[timr]

The Leung et. al. paper (first link) should never have passed editorial review. Look at figure 1: the only significant result in the entire paper (1a, panel 3) is based on four data points. I don't know if this is p-hacking, per se, but it's not a robust result. And not that I give the paper a lot of credence, but it's worth nothing that the overall trend across all pathogens is that masks are ineffective against aerosols.

Link two is a news article and contributes no data.

Link three is a survey of people self-reporting a bunch of different things, where they've thrown out 32% of the data for non-response, selected the ones who did respond (bias!) and used that to make claims about face masks. Moreover, their data shows that as people wear face-masks more often, their chance of getting Covid goes up! This paper has so many confounders and potential sources of bias that it simply cannot be taken seriously.

When I said that the research literature on masks over 2020 hasn't gotten any better, these kinds of papers are exactly what I meant. They're terrible, they're littered with methodological and data analysis flaws, and provide no useful information to the debate. If these "seem clear to you", you don't understand what you're reading and lack the ability to assess research literature.

Sadly, top-tier journals publish a lot of garbage, particularly when that garbage is topical and will get press. You cannot use "Nature paper" as a badge of quality. You still have to read and understand the paper.

[kgarten]

For me that these papers get accepted is rather a sign that scientific consensus is on the side of masks work.

Usually, it takes 2 -5 years to get a nature paper ready (at least for the labs I know). Seeing also the pre-prints that were in the article that you dismissed as "news article." and others it seems a strong indication that most virologists and epidemiologists think masks work.

They might be wrong, yet I haven't heard any convincing theory or model on why that should be the case.

[timr]

> For me that these papers get accepted is rather a sign that scientific consensus is on the side of masks work.

Well that's a great way to confirm your biases. The "scientific consensus" was that the earth was the center of the universe, that infection was transmitted by miasma, that plate tectonics were a crackpot theory...I could go on.

Science is the history of well-controlled experiment overturning consensus thinking.

In any case, editorial review is a human process, and like all human processes, has ample sources of error. In this case, a big source of error is that the major scientific journals have spent the better part of 2020 falling all over themselves to publish garbage about Covid that gets press hits.

> Usually, it takes 2 -5 years to get a nature paper ready (at least for the labs I know).

This obviously isn't true in this case. The pandemic hasn't been around for that long!

Also, no, it doesn't take 2-5 years to put together an editorial. Even in normal times.

[kgarten]

> The "scientific consensus" was that the earth was the center of the universe.

Yes, and then somebody found a better model (that's also still wrong) yet explained more and we used it. So my question, what's your better model you base your assumptions on? Any scientific paper is wrong, you can find problems with any of them. Just, critique is easy and often doesn't lead anywhere.

my point: The scientific consensus right now is wearing masks help. This is supported by physics (particle simulations), medicine (respiratory disease and their transmission .. and a lot of other fields.

You say there's little data ... come on than make this > well-controlled experiment overturning consensus thinking

you can be famous!

https://www.japantimes.co.jp/news/2020/10/22/national/scienc...

Check the video, people are trying what would be an experimental setup you would accept? That is also ethically OK.

We have both controlled experiments that have reduced validity in the real world because of the artificial conditions and we have some more messy real world evidence. Seems OK to me.

The people I see on the other side don't offer an alternative model. Also a lot of the criticism comes from people who don't seem to have virology /immunology etc background.

> Usually, it takes 2 -5 years to get a nature paper ready (at least for the labs I know).

Yes, my point was these papers are rushed. It can be bias, it can be also that reviewers have their expertise that tells them the paper fits the evidence.

Change happens over models/explanations. I don't see any new ones from the "Masks don't work" folks. Just criticism (that's easy ... I'm happy to point out all the flaws in any paper you present me. If you want to kill a paper in peer review it's super easy ...)

For policy decisions shouldn't you use the current scientific consesus,however flawed it is? Is there a better solution? Otherwise you are just relying on your gut feeling. Good luck.

[kgarten]

forgot: here's the paper (it's just mentioned in Japanese in the video) https://journals.asm.org/doi/10.1128/mSphere.00637-20

your controlled experiment with Sars-COV-2 particles :)

[timr]

from the paper (emphasis mine):

> Here, we developed an airborne transmission SIMULATOR of infectious SARS-CoV-2-containing droplets/aerosols produced by human respiration and coughs and assessed the transmissibility of the infectious droplets/aerosols and the ability of various types of face masks to block the transmission.

Mannequin heads in a box is perhaps evidence of something, but it has little to do with actual humans who don't live in tiny boxes with masks sealed to their real-world faces.

[kgarten]

https://med.stanford.edu/news/all-news/2021/09/surgical-mask...

Another one ...

[kgarten]

lol ... yes simulations and models are completely useless (wake up: ALL models are wrong, but some are useful). If you cannot come up with a better explanation why masks should not work, you are just somebody who's not accepting the current scientific consensus.

So where's your experiment showing the contrary? (With real humans and covid particles?) Where is your evidence?

So we have simulations, evidence that it controls community spread, etc. Every expert that I hear speak out says that masks work. Can you find any virologist working in a lab on COVID who claims masks don't work?

https://www.pnas.org/content/118/4/e2014564118 https://jamanetwork.com/journals/jama/fullarticle/2776536 https://pubmed.ncbi.nlm.nih.gov/33431650/

[kgarten]

Also they just tested one mask. In HK, Taiwan and Japan a lot of people are wearing 2 masks: 1 surgical mask and a cotton mask on top :)

Doesn't help at all (check Taiwan's and HK's numbers compared with Texas ... much higher population density).

[kgarten]

we have the controlled experiments, correlations in infections and annectodal evidence in individual incidence: https://www.cdc.gov/mmwr/volumes/70/wr/mm7035e2.htm

what experiment setup that has not been done would convince you? please descirbe it.

[kgarten]

in terms of physics I see a good reason to wear a mask. If COVID did not invent a sneaky way to go around that

Article 2 has a whole lot of pre-prints with data in the citations ... :) Here's a follow up (also with a lot of paper citations : https://www.nature.com/articles/d41586-021-01394-0

Talking with friends who are virologists, they all seem to agree masks help.

What's wrong with the articles the nature commentary are based on And articles like that: https://journals.plos.org/plosone/article?id=10.1371/journal...

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7883189/#__ffn_...

[timr]

> Article 2 has a whole lot of pre-prints with data in the citations

This is yet another news article, written by a journalist (Lynne Peeples, who wrote the other Nature news article you've shared upthread) who is mostly re-hashing her previous article. Did you read it or follow the citations? This is an excellent example of hearsay: if you don't take the time to read the citations in detail, you won't realize what you're missing.

Most of the citations here are same citations in the previous Nature news article. Only citations 4-8 deal with the question of masks' impact on transmission (the remainder are for other questions, like whether vaccinated people can be infected). Of the papers in citations 4-8, two are new. Both are based on mathematical models using self-reported survey data and/or state-level case data. I don't know how good these papers are, but all studies of this form are low-quality medical evidence.

Unfortunately, bad journalists can re-hash the same crappy studies faster than skilled people can debunk their work.

> Talking with friends who are virologists, they all seem to agree masks help.

That's meaningless. You can find a sample of scientists to support pretty much anything.

> What's wrong with... https://journals.plos.org/plosone/article?id=10.1371/journal...

So many things. The paper uses survey responses (= bad data), and tries to correlate it with state-level infection curves (= blurry data), and only does it for a month-by-month difference for every month between April and September 2020, when cases were falling across all of north america. It sets arbitrary thresholds for "high cases" and "high mask wearing". It doesn't control for confounding factors, like other policies the states in question might have enacted.

This is a terrible paper. I cannot believe PLOS One published this. It is embarrassing.

> https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7883189/#__ffn_...

This is not a paper, it's a review. It presents no data. It summarizes the authors' opinions on other papers.

[kgarten]

So I don't know what you are arguing: We have a lot of evidence on the effectiveness of masks in relationship to preventing droplet infection.

We don't have so much about COVID 19. On the community level it looks promising for me, the papers are flawed but coming out.

What I am missing ... where is the other side of the coin. All models are wrong but some are useful. The particle simulations are wrong, but pretty useful (and I don't see why COVID should behave differently). What model do you base your assumption that masks don't work? Is there research that supports your claims? All science is bad science :) Some is useful.

> survey responses (= bad data)? So any research based on survey responses is bad? It's a standard method in policy research. What data would you use to show the community effect of mask wearing?

I'm usually listening critically to experts in a field (not to dentists or surgeons), it seems people who claim masks don't work are not virologists and have not worked with the virus (COVID19): https://en.wikipedia.org/wiki/Christian_Drosten https://en.wikipedia.org/wiki/Sandra_Ciesek

Both strong in the pro mask camp. Are there virologists working on COVID 19 (in the lab) that are on your side of the argument?