Just that increases in risk are usually approximated pretty well by a linear change with dose. Radiation, poison, allergens, etc. Drugs too – the ones where the toxic dose is close to the therapeutic dose are generally pretty nasty even when used as intended. It's possible for there to be a transition between "very safe" and "noticably unsafe" as the quantity of something increases twofold, but a priori unlikely. For this to happen not only in vaccines, but through an unknown mechanism that operates precisely in this particular range, would be truly rotten luck. It's also contradicted (a posteriori) by the evidence I referred to (overdoses and animal models).

This isn't right. Risk from overdoses of drugs, poison, and allergens is *highly* nonlinear. It'd be exceptionally unlikely in the case of mRNA vaccine, but that's the case the majority of the time.

There's a specific minimum dosage most people have of allergen before they hit on a life-threatening reaction. You might be okay with trace amount of peanuts, swell up with one peanut, and die with five. It's pretty consistent, and very nonlinear.

That's why you can build up a tolerance to iocane powder too; small doses won't do anything, while large ones are guaranteed to kill. If it was linear, a 1/10th lethal dose would have a 10% chance of killing you.

I mean, they're all nonlinear at some point in the range of possible doses. Radiation too. I should have specified I was focusing on the "very low" bit of the curves. Chest X-ray vs two chest X-rays, rather than Louis Slotin vs the guy at the back of the room.

Nope. Most medicines are totally nonlinear. Almost any function is affine if you look close enough, but that's different from linear. If you half the dosage of ibuprofen, your odds of complications rarely fall by 50%.

Chest x-ray is pretty linear within reasonable dosage ranges, for much the same reason I'd expect mRNA damage to be.

Let's say a piece of mRNA has a 1E-12 chance of causing you to grow a third arm by virtue of a defect. Two pieces of mRNA would have very close to a 2E-12 chance of causing you to grow a third arm. That's linearity. If you've doubled the defect rate, you've doubled the risk.

This contrast with e.g. sports brain trauma, where one sports brain trauma might be relatively moderate risk, but two might be very high risk (much more than double). Or risk from ibuprofen, where 600mg is has close to zero risk, but an overdose of 60,000mg poses a significant risk of death (much more than 100x the risk of death of a 600mg dose). That's nonlinear.

I'm not sure how risk from mRNA defects would be anything but linear. It would take an extraordinary set of coincidences.

it would appear you are sampling an unknown distribution, finding it mostly to be negligible in its effects, and calling it linear?

I’m confused so let me try an analogy with DNA: most mutations do nothing. Every now and then, cancer. Sample a large number of mutations for a short duration, and you have proven nothing regarding nonlinearity globally. Roughly speaking.

Another example: I can measure a the air near (but not to near) a shockwave, or just under (By human scale) the free surface of the ocean, and find pretty much linear behavior if no other pathologies are present. The underlying physics however, is not linear, it is only my approximation which is. Maybe it’s valid over configurations of interest. But then I better know something about the limits of my approximation.

The thing about DNA damage is that it replicates. mRNA is pretty much designed to be degraded almost immediately after it is transcribed.

If I understand woofie11 correctly, it means that the probability that an error in a single mRNA corresponds with an error in another mRNA to cause a larger perceivable effect is small. The errors would have to be made in the exact same way.

Ah okay. Thanks for that insight! That seems key.

Or alternatively, the odds of a particular mRNA mutation doing something harmful are astronomically low. With 7 billion people being injected with large amounts of mRNA, you have a very small number times a very big number, so I can't estimate how likely something is to happen, but we know it's small since we haven't learned about anything critical with tens (hundreds) of millions of injections so far.

If an mRNA mutation were to make a self-replicating virus, or make something which damages DNA leading to cancer, or something wonky like that, I guess that's in abstract possible. It can make abstract proteins, and we don't know what those will do.

In that case, if I have twice as much mutated mRNA in my body, the odds of that happening almost exactly double. It's like rolling a 1-trillion sided die twice instead of once and seeing if I roll a 1. With one roll, I have a 1-in-a-trillion chance. With two rolls, I have a 2-in-a-trillion chance (minus 1-in-a-septillion of having the same thing happen twice, which we don't know what it would do, but is a small enough possibility we can ignore).

But yeah, unless something really wonky going on, I'm not going to end up with hundreds of strands of mutated mRNA in my body *all doing the same thing*.

The large sample size in cross section is not compelling to me - On its on - over a short duration of time. Because I know nothing about long term possibilities from that.

The statistics as derived from the dynamics of “how rna works” - yeah that’s compelling.

This:

“If an mRNA mutation were to make a self-replicating virus, or make something which damages DNA leading to cancer, or something wonky like that, I guess that's in abstract possible. It can make abstract proteins, and we don't know what those will do.”

Sounds astronomically I likely, but with unknown error bars. Sounds like our If our cross section test Is large enough going on long enough, we might just get there.

Given that the mRNA in the vaccine encodes for just a single protein of a given virus, it seems totally impossible to get an entire virus out of an error. :)

A prion (misfolded protein that causes other proteins to misfold) is more likely, but even then prions need to misfold in specific ways that encourage their replication as well. In all likelihood the error is just going to prevent the ribosome from finishing transcription and we'll have just a fragment of whatever protein the RNA initially encoded for.

Thanks! This is good to know as well. And thanks for pausing to help someone who honestly has no spare time whatsoever to dig in. I know that is probably frustrating for some (not necessarily you, it’s just I’ve been there and it can be exasperating)

Is there any reason to think it's unlikely that mRNA defects would have the same risk profile as Ibuprofen? The body is full of thresholding processes, where up to X% blood concentration the organs filter things out fine, and there's no issue, but over that level danger escalates quickly. I don't know immunology basically at all, but is there some reason to think it doesn't have processes that behave like this?

Yes; they're vastly different mechanisms.

Ibuprofen reaches a threshold of toxicity; a level at which our body's ability to manage it gets met and then exceeded. Things go rapidly wrong at that point.

Degraded mRNA just... doesn't work. It'd be a problem if all the mRNA in your cells started suddenly degrading faster than usual, for sure, but that's because you need the resulting protein output to live. If the vaccine degrades too fast, it just doesn't produce the proteins it was supposed to, but we didn't need them to function. Lost efficacy, but you don't wind up growing extra arms.