Research

How Expert Is That Expert?

To know more about recent developments in their field, many scientists form “journal clubs”, where they gather regularly and read recent research papers and discuss. By dipping their toes into topics adjacent to their field of study, they’re able to keep an eye out for future research topics. They’re less knowledgeable than experts who have done research on those specific topics. You can determine if an expert’s claims are journal club-level knowledge if they cite papers, researchers, and specific results without referencing any personal research they’ve done on the topic. Read full article here

Correction Of Scientific Literature: Too Little, Too Late!

Scientific papers have never been as quickly disseminated as they are today, which is a double-edged sword. While the public finds out about new studies within a few days of publication, it also means that misinformation and poor-quality research reaches the masses at a breakneck pace. The latter is a problem because there is a stigma around research correction that prevents experts from revising studies without negative effects on their careers or throwing away tempting benefits from ignoring their errors. Read full article here

Desci: Decentralized Science?

Two problems researchers face when getting their findings published are the threat of losing access to their published work and the inaccessibility of research to the general public. Decentralised Science (DeSci) aims to solve these problems. Built on Web 3.0 technology like blockchain, DeSci claims it will let researchers retain the rights to their published work and allow for replication of research for easy dissemination without needing to pay expensive fees to access them like you currently need to do with research journals. Read full article here

Language Homogenization At Harvard

Analysing the cosine similarity of word diversity among research papers and Harvard news articles reveals that there’s a trend towards homogenisation, suggesting a “constriction” of ideas. Cosine similarity is the measure of the cosine of the angle of two arrows; a smaller number means a closer distance and more homogenisation. The cosine similarity of these documents was determined by feeding them through a machine learning algorithm that assigned values to the words, plotted them on a graph, and determined their cosine similarity. Read full article here

On (Not) Reading Papers

“[A] proper reliability estimate might simply be impossible given just what is written in the text.” With the plethora of scientific papers published each year, scientists have taken to skimming papers. One reason why this isn’t so bad of a habit is that not all the relevant information is actually available in a paper. For example, sample sizes are frequently larger than reported in the paper, often due to botched samples. However, while the data may not be complete, the underlying consensus is frequently carried through to the abstract. Read full article here

On (Not) Reading Papers

“[A] proper reliability estimate might simply be impossible given just what is written in the text.” With the plethora of scientific papers published each year, scientists have taken to skimming papers. One reason why this isn’t so bad of a habit is that not all the relevant information is actually available in a paper. For example, sample sizes are frequently larger than reported in the paper, often due to botched samples. However, while the data may not be complete, the underlying consensus is frequently carried through to the abstract. Read full article here

Why Isn’t There a Replication Crisis in Math?

The replication crisis describes the growing problem within the sciences where researchers can’t replicate the results of many published studies. But one field that remains unaffected is mathematics. Jay Daigle says this is the case because reading mathematics papers is a form of replication already. This is because math papers are the “real work” and not descriptions of the actual work done in experiments. By going through and checking the proofs in a math paper, you effectively replicate the work done by the author of the paper. Read full article here

What A Dead Salmon Can Teach Us About How We Use Machines

A dead salmon shed light on the shortcomings of fMRI (functional magnetic resonance imaging), which is used in research to visualise brain activity. Researchers stuck a dead salmon in an fMRI machine and presented it with pictures of human interactions. Surprisingly, the machine found significantly increased brain activity at several points. Of course, after applying corrective methods for multiple comparisons — which very few fMRI studies implement — these results were found to be false positives and were discarded. This raises questions about the accuracy of many fMRI studies. Read full article here

T-Shaped People And Academia

“T-shaped people” are what Bastian Rieck calls people with knowledge in one or more different fields, as opposed to “I-shaped people,” who are experts in one field. In academia, T-shaped people are often looked over by I-shaped people because they seem unfocused or lacking in expertise. This is most visible when looking at a T-shaped person’s record of academic publications versus that of an I-shaped person’s; an I-shaped person often has multiple papers under their belt, while a T-shaped person has comparatively few, presumably because they work on broader projects. Read full article here

The Ultrafast Snap Of A Finger Is Mediated By Skin Friction

Researchers investigated the mechanics behind the finger snap under the scope of a latch-mediated spring-actuated system (LaMSA), where, in technical terms, “energy is loaded in a mass–spring system by an external motor over a relatively long period of time before being held in place with a latch.” A snap of the finger is found to deliver the greatest rotational acceleration seen in humans, reaching an accelation of 1.6 106° s2 in 7 ms. The key seems to be in the friction that our skin provides, acting as the “latch”. Read full article here

FASCINATING READS, SUMMARISED

The occasional email full of conversation-worthy content