Tag: science

Skeleton Lake

A new study, dated and analyzed the DNA from the bones of 37 individuals found at Roopkund. The majority of the deceased indeed died 1000 or so years ago, but not simultaneously. And a few died much more recently, likely in the early 1800s. Stranger still, the skeletons’ genetic makeup is more typical of Mediterranean heritage than South Asian.

“It may be even more of a mystery than before. It was unbelievable, because the type of ancestry we find in 33% of the individuals is so unusual for this part of the world.”

GPT-2 6-Month Follow-Up

people find GPT-2 synthetic text samples almost as convincing (72% in one cohort judged the articles to be credible) as real articles from the New York Times (83%) we expect detectors to need to detect a significant fraction of generations with very few false positives. Malicious actors may use a variety of sampling techniques (including rejection sampling) or fine-tune models to evade detection methods. A deployed system likely needs to be highly accurate (99.9%–99.99%) on a variety of generations. Our research suggests that current ML-based methods only achieve low to mid–90s accuracy, and that fine-tuning the language models decreases accuracy further.

The Greening Earth

The earth is getting greener, in large part due to increased CO2 in the atmosphere. Surprisingly, however, another driver is programs in China to increase and conserve forests and more intensive use of cropland in India.

2022-07-21: While Amazon deforestation continues, in other places it is reforestation.

England has 2x the amount of forestland in the past 150 years, and now has as much land dedicated to forests as the year 1350.

Life sciences progress

Academia has a lot of problems and it could work much better. However, these problems are not as catastrophic as an outside perspective would suggest. My (contrarian, I guess) intuition is that scientific progress in biology is not slowing down. Specific parts of academia that seem to be problematic: rigid, punishing for deviation, career progression; peer review; need to constantly fundraise for professors. Parts that seem to be less of a problem than I initially thought: short-termism; lack of funding for young scientists.

2022-06-02: A contrarian perspective

Stepping back, I’m claiming that science is getting harder, in the sense that it is increasingly challenging to make discoveries that have comparable impact to the ones in the past. Diverse groups – the Nobel nominators, contemporary surveyed scientists, academics, and inventors – all seem to have an increasing preference for the work of the past, relative to the present. And looking at growth in the number of topics covered by scientists also suggests it has become harder to make forward progress.

Approximating Pi

Under what circumstances is it possible to represent irrational numbers that go on forever—like pi—with simple fractions, like 22/7? The proof establishes that the answer to this very general question turns on the outcome of a single calculation. The Duffin-Schaeffer conjecture has you add up the measures of the sets of irrational numbers captured by each approximating fraction. It represents this number as a large arithmetic sum. Then it makes its key prediction: If that sum goes off to infinity, then you have approximated virtually all irrational numbers; if that sum instead stops at a finite value, no matter how many measures you sum together, then you’ve approximated virtually no irrational numbers.

Pleasure & Pain Log Scales

Based on: the characteristic distribution of neural activity, personal accounts of intense pleasure and pain, the way various pain scales have been described by their creators, and the results of a pilot study we conducted which ranks, rates, and compares the hedonic quality of extreme experiences, we suggest that the best way to interpret pleasure and pain scales is by thinking of them as logarithmic compressions of what is truly a long-tail. The most intense pains are orders of magnitude more awful than mild pains (and symmetrically for pleasure).

Against High Energy Physics

high energy physics thinks that bigger colliders will let them find new particles, but this view is probably wrong.

You could build a circular machine 3x the size of the Large Hadron Collider to collide electrons and positrons; you could upgrade the LHC, or even build a next-generation linear accelerator. Probing higher energies offers the hope of new physics — it could be supersymmetry, it could be something else, I don’t know what. But before exploring higher energies, it makes sense to me to build a muon collider, and to clarify the question of the Higgs first. Here we already have a particle that we want to explore. We may even find signs of new physics by studying the Higgs very precisely. For that we don’t need to go to a 100-kilometer-around tunnel. Think about how many days it takes to walk 100 kilometers! And it all has to be extremely functional, every single piece has to work — it’s a miracle if people succeed in making it work.

5x heat exchangers

Possible 5X improvement in heat exchangers

Researchers have discovered a simple way to give a major boost to turbulent heat exchange, a method of heat transport widely used in heating, ventilation and air conditioning (HVAC) systems. In a paper published in Nature Communications, the researchers show that adding a readily available organic solvent to common water-based turbulent heat exchange systems can boost their capacity to move heat by 500%