@CamiHarland I'm very excited about this idea and have quite a bit of experience in closely related fields, I'd love to connect for a chat.

@AlexDemarsh Hi Alex, thanks for getting in touch and for your interest in the project! I would love to connect for a chat - I'll send a message over on Linkedin. Best, Cami

This is not a bad project to the extent they are tracking outbreaks (overwhelmingly likely to be caused by another NIH-funded lab).

Actual “biological weapons” are still 1000x less likely than that, especially state-sponsored ones, and there remains no path for their findings to materially change things.

If anything, the fact that gain of function isn’t even on the table to be banned (because Covid was the best thing ever to happen to authoritarian regimes, from Canada to China, and the “vote harvesting” portion of the US regime) shifts the odds even more toward the next outbreak being from the same source as the last one.

Approve to the extent they seem to focus on illness tracking near BSL labs, as that is buried in there somewhere.

Hi, I’m the lead researcher on this project (naturally I will not be betting). Thank you for the feedback already given.

I’ve seen some comments that this project is difficult to assess, so I wanted to provide a better introduction to the methodology for anyone following. I also want to respond on the risk profile and the objection from one of the users below that it could be exposing information that should only be held by states.

Some organizations that use similar techniques for research and reporting are below:

Bellingcat https://www.bellingcat.com/

Forensic Architecture/Forensis https://forensic-architecture.org/

The Center for Spatial Technologies https://spatialtech.info/

Examples of open source data being applied to chemical weapons control can be found here:

https://forensic-architecture.org/investigation/the-shelling-of-khudair-warehouse-chemical-warfare-by-indirect-means

https://forensic-architecture.org/investigation/tear-gas-in-plaza-de-la-dignidad

Our plan is to establish a program carrying out similar work on biological weapons control. Of course, the verification and mapping of biological weapons will require development of new methodologies that are better adapted to monitoring bioweapons research and disease spread. We are proposing to begin with network analysis [https://towardsdatascience.com/network-analysis-d734cd7270f8] and the social media-based and geolocation methodologies in the application. However, satellite imagery can also be used in imaginative ways. This study by the London School of Hygiene and Tropical Medicine and the Satellite Applications Catapult used satellite imagery for population mortality estimation associated with COVID-19: [https://www.lshtm.ac.uk/newsevents/news/2020/satellite-images-suggest-deaths-more-doubled-aden-yemen-during-intense-covid].

In response to one of the commenters, this is why we suggest that finding reporting disparities related to state cover-ups or poor data collection practices is possible. All of this work has a wide range of potential applications by other researchers and reporters. During the development of this proposal I have been in contact with researchers on some of the projects listed above to discuss the potential and feasibility of this work.

Ethics

Major think tanks and universities have given ethical clearance for projects with similar implications, for example at the Stanford Cyber Policy Center [https://cyber.fsi.stanford.edu/]. Trust and safety is clearly something to be taken seriously and best practice is constantly being discussed between researchers in this field. The reason why we would not seek to be funded by government agencies is that our purpose is completely different and would require independence.

I would be interested to hear any other feedback/concerns

@CamiHarland Hi Cami, thanks so much for you post! You linked some fascinating research that I was completely unaware of. I've warmed up to this proposal a lot since I first discovered it as I've learned more about the space. I do still have a couple reservations that I'd appreciate if you could address:

1. On a more technical level, how confident are you that enough information exists in your sources to get good insights? I.e. for Twitter in Russia, is 10% usage enough to create a good network model? And how different are various labs in their satellite footprints and research outputs?

2. Can you illustrate your theory of change a bit further? Say you see abnormal activity from one of the BSL 4 facilities in China. In your best case scenario, what does your evidence look like, what actions by countries or international bodies are taken that they weren't willing to take before, and does the Chinese response look like?

@CamiHarland Hi Camille, I am applying for the MSc in Modelling for Global Health programme starting at Oxford in 2023. I think your project is really exciting and there will be much more focus on this in the next five years. You might want to check out the work of Dr Marc Choisy at Oxford, who is working on computational models of disease spread for measles, influenza and dengue. How dependent will your project be on the right programmers to develop the models? How sure are you that you can hire the right person if you get the funding?

@JenniferKeyes Hi Jennifer, Thanks so much for your interest in the project! Having others on the project with relevant experience will of course be very important for it to reach its full potential. As you say, this is a growing field and we hope to eventually bring in people with a variety of backgrounds, including computational or mathematical modelling, intelligence roles or non-proliferation studies. I will definitely take a look at the work you have referenced as well - that sounds great!

No one has made any semblance of a defense of this proposal.

“It’s maybe feasible” - we all agree that they might be able to say Country X is a bit more likely than Country Y to have bio labs.

No one will care, the “pseudo-intelligence” will go nowhere, and might help the countries better evade detection if they publish this openly.

Unclear how many of the defenders are either affiliated with the project or are pumping its stock—as absent some actual demand by the UN or any form of real-world organization, this is somewhat below a file-drawer project or summer project.

@Gigacasting Dude you're commenting on every single project on this site to try to win money and I don't think you've read any of them. Look at this feedback on your comments on the extinction shelters project by @Nuno Sempere:

"I would give essentially ~no credence to the comments by Gigacasting on this thread; they seem like the come from a system 1 that I don't particularly trust."

Your comments are thoughtless (“visual networks for Covid means this is a good idea”)

If you aren’t intelligent enough to analyze even the slightest detail about the projects—then while it’s tragic you don’t “trust me” you can be silent if you’re not able to rebut even a hint of any of the dozens of arguments made against the proposal—almost all of which mirror mine.

You realize you look like a small child “Nuno doesn’t like you and I don’t either!!”

I don’t care. Make a thoughtful argument.

Hint: you make a good cult follower but are incapable on independent thought.

The task is whether the projects are good not whether you think they’re so hot right now

“I know there is a lot of interest in the Biological Weapons Convention and open data methodologies from different corners of the Effective Altruism movement right now. I would be very surprised if this doesn't get some funding. “

@Gigacasting You’re being abusive to dozens of people on this forum and should really be banned. The point is to provide quality feedback, not just trolling.

The mechanism here seems extremely dubious, I don't think this should get funding unless the requester can provide some proof of concept.

They cite Chinese naval bases in Cambodia and North Korea's nuclear program as examples of satellite analysis being useful. Naval bases aren't themselves hard to find, they're massive fortified docks. The intel that the Ream Naval Base was for Chinese use was from traditional intel sources citing a secret deal in 2019 (https://www.wsj.com/articles/secret-deal-for-chinese-naval-outpost-in-cambodia-raises-u-s-fears-of-beijings-ambitions-11563732482?mod=e2tw). Nuclear testing sites in North Korea involve massive tunnels and retaining walls whose purpose is pretty obvious from the outside, and which normal news sources already report on (https://www.cnn.com/2022/04/07/politics/north-korea-nuclear-satellite-images/index.html). The only other example provided is a report from a Russian defector that Russia found a lab in China linked to a fever outbreak in the '80s. Aside from being dated and unverified, it doesn't appear that any action was taken as a result of that finding.

I can't imagine satellite images telling you anything other than that there's a bio lab somewhere. Bio labs doing illegal research should look identical to those doing legitimate research, and a lot of research is dual purpose to begin with. Labs aren't like huge naval bases or nuclear test sites whose purpose is obvious from the outside. Further, any finding of nearby disease from social media would be minimally useful circumstantial evidence, case in point COVID in Wuhan. Ultimately, we're already suspicious of other countries' BWC compliance. Even on the off-chance the researcher finds something suspect, I don't see how we end up anywhere different.

Based on the evidence presented in the request, the chance of finding anything here, not to mention it leading to any concrete action, seems extremely remote. There needs to be some additional evidence that this research is worth prioritizing at the expense of other projects.

@GregJustice The North Koreans also are well known to use fire/smoke to cover areas from satellites.

@GregJustice Good comment, but I think you're focusing on the low-hanging fruit for a negative eval of this proposal, since this issue with satellite imagery is already well-known and the authors say it's not their main focus:

"Although monitoring of biological research is more challenging, some identification of facilities is possible. Kanatjan Alibekov, defected former director of one of the Soviet germ-warfare programs, asserted that Soviet reconnaissance satellites identified biological weapons laboratories in China in the 1980s, in one case connecting them with hemorrhagic fever outbreaks in the area."  

The use of satellite imagery for visualization seems to be the bones of the project, which is something different.

"Satellite imagery will also be used in the project for visualization purposes. It is intended to make use of a tool called TimeMap, developed by the research group Forensic Architecture, University of London. This program, which maps particles of data while storing the open data source information (e.g. Twitter post) could be useful for verifying exchange of information on outbreaks of infectious diseases (CBM B). This analysis could reveal unusual patterns of disease symptoms reported by individuals, potentially circumventing state repression of case numbers or total cover-ups of outbreaks."

This has much wider applications to data visualization in epidemiology. There has been a lot of focus on visualizing transmission networks for COVID-19 over the last few years using network graphs for instance Luo, Ma et al in Nature (https://pubmed.ncbi.nlm.nih.gov/33883590/) and balloon charts (https://idpjournal.biomedcentral.com/articles/10.1186/s40249-021-00800-z). Interest in these types of techniques is high at the moment and proof of concept is already appearing in peer-reviewed journals. Because of dual-use factors in biosecurity preventative research has a double value for intended and unintended harms.

I know there is a lot of interest in the Biological Weapons Convention and open data methodologies from different corners of the Effective Altruism movement right now. I would be very surprised if this doesn't get some funding.

@felixeno1 The main recommendation for addressing biological risk made by Toby Ord in The Precipice, for example, was to bring the Biological Weapons Convention to parity with the Chemical Weapons Convention in funding and investigative powers, but unfortunately politics has prevented the latter for decades.

@felixeno1 That does assuage some of my concerns, selling some of my position. I am still left with the concern that the best case result would be "there's some sick people near this facility," and then you're just stuck with that. Visualizing spread is really cool, but the proof of concept I'm looking for is specifically that this can detect secret biolabs. You might be right that interest from the EA community will lead to funding, that does scare me.

What I can't get past is imagining what this would've looked like with the start of COVID. Imagine some individuals get sick in Wuhan and all of them post about it on Weibo and China doesn't censor it. You could create a list of people who got sick, you can see that they made social media posts in a city with a bio lab, but what do you do with that? It's not clear evidence of anything, and China blocked any further investigation. With COVID we had the type of evidence that this project would've provided and it was functionally a dead end.

Multiple people getting sick could be suspicious, but I don't see how it can be conclusive. I don't see how you get anything done as long as countries can simply stonewall due to lack of an enforceable inspection mechanism.

Seems like fairly decent value for money. At least offering original research rather than a duplicate ‘literature review’ by someone outside the field like a lot of projects on here

@gigacasting The benefit seems to be to have intelligence that is not tied to a single country’s intelligence agencies, like independent think tanks and universities vs government research. Also doubt that the CIA has the same benign intentions as effective altruism!

“analysis of commercial satellite imagery and other geospatial analysis, media and network analysis relating to scientific activity, including collecting data from social media, analyzing publication and download records of significance for biological weapons programs.”

🤔

sort of seems like a joke, watching Twitter to detect international treaty violations (and shouldn’t they be pitching the CIA, not freelancing this, who’s going to care about their “intel”)

@Gigacasting Satellite image analysis for Ukrainian groups has given rise to the do it yourself defense industrial complex (DIYDIC?)

Without a “go to market” (any evidence that an intelligence agency or decision maker would care about much less be responsive to their work) it’s just someone pursuing a rogue foreign policy.

Pitch In-Q-Tel or Palantir; otherwise this is, at best, a file-drawer report, or at worst, harms the world by letting crucial methods and signatures of activity be known to those seeking to evade them

There’s a reason “sources and methods” are kept close to the vest; If they are successful and publish this openly it’s probably going to be duplicative with true intelligence work and actively harmful; there’s just no case where this makes sense.

(Unless they are adopting a do it yourself drone strike on Chinese bio labs 🤔)

@Gigacasting It makes sense if you are in the defence analysis world. OSINT is the future and already part of the intelligence ecosystem. If you look at Ukraine OSINT has forced more accountability. In February the U.S. went public with its intel on Ukraine, partly because so much analysis was already out there. That ramped up scrutiny and publicity. Intel on Ukraine has been far more accurate than for other 21st Century wars, partly b/c of OSINT.