1. In short, no.

This is going to take a detailed thread to unpack. https://twitter.com/mattwridley/status/1270010444279877633

2. First, let's get the blog post out of the way. Ridley is speculating that the Asiatic flu of 1889-90 was not an influenza pandemic at all, but rather the emergence of the now-mild OC43 coronavirus into humans. According to his story, it subsequently evolved lower virulence.

3. The basis is a single paragraph in a 2005 paper about the OC43 genome, quoted below. The main evidence seems to be chronology and symptoms. But phylogenetic dating is notoriously imprecise, and some serological evidence suggests that the 1889-90 pandemic was a flu.

4. Ridley goes on to speculate that perhaps if the Asiatic flu was OC43, it then rapidly evolved lower virulence. Maybe, he suggests, SARS-CoV-2 would do the same thing—if we just ended the lockdowns and let it spread through the population.

5. Even if OC43 was the Asiatic flu—a pretty huge leap—I wouldn't want to gamble that SARS-CoV-2 would follow its trajectory. The final sentence is striking. No speculation, no "maybe"—just the assertion that unchecked COVID-19 will become indistinguishable from every other cold.

6. We simply do not have the evidence to draw that conclusion with any kind of certainty. I can't think of a human viral pathogen that has evolved this level of reduced virulence in recent enough history that we were able to monitor it. In any case, it's clear that not all do.

7. Other endemic virus remain highly virulent. Influenza remains far worse than a common cold. Measles, even more so. Smallpox, worse yet.

When spinning these evolutionary tales, people often invoke myxoma virus, introduced into Australia in the 1950s to control wild rabbits.

8. Indeed, myxoma virus did evolve reduced lethality, from nearly 100% to less below 50%. At the same time, the rabbits also evolved increased immune resistance.

But here is the scary part: that trend turned around. Myxoma became more lethal again.
https://www.pnas.org/content/114/35/9397

9. Virulence evolution is not a simple, predictable, one-way trajectory.

10. Ridley is telling us an evolutionary story based in the theory of virulence evolution.

So what does that theory actually say about what we should expect SARS-CoV-2 to do over the next few years? Let's take a look.

11. The most basic idea behind the theory of virulence evolution is that viruses face tradeoffs between virulence (the harm they cause to the host) and transmissibility (how well they spread to other hosts.

12. If a strain is to transmit, it needs to reproduce within the host and facilitate its own spread. This is hard to do without harming the host somewhat. But if it is too virulent, it may kill the host too quickly and miss out on opportunities for transmission.

13. So if we just look this far, we might expect to see the evolution of intermediate levels of virulence.

But of course this only scratches the surface. There are a whole host other considerations and there have been literally hundreds of papers written about these.

14. These include the vertical vs. horizontal transmission, competition between co-infecting strains, effects of an immune system, heterogeneity in host condition or genetics, and the role of vector spread.

I sometimes spend 2 weeks on this in my course. It's not nearly enough.

15. I can't cover all that here but I do want to raise a few key points.

Let's start with time scale and mutation supply.

*Even if* evolutionary theory unambiguously predicted a major decrease in COVID virulence, we still ought to be asking how long it would take.

16. If we could expect it to happen over the course of a year or so, that could help should a vaccine fail to materialize.

If it would take 100 years that's little consolation to me—and while Viscount Ridley may not agree, I think we're going have other, bigger problems by then.

17. When a virus emerges into the human population from another species, we don't expect it to be perfectly adapted for transmission. It may be inefficient at replicating, or may colonize the wrong parts of the body. It may be too virulent, or not virulence enough.

18. That's SARSCoV2 as of last November. What happens next? Well, that depends on what mutations happen to arise and what those mutations do. In general, those that increase how well the virus transmits will also increase in frequency. (But we'll revisit this; it's also subtle.)

19. If the virus is originally more virulent than optimal, the first successful mutations could make it more virulent still, so long as they help it spread. Only once the virus gets "really good" at spreading in humans could we expect it to approach any sort of virulence optimum.

20. So early on, basically any old thing could happen to virulence, depending on what mutations arise first. In this short piece below, I briefly describe a lovely modeling paper by Jim Bull and Dieter Ebert, that explains how this works.

http://blogs.nature.com/journalclub/2008/05/carl_bergstrom.html

21. Given the limited genetic variation we are seeing in SARS-CoV-2 and some other factors I'm about to discuss, "early on" will be on the order of years or even decades.

So any predictions of the sort Ridley is making pertain to the 2030s or later better than to 2021.

22. As we think about timescales, the next thing to keep in mind is that we are currently in expansion phase of a pandemic to which comparably few people on earth are already immune.

Why does this matter?

23. It matters because of the nature of competition and the nature of exponential growth. Suppose that tomorrow a new mutation arises that decreases virulence but increases transmissibility—just the kind that Ridley is hoping for in his argument. What happens to it?

24. (Technical aside: if R0 isn't much bigger than 1, the new mutation probably goes extinct. We can approximate transmission as a Poisson branching process, and calculate exactly what that probability is. But let's ignore that, by supposing the mutation arises and persists.)

25. A new mutation is *very rare* because there are millions of active cases worldwide, and it is just one of them. If it transmits say 10% more effectively (a very big advantage by evolutionary standards) it would take hundreds of generations—multiple years—to become common.

26. Taking stock, the theory is telling us that (1) the first successful mutations may not move us toward reduced virulence and (2) even if they do, it'll take years to get there.

Now let's look at how natural selection operates on virulence in a pathogen like SARS-CoV-2.

27. There are really a couple of key observations to make about the virulence of SARS-CoV-2 in this respect.

First, note that disease severity varies widely among patients. Most infected people do not suffer severe disease. Approximately 5% are hospitalized. 0.5-1.0% die.

28. Second, when death occurs, it typically occurs long after the usual window of disease transmission, a couple of weeks or more after the onset of symptoms.

Why do these thing matter?

29. The really bad things that COVID-19 does to people happen (1) only in a small subset of people and (2) after most or all of the transmission has already taken place.

30. That means that evolutionary changes in virulence will not necessarily not change transmission much, unless they change other things by coincidence as well.

And *that* means that selection on COVID-19 virulence is likely very weak.

31. Another way to think about it is that the bad things that COVID does occur after transmission and only to a small fraction of people, so natural selection can't really "see" those things or operate effectively on them.

32. (There's a non-trivial analogy to the evolutionary theory of aging as developed by Peter Medawar and George Williams. But that's a topic for another thread. If you can't wait: https://www.nature.com/scitable/knowledge/library/the-evolution-of-aging-23651151/#:~:text=Medawar%20and%20George%20C.,mathematically%20formalized%20by%20William%20D.)

33. This brings us to an important paper by Jim Bull and Bruce Levin (who turned 80 yesterday; Happy Birthday, Bruce!)

Unfortunately, it's behind an Elsevier paywall. I apologize.
https://www.cell.com/trends/microbiology/pdf/0966-842X(94)90538-X.pdf

34. The paper proposes that for many diseases, the harm that is done to the host is essentially disconnected from any benefit to the pathogen.

Virulence is not a matter of evolutionary fine tuning, it is, as Bruce is fond of saying, simply a matter of "shit happens."

35. They provide a number of really compelling examples.

The bacterial species commonly responsible for meningitis—H. influenzae, N. meningitidis, and S. pneumoniae—are transmitted by respiratory droplets and usually causes little or no pathology.

36. But sometimes they get into the cerebrospinal fluid, and cause terrible pathology. They don't transmit from there—the CSF is a transmission dead end for them—but in a small fraction of patients, they get there. For meningitis, virulence is wholly uncoupled from transmission.

37. Poliovirus is an RNA virus that is transmitted by an oral-fecal route. Along that route, it causes no pathology. But sometimes it gets into the CNS. There, it causes severe pathology—but again that's a dead end for transmission.

38. For somewhat more complicated reasons, AIDS seems very unlikely to be an adaptation to further the transmission of HIV. Levin and Bull go into this in detail in their paper as well.

And so on, and so forth.

39. (The original Levin and Bull paper focuses on within-host evolution as a driver of shit-happens virulence, but in my opinion this is largely unnecessary. Within-host ecology, e.g. sites of colonization/reproduction, goes a long way by itself.)

40. This is more hypothetical: COVID pathogenesis fits fairly well with many aspects of the Levin and Bull model. The majority of individuals have mild disease, but some are unlucky. Transmission from deep in the lung, where the greatest damage is done, is unlikely.

41. To wrap up, we've seen that

1) Empirically, we don't commonly see the patterns that Ridley is hoping for, and we do see that virulence can do all sorts of things, including make a u-turn in those poor rabbits.

42.

2) Even if COVID did evolve decreased virulence, it would not likely do it on a timescale that would help us much.

3) Selection of COVID virulence is likely very weak.

4) Empirically, many diseases have virulence uncoupled from transmission entirely.

43. Ridley's hope that virulence will decline due to selection (and his fear that we are slowing the process by not allowing the disease to rip through our communities) seems misplaced.

And his confident assurance that COVID will become just another cold is utterly unfounded.