3 ms·

This is not stupid. The reporting is stupid. The way science here works: Step 1: People come up with a hundred hypotheses Step 2: Six of them are confirmed in

This is not stupid. The reporting is stupid. The way science here works: Step 1: People come up with a hundred hypotheses Step 2: Six of them are confirmed in preliminary studies, five by fluke (false positives) and one true positive Step 3: The one true positive is confirmed in multiple replication studies Step 3 is when we get to a reasonable scientific theory or scientific fact. The problem is step 1.5 or 2.5: Popular press picks up a hypothesis as "science" or (at 2.5) as peer-reviewed science, often promoted by the original researcher. We absolutely should be investigating heparin, ivermectin, vitamin D, zinc, and other speculative drugs for effects. Outside of exceptional circumstances, we absolutely should not be recommending those for general use until they're validated. As a footnote, machine learning supercharges this process. It can generate countless speculative drugs very rapidly. The vast majority are nonsense, but the tiny minority which work make it still worthwhile. If we have a something with a 1% chance of curing a disease, it takes, on average, 100 similar trials to cure that disease. For major diseases, that (or even much longer-shot cures) is completely worth it.

SketchySeaBeast 3y ago
> As a footnote, machine learning supercharges this process. It can generate countless speculative drugs very rapidly. The vast majority are nonsense, but the tiny minority which work make it still worthwhile. Is this actually happening right now or are we just creating a speculative future for ourselves. Who is running the sanity checks on these huge number of potential medications to make sure the drugs aren't obviously poisonous?
- frognumber 3y ago
  Yes. This is happening. Ultimately, drugs created this way go through the same checks as any other drug. Most drugs, created this way or otherwise, don't work and are potentially poisonous. That's why the processes here are super-rigorous.

SketchySeaBeast 3y ago

> As a footnote, machine learning supercharges this process. It can generate countless speculative drugs very rapidly. The vast majority are nonsense, but the tiny minority which work make it still worthwhile. Is this actually happening right now or are we just creating a speculative future for ourselves. Who is running the sanity checks on these huge number of potential medications to make sure the drugs aren't obviously poisonous?

frognumber 3y ago
Yes. This is happening. Ultimately, drugs created this way go through the same checks as any other drug. Most drugs, created this way or otherwise, don't work and are potentially poisonous. That's why the processes here are super-rigorous.