What new cause area should Open Philanthropy consider funding?
I’m not a frequenter of EA forms, but I thought I’d be fun to try and give what I thought of as the “obvious” answer. Probably not up to scruff, but I had fun speedwriting it and I’d be curious about the obvious and less obvious criticisms people have of this.
I think fundraising around the creation of DIY vaccine platforms has a lot of potential.
i—Delays in Vaccine Distribution
The fact that covid vaccines are made by targeting the spike protein is pretty standard intuition, this is obvious without hindsight by looking at early vaccine candidates against covid.
Similarly, with many bacteria, viruses, and parasites, there are pretty obvious heuristics for which surface proteins to make the target of a vaccine, and one can always mix a few.
The FDA and EMA approval process respectively have added a roughly 11 months, and scaling up production of the vaccines took yet more time.
This is not at all standard, approval agencies were especially lax with covid vaccines due to necessity; Normal vaccines can take decades from creation until approval. This is fine, except in cases where people that want to participate in trials of the vaccines based on their own CBA can’t, or in cases where the utility calculation would encourage one to get vaccinated almost no matter what (e.g. malaria), or in cases where we can theoretically assume the vaccine to be of very low impact (e.g. very small dose LDN mRNA vaccine against the spike protein of a non-sars-cov-2 cold-causing coronavirus) … which is to say, this is not-at-all fine.
To this, we can add delays in production, in part because of the aforementioned regulatory burden. Vaccines have to be centrally made, then shipped through medical couriers to centers of distribution where they might be administered or further spread out to more local points of innoculation. This step seems entirely replaceable, provided open source schematics, by something as simple as an mRNA printer and mixing the outputs thereof with some lipid particles in a vial.
This means that with a disease like a covid, there was roughly a year and two-thirds worth of wait between a working vaccine existing and most of the developed world’s population having access to it.
ii—Ease of Vaccine Creation
There are “generic” ways of making vaccines, with the latest one being mRNA inside LDN capsules. Given that this is now being prototyped (with the typical 4-10 year slog through clinical trials approach) against HSV1/2, malaria, dengue, several types of encephalitis, and others, I’d say we can call it fairly hit-all at this point. On top of that, it seems to principally work with IM, IN, and oral administration routes.
There are heuristics for selecting the protein target of vaccines, and while these might be difficult to understand for a layman, in most cases a prediction market of experts (or some simpler consensus mechanism) could arrive at an educated guess.
With COVID, I’m personally aware of two documented attempts at making a DIY vaccine [1][2]. Also, Preston Estep’s nasal vaccine (documented in [1]) was, I think, used by various others (including George Church, of course), but most were biotech researchers, so had some amount of manufacturing expertise.
Still, it seems like vaccines could be quite trivial to both design and produce for many diseases. Which is to say that a reasonably-apt Ph.D. in immunology with plenty of labwork experience could knock out a candidate in a few weeks or months, and a reasonably apt human that took a few labs in college and has access to UPS delivery could replicate said vaccine in a few days or weeks.
iii—Vaccine Blueprint Platform
Thus my proposal would be for a platform that provides blueprints for making vaccines and collects data on their efficacy.
More specifically the platform would have to provide:
Instructions for producing vaccine candidates against commonly circulating diseases
Information about commonly circulating diseases in a given area (local institutes usually produce these from air and waste-water)
Information about where to order cheap reagents or equipment for producing many vaccine candidates.
My bet if such a thing were created, is that the vaccine candidates are likely to all be mRNA based and the recommendations will be to buy an mRNA printer and a few types of “stock” LDNs to store in your fridge.
The vaccine candidates themselves would be selected by independent experts for expected efficacy and ease of use and classified in terms of cost and risk. The selection process would also favor vaccines that are cheap to make or can be made with the same equipment used for making most others.
Besides the creation instructions, vaccine candidate proposals would also contain instructions about how to check for immunity (e.g. via specific ELISA tests for whatever antibodies the vaccine is hoping to generate, profile changes on 2 or 3 flow cytometry-based analyses of immune cells in the blood, or other readily available lab or home tests) and which, if any, specific adverse reactions to look out for.
Data would be gathered from users about the vaccine candidates, which would give some idea of efficacy as well as safety, thus giving the option for people to make the jump and try a vaccine whenever they feel comfortable with the statistics around it (these can be computed on-the-fly with every new data point being added).
Making and maintaining the platform itself ought to be pretty cheap, almost negligible, my personal expert guess would be 40 to 200k to build, 5-20k yearly to maintain without expansion. The cost of incentivizing researchers to use it and post ideas seems more complex to estimate, as is the cost of those researchers donating their time to this and not working on something else. Finally, there’s the cost of potentially supplying people directly with the reagents and equipment needed to make vaccines, but this step is optional, as the project would originally rely on people to volunteer their own money to buy the necessary equipment for benefits to themselves and their circle of care.
iv—DALY estimates
Estimating DALY loss from diseases seems very hard, especially since we’d be up against the hard task of estimating DALY gain conditional on the thymic capacity for individuals vaccinated against all of these diseases.
So instead think of it from the perspective of:
Disease is the terminal cause of death in most cases, and immunity is the only way to fight against the disease at a zero-cost way. Malaria nets might be more cost-effective than a Malaria vaccine, but they don’t lead to a Malaria free society, and the benefits of that are higher than just the DALYs saved. Extrapolate to every other neglected disease.
We usually ignore deaths that are triggered by the “common cold” or damage from viruses such as HSV and HPV because they are too fuzzy to understand. But assuming some damage does exist, and assuming there are thousands of such sub-clinical viruses, hundreds of which an individual might encounter during their life and vaccinate against, we’d be defending against quite a lot of unknown unknowns (e.g. multiple sclerosis, or CFS/ME)
Such a platform would generate a lot of value in terms of knowledge by simply having individuals exchange private know-how around vaccine creation and data around vaccine response for altruistic reasons.
If you really want a DALY estimate you could do something like taking <nr you believe for covid global DALY> multiply by 1/5th to get average cost from new season colds, factors in DALY you think is lost to parasites and oral bacteria, factor in DALY from neglected or subclinical viral diseases, divide by world population, divide each disease class by assumed vaccine efficacy, multiply by assumed uptake. I got numbers ranging from 3,200DALYs/year to 41,000,000DALYs/year depending on what assumptions I chose to make, so I gave up writing the analysis since it’s too fuzzy.
v—Issues
I see a few large issues around this proposal which I’ve not touched:
It might be that all diseases against which a healthy immune system cannot react properly (e.g. malaria, yellow fever, CMV) are either inconsequential, have readily available vaccines, or can be eliminated more easily via reasonable levels of sanitation
Any such platform for vaccine creation would probably raise some political ire. But given the decoupled nature of the platform from the actual process, it should be sufficient to entrust its hosting either to an anonymous distributed hosting service or to some guy living in the Cayman islands to avoid all legal responsibility.
The target audience for the platform will initially be between the hundreds and dozens of thousands. This reduces the scope of DALY saved quite a lot. But provided a new pandemic arises, which seems almost certain, people might gain trust in the platform by being provided with timely vaccines through individuals in their network using it. One could also see “evangelists” using this platform in their own communities to provide vaccines for high-risk diseases to anyone.
The most suitable type of vaccines to distribute this way is intranasal vaccines (perk: mucosal immunity) and oral vaccines (perk: better efficacy against many bacteria and almost all parasites). The most generic vaccine technology right now (NLP & mRNA) is still not used in any widespread oral or mucosal vaccines. But in principle this is doable and there’s late-stage active research in both intranasal and oral.
All of these issues seem principally addressable (2, 3, 4) or easy to investigate (1).
A platform for on-demand vaccination
Open phil aksed:
I’m not a frequenter of EA forms, but I thought I’d be fun to try and give what I thought of as the “obvious” answer. Probably not up to scruff, but I had fun speedwriting it and I’d be curious about the obvious and less obvious criticisms people have of this.
I think fundraising around the creation of DIY vaccine platforms has a lot of potential.
i—Delays in Vaccine Distribution
The fact that covid vaccines are made by targeting the spike protein is pretty standard intuition, this is obvious without hindsight by looking at early vaccine candidates against covid.
Similarly, with many bacteria, viruses, and parasites, there are pretty obvious heuristics for which surface proteins to make the target of a vaccine, and one can always mix a few.
The FDA and EMA approval process respectively have added a roughly 11 months, and scaling up production of the vaccines took yet more time.
This is not at all standard, approval agencies were especially lax with covid vaccines due to necessity; Normal vaccines can take decades from creation until approval. This is fine, except in cases where people that want to participate in trials of the vaccines based on their own CBA can’t, or in cases where the utility calculation would encourage one to get vaccinated almost no matter what (e.g. malaria), or in cases where we can theoretically assume the vaccine to be of very low impact (e.g. very small dose LDN mRNA vaccine against the spike protein of a non-sars-cov-2 cold-causing coronavirus) … which is to say, this is not-at-all fine.
To this, we can add delays in production, in part because of the aforementioned regulatory burden. Vaccines have to be centrally made, then shipped through medical couriers to centers of distribution where they might be administered or further spread out to more local points of innoculation. This step seems entirely replaceable, provided open source schematics, by something as simple as an mRNA printer and mixing the outputs thereof with some lipid particles in a vial.
This means that with a disease like a covid, there was roughly a year and two-thirds worth of wait between a working vaccine existing and most of the developed world’s population having access to it.
ii—Ease of Vaccine Creation
There are “generic” ways of making vaccines, with the latest one being mRNA inside LDN capsules. Given that this is now being prototyped (with the typical 4-10 year slog through clinical trials approach) against HSV1/2, malaria, dengue, several types of encephalitis, and others, I’d say we can call it fairly hit-all at this point. On top of that, it seems to principally work with IM, IN, and oral administration routes.
There are heuristics for selecting the protein target of vaccines, and while these might be difficult to understand for a layman, in most cases a prediction market of experts (or some simpler consensus mechanism) could arrive at an educated guess.
With COVID, I’m personally aware of two documented attempts at making a DIY vaccine [1] [2]. Also, Preston Estep’s nasal vaccine (documented in [1]) was, I think, used by various others (including George Church, of course), but most were biotech researchers, so had some amount of manufacturing expertise.
Still, it seems like vaccines could be quite trivial to both design and produce for many diseases. Which is to say that a reasonably-apt Ph.D. in immunology with plenty of labwork experience could knock out a candidate in a few weeks or months, and a reasonably apt human that took a few labs in college and has access to UPS delivery could replicate said vaccine in a few days or weeks.
iii—Vaccine Blueprint Platform
Thus my proposal would be for a platform that provides blueprints for making vaccines and collects data on their efficacy.
More specifically the platform would have to provide:
Instructions for producing vaccine candidates against commonly circulating diseases
Information about commonly circulating diseases in a given area (local institutes usually produce these from air and waste-water)
Information about where to order cheap reagents or equipment for producing many vaccine candidates.
My bet if such a thing were created, is that the vaccine candidates are likely to all be mRNA based and the recommendations will be to buy an mRNA printer and a few types of “stock” LDNs to store in your fridge.
The vaccine candidates themselves would be selected by independent experts for expected efficacy and ease of use and classified in terms of cost and risk. The selection process would also favor vaccines that are cheap to make or can be made with the same equipment used for making most others.
Besides the creation instructions, vaccine candidate proposals would also contain instructions about how to check for immunity (e.g. via specific ELISA tests for whatever antibodies the vaccine is hoping to generate, profile changes on 2 or 3 flow cytometry-based analyses of immune cells in the blood, or other readily available lab or home tests) and which, if any, specific adverse reactions to look out for.
Data would be gathered from users about the vaccine candidates, which would give some idea of efficacy as well as safety, thus giving the option for people to make the jump and try a vaccine whenever they feel comfortable with the statistics around it (these can be computed on-the-fly with every new data point being added).
Making and maintaining the platform itself ought to be pretty cheap, almost negligible, my personal expert guess would be 40 to 200k to build, 5-20k yearly to maintain without expansion. The cost of incentivizing researchers to use it and post ideas seems more complex to estimate, as is the cost of those researchers donating their time to this and not working on something else. Finally, there’s the cost of potentially supplying people directly with the reagents and equipment needed to make vaccines, but this step is optional, as the project would originally rely on people to volunteer their own money to buy the necessary equipment for benefits to themselves and their circle of care.
iv—DALY estimates
Estimating DALY loss from diseases seems very hard, especially since we’d be up against the hard task of estimating DALY gain conditional on the thymic capacity for individuals vaccinated against all of these diseases.
So instead think of it from the perspective of:
Disease is the terminal cause of death in most cases, and immunity is the only way to fight against the disease at a zero-cost way. Malaria nets might be more cost-effective than a Malaria vaccine, but they don’t lead to a Malaria free society, and the benefits of that are higher than just the DALYs saved. Extrapolate to every other neglected disease.
We usually ignore deaths that are triggered by the “common cold” or damage from viruses such as HSV and HPV because they are too fuzzy to understand. But assuming some damage does exist, and assuming there are thousands of such sub-clinical viruses, hundreds of which an individual might encounter during their life and vaccinate against, we’d be defending against quite a lot of unknown unknowns (e.g. multiple sclerosis, or CFS/ME)
Such a platform would generate a lot of value in terms of knowledge by simply having individuals exchange private know-how around vaccine creation and data around vaccine response for altruistic reasons.
If you really want a DALY estimate you could do something like taking <nr you believe for covid global DALY> multiply by 1/5th to get average cost from new season colds, factors in DALY you think is lost to parasites and oral bacteria, factor in DALY from neglected or subclinical viral diseases, divide by world population, divide each disease class by assumed vaccine efficacy, multiply by assumed uptake. I got numbers ranging from 3,200DALYs/year to 41,000,000DALYs/year depending on what assumptions I chose to make, so I gave up writing the analysis since it’s too fuzzy.
v—Issues
I see a few large issues around this proposal which I’ve not touched:
It might be that all diseases against which a healthy immune system cannot react properly (e.g. malaria, yellow fever, CMV) are either inconsequential, have readily available vaccines, or can be eliminated more easily via reasonable levels of sanitation
Any such platform for vaccine creation would probably raise some political ire. But given the decoupled nature of the platform from the actual process, it should be sufficient to entrust its hosting either to an anonymous distributed hosting service or to some guy living in the Cayman islands to avoid all legal responsibility.
The target audience for the platform will initially be between the hundreds and dozens of thousands. This reduces the scope of DALY saved quite a lot. But provided a new pandemic arises, which seems almost certain, people might gain trust in the platform by being provided with timely vaccines through individuals in their network using it. One could also see “evangelists” using this platform in their own communities to provide vaccines for high-risk diseases to anyone.
The most suitable type of vaccines to distribute this way is intranasal vaccines (perk: mucosal immunity) and oral vaccines (perk: better efficacy against many bacteria and almost all parasites). The most generic vaccine technology right now (NLP & mRNA) is still not used in any widespread oral or mucosal vaccines. But in principle this is doable and there’s late-stage active research in both intranasal and oral.
All of these issues seem principally addressable (2, 3, 4) or easy to investigate (1).