Let's hope for the best. But his article reads too much like Michael Crichton for my taste.<p>> Hospitalized COVID-19 patients on famotidine appeared to be dying at a rate of about 14% compared with 27% for those not on the drug, although the analysis was crude and the result was not statistically significant.<p>That's a pretty large difference. For it not to be statistically significant, n is unlikely to be >50.<p>Since I doubt anything close to half of any random sample (?) takes this drug, we'd ideally want the control group significantly larger than the treatment group.<p>The solution I've come up with is 1/7=14% and 3/11=27%. Which also explains why they don't disclose these numbers.<p>It's important to note that there's really isn't a single virus that we can cure with a cheap and benign drug. Or really any drug. HIV took 40 years, untold billions, and is an uneasy stalemate more than a cure. HEP C is curable in the sense that most patients will rapidly improve upon hearing about the costs and side effects. Herpes? Influenza? Nope & no. Rabies? Standard of care is, or should be, a gun. Measles/Mumps/Ebola/Marburg/Lassa/HEP A/B/D/E? Not one cure!<p>Then, there's the "classified" protein docking simulations. I used to work in unclassified protein docking studies at a Max Planck I. If we had known there was an option to get our work classified, that would have done wonders for our careers! Unfortunately, all our failures happened out in the open, just as everyone else's.<p>This idea of "designing" molecules to act with some specific mechanism that you come up with by studying 3d-structures and running molecular dynamics is... controversial. As of five years ago when I left it was about as useful as astrology. Maybe it has improved: some parts are near-perfect scenarios for DNNs. But I somewhat doubt it. Even for molecules where we know the exact interactions they participate in, we suck at predicting the effect: you meddle with a bit of NO and every physiological textbook predicts it will lower blood pressure. Then you test it with some volunteers and they start stashing the stuff because it gives them a boner. And blue vision, somehow...<p>In as far as small-molecule drug research is still happening (as opposed to more lucrative and potentially far more powerful biologicals) it still happens the old-fashioned way of screening "randomised" candidates, only on far larger scales than ever: start with some structures that have been successful in similar situations, collect a few thousand variations based on your gut instinct, some simulation, and what's left over from that experiment the PhD candidate was running until she bit of that hamster's head in frustration Sunday night. Throw away everything that would kill you, throw away everything that might explode, throw away everything your target bacterium eats like ice cream, throw away everything that makes the mice question their job choices and intentionally fail the swim test, throw away everything Martin Shkreli has patented. Then spend a billion trying to find out how cancerous the two or three candidates left over <i>really</i> are. Yes, that's the <i>good</i> process.<p>Maybe we get lucky! This probably deserves to be tested, especially considering how cheap and safe it is. (Far too much bandwidth was wasted on you-know-what because of you-know-who that should have been used to go wide).<p>But I'd put chances of a successful drug treatment at <5%, some improvements in treatment protocols (ventilation etc) that cuts deaths by half or more at maybe 25%, and a vaccine at >80%. It also seems to me that our chances of eradication-through-behavioural-changes are underestimated: finding a way to sustainably push R<1.0 will, eventually, lead to eradication. Get < .5 (tests! tests! tests!) and it's a matter of weeks.