If anyone had been in the mood to argue with my previous missives about getting a vaccine out quickly, they could have easily found an objection. “You’re only addressing testing”, they might have said, “and the real problem is manufacturing. It’s fine to test a vaccine fast, but the real struggle is going to be in the process of gearing up to manufacture hundreds of millions of doses. Even most of the regulatory work deals with details of manufacturing. And since we can do testing in parallel with gearing up for manufacturing, and the gearing up for manufacturing will take longer, we don’t need to do testing as fast as you suggest.”

That wouldn’t have been 100% right; at least it was widely reported that fielding of vaccines was slowed down due to waiting for trials to complete. But it still would have been very largely correct. There’s a whole world out there of ingredients to make vaccines, sub-ingredients to make those ingredients, and companies each of which can supply one or more of those ingredients or sub-ingredients at some quantity and some quality level. Some of that is acknowledged officially (as in companies that have all the paperwork in place to prove that they make ingredients to pharmaceutical standards), and some exists unofficially. There’s also the equipment those companies use to manipulate and to test those ingredients, and another corporate/regulatory ecosystem that provides it. I probably have a better idea of what that world contains than 99.9% of the population does, but in absolute terms that just gets me to the point of knowing how much I don’t know about it. Indeed, though the people who did the hard work of getting the vaccine manufacturing done must have known the details of the industry much better than I do, there must have been plenty they learned on the job by scrambling for suppliers to deliver a key ingredient or device. It’s not the sort of environment where companies list prices and quantities on their websites; it’s the “contact us for more information” sort of environment, where until you get people’s full attention and interest and engage in prolonged negotiation you never quite know what they’re capable of – and they may be fooling you even then.

But although the “manufacturing is the main problem” objection would have been mostly valid at the time, it’s not true any more. The machinery to produce mRNA vaccines is now up and running and producing hundreds of millions of doses. That machinery is much easier to switch to making a new vaccine than is machinery for older methods of vaccine production: the processes of making and purifying mRNA aren’t entirely independent of the actual sequence the mRNA codes for, but they’re much less dependent than in other areas. When purifying a protein, for instance, different amino acids have different shapes and different binding affinities, so the process takes considerable tailoring, whereas mRNA is made up of units which are very close to the same shape as each other. So we actually are now in a world where the limiting factor in getting a new vaccine out is testing.

Even aside from that, faster testing would make manufacturing easier, by making it feasible to improve the manufacturing process. In the current environment, once the manufacturing process for “biologic” drugs is approved, the company has to keep running that process as described in the approval documents or else go through the entire approval exercise again. (The pseudonymous commenter “Mammalian scale-up person” in the comments on this article by Derek Lowe writes of people who “misunderstood the requirements and put in the number of doors and windows in the building”, and then “were unable to transfer manufacturing to another site”.)

Those sorts of rules only apply to “biologics”; the rules for “small molecules” (where the drug is a single molecule whose structure is precisely defined) are simpler. Even for biologics, though, that’s the sort of regulatory regime that would be appropriate if this were all voodoo: if no intermediate stages in manufacturing could be meaningfully tested and the consequences of any change could never be forseen. Really these medicines could never have been developed in the first place if they were that dimly understood. But to the average voter they are that dimly understood, so it’s no surprise that we have the present regulatory regime. And even to experts there are still occasional voodoo aspects of the exercise: places where some step of the manufacturing process is broken by changes that “shouldn’t make a difference”, and broken in a way that isn’t easily tested for. So often it really is appropriate to re-run trials on humans.

For that, quick trials would be a benefit. Indeed, even when the current standards say it is okay to open a new, larger factory to scale up production: are you really sure all the new machinery works the same as the old? It fits the description you wrote down and the government approved, but is it really the same? Maybe it would be best to test the output of the new factory against the old? And maybe not just in lab tests but in humans, if that can be done in weeks rather than years?

Or there may be a good reason to change the production process: scaling up often means a shift in what sort of equipment is optimal for production. There were stories of the mRNA vaccine production being scaled up by purchasing a huge number of small mixing machines to produce lipid nanoparticles rather than by trying to use a larger mixer; the latter could have been more economical, but the former was what was tested and approved. Quicker testing would lessen the extent to which people have to do this sort of thing.

Of course, all this sounds like dimly grappling with a process that is sophisticated beyond any one person’s understanding – and if we are to take full advantage of modern technology, that is perhaps inevitable: no one person understands every detail of it. People are aware of what specialists in related areas do, and understand most of it, but not so much as to actually be able to step into their shoes and do their jobs.

But making a vaccine doesn’t actually have to be that complicated. Pasteur was making vaccines in the late 1800s with laboratory equipment that today is on the high school level, and getting them out rather quickly. Developing a vaccine can be quite hard when going up against a disease that the body does not naturally cure itself of, like AIDS or tuberculosis; they first started trying to make a tuberculosis vaccine in Pasteur’s era, and still don’t have a good one. (There is a widely used tuberculosis vaccine, BCG, but its protective effect is weak enough that some studies show effectiveness and others don’t. Its risks are also not what one would want: it’s a live vaccine, a weakened form of the tuberculosis germ, yet not weakened enough to be thoroughly safe; it commonly establishes itself as a persistent infection and in rare cases can even kill.)

But Covid is the easy category; the body normally purges itself of the virus, after which reinfections are mild. Indeed, the Sinopharm vaccine (a preparation of killed Covid), the main Chinese offering, is not far from something Pasteur might have produced. In the West, people who were trying to make a vaccine powerful enough even for AIDS turned with relief to a much easier problem, and seem to have produced superior Covid vaccines.

As for what might be done today by people who had something of Pasteur’s daring and ability, the RaDVaC project is worth mentioning. Their white paper is a masterclass in modern vaccine design. There’s no big money behind the project, but quite a lot of skill; it’s a labor of love (and fear; they’ve been testing it on themselves). It hasn’t gone through any formal double-blind testing (that would need big money), but it might be the best vaccine out there. Not being constrained like the big companies are by the need to stick to what has been formally approved, they’re now on the eleventh iterative improvement. Their vaccine is administered by inhalation into the nose, which might be what we need to counter transmission of the latest variants even by the vaccinated: when the body encounters a virus (or a vaccine) in the nose, it produces “mucosal immunity” there, in the form of secreted IgA antibodies and resident T cells. There’s a good chance that this would stop infections before they grew to a level where they spread to other people.

Still, RaDVaC doesn’t really address mass manufacturing and distribution. They write of “open source”; but where open source has really succeeded is in software, where mass manufacturing and distribution are easy. There doesn’t seem to be any huge difficulty in manufacturing or distributing their vaccine; they’ve chosen relatively simple ingredients which can be (and have been) sourced by amateurs. The most complicated ingredient is short peptides from the virus, which are ordered via custom synthesis from companies serving the biological research community. Still, actually having to deliver something in large quantities always involves more difficulties than can be anticipated. Now, the quantities here aren’t large by ordinary standards; indeed, they’re near-invisible: “50 to 100 micrograms total peptide per dose” of vaccine. And safety standards aren’t as stringent for something that’s snorted as for something that’s injected. But putting together and distributing doses for millions of people would still take a lot of doing.

And although they like to cite Linux as an example of successful open source, they don’t call each others’ ideas “idiotic” in public like Linus Torvalds does. Now, he’s more diplomatic than many give him credit for: he normally tells people that their idea is idiotic, not that they themselves are idiots. (The difference might seem subtle, but really it’s huge: even geniuses have the occasional brain fart.) Even so, the only way he gets away with it is that the things he calls idiotic really are idiotic; it’s not a style that can be safely recommended to others. RaDVaC is more like a traditional academic collaboration: internal discussions followed by publication of a consensus document – which is fine, but has neither the risks nor the benefits of “management by perkele”.

Another thing about quick vaccine trials: besides running them in the manner described in my previous article (an escalating exponential trial), it would be best if they were done in a very open way – that is, with a website where anyone could look and check the progress: not just how far the trial had gone and how many people have been enrolled but also the current outcomes, in full detail: how many people from each group had gotten the disease, been hospitalized for it, or died from it, and how many had reported side effects (and what those side effects had been). This would be kept updated in real time: people could follow it like a horse race and even bet on it. (Indeed, the website probably could pay for itself by selling fast data feeds to Wall Street traders so that they could instantly update their pricing models – models not just of the vaccine companies’ prospects but also of economic prospects in general. Wall Street traders can and do scrape public websites to gather data, but they prefer fast direct data feeds and are willing to pay.)

Making results public immediately might lead to concerns over breaking blinding, but it doesn’t seem like those concerns would be well grounded. There still would be much the same sorts of mechanisms in place to validate reports as there are today: it’s just that when they hit the tabulating computer (the first entity in the chain that would know who got vaccine and who got placebo) that computer would update the website accordingly. A doctor who reported a case could in theory check the website just before and just after reporting, to see which column his patient ended up in, but it’s unlikely a doctor would go to that trouble for something that wouldn’t affect the care provided. Besides, the blinding in the coronavirus vaccine trials was never really thorough: injection-site reactions already gave the trial subjects a good idea whether they got the real vaccine or the placebo.

Making the trials open would eliminate the possibility of politicized data-hiding stunts (such as did in fact occur). And yes, it would give ammunition for the vaccine debate – but to both sides. I can already see the arguments: “the first reports were trending negative, and then the efficacy rate went up: they must have jiggered the results” – and in response, “that’s just random chance”. Well, maybe it is and maybe it isn’t: statistics has hard rules for deciding such questions. The rules are not even that difficult to defend in public debate. It’s common for our cloistered intellectual class to cringe from such debates, and deliver only an ex cathedra pronouncement on what people should do, but that strategy isn’t working well: they’ve screwed up too many pronouncements – on masks (first pooh-poohed, then mandated); on vaccines (first dismissed as something that would take years, then developed at “warp speed” and also mandated); on testing (botched in a number of ways). There’s no wonder people no longer trust them and are ready to believe conspiracy theories. Openness is a way to restore trust. Even if one cares little about whether public health authorities are trusted or not, in a pandemic there is a real need for people outside the public health profession to know how things are going – not just Wall Street traders but others who make business plans (or for that matter any plans).

Speaking of trust, one of my previous statements is no longer operative: now I have seen anti-vaxxers talk of delayed harm from vaccines. One fear is of the mRNA vaccines altering people’s genes; on that front I don’t think I can do better than to link to Linus’s rant, though for a deeper debunking of one such line of suspicions there’s this paper (which refers to the virus rather than the vaccine, but the vaccine introduces less RNA than the virus does).

Another fear is of antibody-dependent enhancement that only becomes apparent as antibodies dwindle; this seems to have happened in the Dengvaxia trials, where it showed up mainly in the third year of followup. The mechanism seems to be that as antibody levels wane, dengue virions which previously would have been swamped by antibodies and neutralized later just have one or two antibodies glom onto them, and use those to infect macrophages. But this mechanism has been tested with Covid, using a very wide range of concentrations of antibodies, and the macrophages didn’t get infected. (They also tested SARS and MERS; and MERS, a much less transmissible and much more lethal virus, did infect macrophages; SARS didn’t. Other papers, though, do report positive results for ADE with SARS, though I find some of them frustratingly unclear about what that meant. When a paper reports that a method was “as described previously (citation)”, one reads that citation and finds it also says “as described previously”, and one follows the chain four-deep without ever getting to a real description, and when it’s a bit of a tree rather than merely a chain since some of those citations are to multiple papers, and when some of those add language like “with minor modifications” – well, one has to conclude that the authors aren’t actually serious about describing their work. And in this case it matters: macrophages take in foreign particles as part of their normal job, so how exactly the study determined that they were being infected rather than operating normally is key. There are also intermediate possibilities such as an infection that ruins the macrophage but does not lead to it producing more viruses. The study I linked with approval above did actually measure productive infection, and said so clearly.)

In general, though, the line “don’t make antibodies; the virus will just use them against you” is on the same level as “don’t buy a gun for self-defense; the criminal will just take it away from you and use it against you”. Yes, such things have happened, but it’s very far from being the usual course of events. Normally guns work against criminals; normally antibodies work against infections; it’s the rare case where they’re worse than useless – though neither possibility is to be entirely ignored as a danger. (One certainly shouldn’t shove the gun up into the criminal’s face the way Hollywood loves to show; that’s just asking for it to be taken away. Likewise, if dealing with a known trickster like dengue, one needs to be very much on guard.)

Now, if one isn’t fixated on the dangers of the vaccine as a vaccine, one can come up with much scarier scenarios. Greg Cochran, five years ago, mused about the episode where poliovirus vaccines were grown in monkey cell culture and thereby contaminated with the monkey virus SV-40, and asked: what if the contaminant had instead been HIV? The deaths could have been in the millions. Indeed, the recent talk of HIV due to the coronavirus vaccination campaign in Brazil shouldn’t be entirely dismissed: perhaps there has been needle reuse. Even contamination during manufacture is not impossible, though improbable for most of the Covid vaccines: vaccines that are not live vaccines have much less chance of having a live contaminant. And whether the contamination be live or dead, we have much better methods to test for it these days. But though improbable, it’s not impossible; contamination is the sort of problem that requires continual vigilance.