> they select for the spike protein of the original variant and ignore others
Repeated exposure causes you to get many different kinds of antibodies that react against many things like the original spike protein. There's ample research showing that a 2 shot (and 3 shot) series of mRNA vaccination creates a very broad response that cross-reacts to many spike protein variants.
Reaction against the spike protein is best, because this binding doesn't just activate the rest of the immune system, but it's neutralizing against the virus: it prevents its entry into cells.
Compare this figure:
https://cdn.jamanetwork.com/ama/content_public/journal/jama/...
Many convalescent (prior infected) individuals had very low neutralization activity against variants. All vaccinated individuals had good neutralization titers against all variants measured in this study, though they were still definitely better against A.1 than other types.
> whereas natural infection generates a broader response that will target a variety of proteins.
Natural infection will generate a variety of nucleocapsid antibodies, too (against non-spike portions of the virus).
This is a mixed bag: it means the immune system can be activated and it also means that the immune system will try and break down things that have antibodies bound elsewhere on the virus... but the spike antibodies around that could directly prevent infection of new cells are less effective against variants.
And as for immunity from infection, your body makes antibodies for the prefusion spike (the free floating form that infects cells), the post fusion spike (after it binds to ACE2), and the nucleocapsid. Since prefusion spikes are what bind to cells for entry, you’d want your immune system focused entirely on attacking that vector. China’s sinovac vaccine generates nucleocapsid antibodies and it’s failing in the face of omicron.
Against Delta, if you compare a vaccinated population with 2 doses of the mRNA vaccine against an equivalent unvaccinated population that ended up with 100 infections, with about 70 people having symptoms and 10 people with severe illness, you'd expect:
* 60 people who would have otherwise been infected to show no evidence of infection. That is, the vaccine is so strongly neutralizing that a large proportion of people that would otherwise be infected show no evidence of virtal replication at all.
* Another 30 people or so to end up infected, without symptoms. The virus replicated some in their cells, but the severity of the disease before an immune response was mounted was so low that it went unnoticed. Note that this number has not changed too much from the unvaccinated population.
* About 10 people with symptoms, of whom 1 has severe illness.
You could view this as mostly shifting severity "downwards" -- some people who would have been infected ending up not; some people who would have had symptoms instead just being infected; some people who would have had severe illness instead just having symptoms.
From a pure mechanical view: if an infected cell produces 1000 copies of the virus, ordinarily resulting in 50 other cells being infected... and antibodies from a vaccine glom up on spike proteins and reduce this 50 number to 10-- the virus still grows in your body, but the rest of your immune system has a whole lot longer to react before illness becomes severe. And this ignores other benefits from vaccination (earlier mobilization of immune system, T-cell mediated immunity, training of B cell lines to target spike proteins resulting in faster maturation of variant-specific antibodies, etc). So even if the antibody concentration and affinity isn't quite sufficient it's still a benefit.
The opposite is the case.
No vetting of that Israeli study, no peer review, no competent efforts to account for bias in the sample, and it just does not hold up to scrutiny when compared to other results.