Here's an example illustrating how bad of an argument that is:
It appears to be reasonable to assign the "your mother is a [insert derogatory term]" hypothesis a higher prior, in Bayesian terms, just based on the many familiy members whose image/status would it directly or indirectly damage if found true (your dad, your uncle, your sister, your grandparents), and which therefore would be compelled to paint it as unlikely.
But I choose to keep this non-insult since it has a very strong impact in just a few words due to the high cognitive dissonance it causes. (Partly responsible for the chuckle you mentioned).
The earth being flat is not a good substitute btw since it's also in the topos of conspiracy theories thus reducing the cognitive dissonance.
But I'd be happy for other suggestions.
> It appears to be reasonable to assign the "your mother is a [insert derogatory term]" hypothesis a higher prior, in Bayesian terms, just based on the many familiy members whose image/status would it directly or indirectly damage if found true (your dad, your uncle, your sister, your grandparents), and which therefore would be compelled to paint it as unlikely.
If something major happened that made it a difficult binary choice between either that and <insert an alternative>, then certainly it would be reasonable to assign appropriate priors if you were forming your opinion exclusively by interviewing my relatives. (On the other hand, if my family members or relatives were in positions of significant power/esteem/status that they stood to lose if what you say were true, had significant social influence and a degree of control over mass media, then reweighing priors in such fashion would be advisable regardless of your source.)
Hi there. My intention was not to make this about your literal family. The "your" was intended to refer to a general addressee. Sry for that.
In fact, it would make much more sense to argue the opposite (not that I would necessarily agree with this reasoning either). The fact that so many institutions would see their reputation damaged if a lab leak was known to have happened makes it less likely for the virus having originated from a lab leak a priori, as said institutions would put in place strong measures to avoid it.
On the Bayesian level though, what can make sense would be to argue that the fact that so many institutions would be damaged post hoc if a lab leak were to be known, then the weight of evidence against the event which has been provided by such institutions should be lower.
Wouldn’t weighing evidence play a key role in assessing relative likelihoods of two possible causes of an event that had already occurred?
Especially since any remaining evidence is not conclusive. (As it happens, there appears to be evidence of some of the parties involved acting in ways that prevent further evidence gathering, too.)
Absolutely, yes. However, it would not affect the priors.
I was going to joke, but this is NOT TRUE AT ALL
Based on my life experience, a more correct idea is, "as said institutions would put in place strong measures to avoid appearing to have caused it"
This tilts the argument back the other way.
Look, I don't know about lab vs natural origins of COVID-19, but I do know that big, human orgs tend to have more CYA theater than actual preventative measures in place.
Regarding your own interpretation of the argument, i.e. that large human organisations are more likely to have ineffectual preventative measures, therefore increasing the prior probability of a lab leak, that would technically be a valid argument. However, in this case in particular, assuming that premise raises the question "Compared to what?". It's not like there is a cottage industry of artisanal HS4 laboratories that we can say are better at following safety requirements than large ones.
I myself have no strong opinion either way regarding the lab-leak vs. natural origin hypotheses. In fact, a while ago I thought of a "consensus" hypothesis, which is that the origin was accidental human infection during sample collection in caves. However, this possibility would fit better with a purely wild-type virus, and thus would not be consistent with the sequence oddities explained in the linked Twitter thread.
This is the same kind of issue as why it’s important to separate the regulator from the regulated.
It's called a prior for a reason.
A perfect coin has a 50/50 chance for heads. That is a prior. Getting 10 heads in a row does not change that prior. Getting 1 billion heads in a row does not change that prior.
Unlikely outcomes don't change the prior. The prior is the probability distribution of beliefs before evidence is taken into account.
However, what about the prior probability of a good-faith statement in presence of a strong conflict of interest? Having no ability to directly verify the truth, wouldn’t you require stronger evidence if you were to rely on that statement in your decision-making?
The definition of your prior would incorporate that.
>Having no ability to directly verify the truth, wouldn’t you require stronger evidence if you were to rely on that statement in your decision-making?
That is exactly the purpose of priors and Bayesian belief systems. If you could verify any truth, you'd not need a prior or Bayesian probability concepts for that.
In none of these cases do you change the prior after some evidence is obtained, You'd use the prior and Bayesian update to make a new belief, which is then the prior for making decisions as future evidence is obtained.
Retroactively changing the prior is demonstrably bad at reaching the proper conclusions over time. E.T. Jaynes "Probability Theory" [1] has detailed proofs of all of this, covered in the first few chapters.
[1] https://www.cambridge.org/core/books/probability-theory/9CA0...
I support neither treating an accidental leak (if it occurred) as a crime, nor blaming any individual for it.
That aside, the key is not quite “existence of people who could be blamed” but “existence of people who stand to (or think they stand to) lose a lot in terms of image/power/funding and have a certain degree of influence over the outcome”.
(Before you ask why am I not blaming Bill Gates for this pandemic then, there are of course other considerations that go into weighing priors. Using only this factor alone makes it indistinguishable from conspiracy theorising.)
Also the principle of adverse interference. China destroyed/made unavailable a lot of documents. It is reasonable to infer that the destroyed documents had evidence of a lab leak.
If you consider those as viable hypotheses in the first place then sure. In current context there is a binary choice between two already established and well-informed hypotheses of similarly high likelihoods, and if one of them is significantly more favorable to a party with control over evidence we might expect a stronger proof.