…the expert woodworker does not have a single "best" tool for any job, the expert researcher knows how to draw on a range of tools to find the ones most appropriate to a given situation.
To claim otherwise reflects a fundamental misunderstanding of how science works. Like any tool, RCTs have their advantages - and they also have their disadvantages. Science is a specialty (with many subspecialties), and just as…
While ranked highly within the EBM (evidence-based medicine) process for decision-making by non-scientists, this is a reflection of their utility as a heuristic, rather than any inherent superiority to other methods. twitter.com/ToshiAkima/s...
🧵(1/n) An unfortunate example of confusion around the value of RCTs common to practitioners with limited research training. RCTs are a useful scientific method, but they are only a small part of the biomedical researcher's deep toolbox. bsky.app/profile/mark...
A thread from 2023.05.08 addressing the over-valuing of Randomized Controlled Trials (RCT) driven by Evidence-Based Medice (EBM) hierarchies. Not everyone realizes that EBM is just a protocol meant to help non-scientists interpret science, as opposed to an objective description of how science works.
"Unnecessary content" placeholder I will use in reposting old Twitter threads with broadly applicable content on bsky. Some were responses to positions taken by others, where the value today is in addressing the idea rather than naming & shaming any one particular individual.
Without a deep understanding of the underlying processes, I get the impression that the current "big thing" AI's effectively average everything in the training dataset into an n-dimensional semantic map, take questions as coordinates, and spit out whatever they find there. Inherently antidiverse.
It'd likely be very similar looking at natural scientists in relation to the pandemic. People like Morawska / Marr / Prather / Jimenez are the archetypes, but the moral injury caused across fields by knowing that millions of people have died unnecessarily is just crushing.
A tool for precisely calibrating 3D printer horizontal expansion, so you can print things that fit exactly into existing parts or other 3D printed objects. The accuracy limit will depend on your printer but my Taz5 can routinely get down into the 10s of microns. www.printables.com/model/519880...
