The right messaging strategy long-term is to be transparent, honest and rational. Shortcutting this is risky, through the three mechanisms I mentioned in last comment.

SWP doesn't primarily focus on ablation. Where they do, they should keep in mind and make it clear that they're talking about <0.1% of farmed shrimp.

Two benefits of giving now rather than in your will in 80 years:

1. It's one small step in normalising giving. People are more likely to consider donating their money if they have people around them that do. You will nudge people.

2. Doing good accumulates compound interest. Empower someone to live a better life today rather than 80 years from now and they have and extra 80 years to be a productive member of society and help themselves, their family, their community.

Sounds like a recipe for:

1. Reinforcing the stereotype of vegans/vegetarians as “bleeding hearts” who don’t think about practicalities.
2. Misguided regulations/legislation in the style of bee bricks that contribute to society’s red tape fatigue without having positive impact.
3. Reduced trust in the effective animal welfare movement (I’ll now be more skeptical, less trusting that I’m being told all relevant info)

It's contrary to the philosophy of Effective Altruism to be relying on or supporting people's "gut level" vibes.

I can think of many examples of ineffective charities that you could justify in the same way because they "capture in a nutshell the low intrinsic moral value that [people] are assigning to..." some neglected group:

1. Soccer balls for village kids in rural Uganda
2. GoFundMes for peoples' sick and injured pets
3. Knitted vests for penguins
4. Free iPads for kids in low-performing schools
5. Drama therapy for prisoners

The risk with playing into people's gut feelings is that down the line when they find out that you're actually talking about fewer shrimp than would feed a whale for a week, they're going to feel duped, trust is lost, reputation ruined. So on.

This sucks

Using different countries as cohorts introduces all sorts of confounding factors.

People will put forward a biased case even without a financial incentive. Maybe a person genuinely believes that the weight-loss or skincare regimen they followed worked, and they want to put the best case forward so other people benefit from it, so they use the worst Before and best After photo to be more convincing. People who saw no difference won’t post at all.

The reason RCTs exist is because time and time again we’ve seen that looking at anecdotes and individual case studies leads to conclusions that turn out to be wrong when you RCT them. Might seem nitpicky but you just end up wrong half the time otherwise.

In 2016 I took part in a novel drug trial in Brisbane, Australia that injected me (and about 6 other men) with the Plasmodium falciparum strain of malaria and then treated us with a new medication called SJ733. The development of SJ733 was funded, I was told, through Bill Gates' Medicines for Malaria Venture. The paper about this trial (and some other trials) came out in 2020: https://pubmed.ncbi.nlm.nih.gov/32275867/
Results were positive!
Seems like work on it continues: https://pubmed.ncbi.nlm.nih.gov/35598441/ 
I gave the $2880 they gave me to the Against Malaria Foundation. It's one of the best things I've ever done.

I didn’t see case-control studies or cohort studies. You should link to those.

Before and after photos are prone to manipulation and bias
Take a photo with your jaw pulled back for the before and then one with your jaw projected forward for the after. Use better lighting for after. Even if the person is not consciously intending to, these are so easy to manipulate.

Maybe surgery was done in some of those photos and the poster is not being forthcoming about that, if they have something to sell.

The evidence for mewing just doesn’t seem to be there. You give a lot of theory and anecdote and before/after photos, which aren’t worth much. The main proper source I see you list is https://academic.oup.com/bioscience/article/70/9/759/5872832?login=false, which itself seems to admit that there’s a lack of evidence for it

Another cost with a failed moonshot is damage to reputation. If I think that there is a 5% chance of another pandemic in the next 10 years and I spend the next 5 years working to mitigate it, there is a 95% chance that everyone who tells me I am crazy will end up looking like they were right.