadn't gotten the vaccine.

[lossolo]

> probably around 200,000 - 2,000,000

There are around 200 strokes per 100 000 people per year [1].

And number of "heart issues" is even more, only deaths because of heart issues are around ~170 per 100k people per year.

Correlation is not causation.

[1] https://www.world-stroke.org/assets/downloads/WSO_Global_Str...

[bayesian_horse]

The numbers are probably even less favorable since the vaccinations have been skewed to the older population for much of the time.

[sethammons]

> The fact that they are focused on the 5 and not the 500 speaks heavily to their biases.

Same guy as before. I know of two people who temporarily lost taste and I heard of one person who died from COVID, a coworker's uncle.

You can rationalize this however you like, but these are the cases I'm aware of.

[cupofcoffee]

You are wrong about what is meant by a "social circle". They are people that he socializes with.

Definition of social circle is "A social circle is a group of socially interconnected people". I doubt anyone would complain about having a small social circle if it contained 2M people. :)

[Apes]

No.

> My daughter's bf's friends mom

[cupofcoffee]

They said: "I don't have a large social circle." You need to read context a bit more carefully.

[Apes]

Sure, he said that, but only one of the complications are actually IN his social circle, the rest are two or more steps removed.

"My dad's coworker" - 2 steps

"my wife's brothers" - 2 steps

"My daughter's bf's friends mom" - 4 steps

"A friend" - 1 step

He might have a small social circle, but even a small circle is going to explode exponentially when you start hoping outwards. By the 4th jump, you're going to start seeing extreme numbers regardless of how small your personal social circle is.

[cupofcoffee]

I stand corrected. Now it makes sense that only 1 side-effect in his circle.

Thanks.

[labawi]

To put the calculation in perspective.

What you have calculated:

(using 4 deaths at 1e-6 odds, because it's simpler and still gives 10x better odds at 1e-24)

Taking 4 death row inmates, jabbing them with AstraZeneca and expecting all 4 of them to die.

What you have not calculated:

Jabbing 1e9 people, 10k of them developing serious issues (expected prior) and there being a multi-hop connection between 5 of them. Two of them actually being brothers (familial clustering), and some with issues likely more prevalent so not even from the 10k group (1/1e5 cutoff).

[throwawayay02]

Your math is wrong, see the birthday paradox. But still, unlikely.

[cupofcoffee]

It's a back of the envelope calculation. I've a strong intuition that the precise answer isn't too far off.

[labawi]

The math is monumentally off. It fixates on 5 specific people having issiues, instead of: There exist some 5 loosely connected people that some HN-er heard about, and probably hasn't even met all of them in person, especially the ones that died.

The implied qualification of - there exists a HN-er so divide by 1e9 to get odds - is nowhere close to accounting for the additional degrees of freedom from the social graph.