I have been in the Light World, then the Dark and I've been working on escaping it for years, with partial success. It felt good to read this essay, the fact that the move is doable but takes a ton of work and time resonates with my expérience. It's a small example, but the example of the child psychopath who recovered thanks to professionnal help gave me hope. I have had to renunce to parts of my integrity, empathy and trust in the world and others in order to deal with difficult situations and build for myself a safe environment. I am working on gaining them back, but woah that's hard ! 

Some things which helped me :

• Internal Family System and learning to communicate with love to all my parts, including the angry hurt ones, including the irrationnal ones
• Going no contact with my mother, who was the main source of emotional violence. Hardest choice of my life, I don't regret it.
• Making new friends who are kind, warm and empathetic - and learn from them. Having old friends I trust a lot helped too.
• (not sure I would recommend it to anyone else, but it was necessary for me to change pattern) temporarily stopping to pursue big goals and limitting my responsabilities. Numbing my feelings was the only way I knew to deal with deadlines and responsibilities. Getting rid of external pressures gave me the space needed to learn how to explore and process my feelings in an healthier way. Cons : not always possible + create a disconnect from the world and from a sense of purpose

I agree with the fact that there are many examples of exponential growth in real life, but this post seems to overstate their importance. 

Here, I feel that approximating a little quantity to zero led to a false conclusion. A lot of cool (and I guess bad) things in life are present in small quantity and it's enough for them to be valuable and have a strong impact. There is one ethical breeder of the breed of dogs you love in France (example taken from a friend ;) ) : that's all you need ! If you were limited to the main topics in conversations, it would get boring very fast. Thankfully, there are plenty of small new topics of conversations that appear when you talk to new people ! And, taking an example closer to LessWrong, you only need one powerful AI with weak safeguards to get the world into trouble.

Ah yes, you're right. I don't know why but I made the mental shortcut that the mutation rate was about the DNA of cows / humans and not the flu virus.

The general point still holds : I am wary of the assumption of a constant mutation rate of the flu virus. It really facilitates the computation, but if the computation under this simplifying hypothesis leads to a consequence which contradict reality, I would interrogate this assumption.
It's surprising to have so few human cases considering the large number of cows infected if there is a human-compatible viron per cow. 

Another cause of this discrepancy could also be that due to the large mutation rate, a non-negligible part of the virons are not viable / don't replicate well / ... 

There are papers which show heterogeneity for influenza / RNA viruses but I don't really know if it's between the virus population (of the same kind of virus) or within the genome. And they are like a factor 4 or so in the papers I have seen. So maybe less relevant than expected.

Regarding the details, my lack of deep knowledge of the domain is limiting. But as a mathematician who had to modelize real phenomenon and adapt the model to handle the discrepancy between the model's conclusion and reality, that's the train of thought which comes naturally to mind. 

I am no biologist but I thought it would be fun to give it a try. Hoping it's not too late to participate.
For the purpose of this experiment, I assumed everything written in sound scientific papers was right, as I had neither the time nor the knowledge to do a proper truth-check.

Here is what I found :  

The fact that we see very few human cases compared to what we could expect for a virus which can become human-compatible with a single point mutation could be partially explained by the heterogeneity of the mutation rate over the genome:

•  in this paper, a 1 to 100 ratio of mutation rate inside the human genome is mentioned : https://pubmed.ncbi.nlm.nih.gov/35218359/
     There are known factors which hints at a low / large mutation rate for a sequence, so one could investigate further whether this would be the case for the "single point mutation" mentioned
  -> you can divide the probability by 100 in the best case

A paper where they infect ferret with H5N1 and test how to mutate the genome to make it airborne:

• https://pmc.ncbi.nlm.nih.gov/articles/PMC4810786/
• "Four amino acid substitutions in the host receptor-binding protein hemagglutinin, and one in the polymerase complex protein basic polymerase 2, were consistently present in airborne-transmitted viruses."

As an alternative to tamiflu which can also be used in combination with it : favipiravir

• https://www.sciencedirect.com/science/article/pii/S0163725820300401, a 2020 paper about it
• licensed as an influenza drug in Japan
• when used in combination with tamiflu, a 2020 paper (with small sample size = 40) suggests that combining tamiflu with favipiravir improve significantly the efficiency
• as it has been envisioned as a drug against COVID, it's easy to find meta-analyses on the side-effects of this drug
• a huge problem with tamiflu is that the influenza viruses develop resistance to it. favipiavir does not seem to have this problem

The discussion on the impact on false positive / false negatives would be more fair if you also discussed the negative impacts implementing bayesian punishment would have. For example, if you start giving small punishments for crimes with low credence of guilt, that would not be punished in the current system, this will add its lot of false positive. 

I would not be confident it would be a good idea to implement this in our current justice systems. It may have a negative impact on people's faith in justice (is it deserved ? yeah ! is it good ? not sure) and my view of the justice system is that it's an essential part of society which is quite fragile as it relies in part on the lie that the justice system is fair and that it's decision is the truth. Plus, more generally, I don't think judges would handle the credences correctly, it's quite a difficult task to transform an heteregoneous and large set of proofs, among which testimonials, into a credence. 

Thanks for the post ! I have tried doing EMDR by myself following these instructions after struggling to find a reliable EMDR therapist and it's definitely a new addendum to my therapeutic toolbox. It's great to process difficult events that happened in the past and the subsequent feelings.

In the link you refer to, there is a phase at the beginning where the therapist teaches self-care techniques to handle strong emotions (or they ensure that the patient already has such tools). This was lost in your post and I think it is important in order to avoid retraumatization and bad experiences. I felt confident trying the technique because I know how to soothe myself if needed and be present with my feelings, but I am pretty sure it would have been a scary or even maybe a painful experience otherwise.