What Adam is Reading - Week of 2-3-25

Week of February 3, 2025

 

I am grateful that our kids, who have been back at college for two weeks now, call and text with stories, updates, and questions.  This week, my wife and I remotely assisted with removing disruptive web browser extensions and strategies for dealing with a flat bike tire.  Each call brings mixed emotions - the satisfaction of being needed and the question, "Am I an inadequate parent for not previously teaching my kid this skill?"  Among the various empty nest activities (mahjong, adult dinner parties, and house decluttering), I did not anticipate being a guilt-driven tier-3 help desk.

 

P.S.  When I shared this opening with my older son and asked for a 1-10 score on my tech support, he replied, "8; during the height of the [browser extension] crisis, the lack of response time was very disheartening.  [You took 12 hours to reply to my initial text for help].  [Your help desk] has yet to provide clear working hours."  Feedback is a gift?!?

 

 

---

Listen to a Google Notebook LM A.I.-generated podcast of the newsletter with two virtual "hosts."

 

My effusive hosts did a good job this week keeping it to 10-minutes and minimizing the pontification.

https://drive.google.com/file/d/17DjcSlf1kYCpOVW8-5K0nIyH1lWfo8Dc/view

 

About NotebookLM: https://blog.google/technology/ai/notebooklm-audio-overviews/

------

 

Science and Technology Trends

 

A new, non-opioid pain medication was approved by the FDA this week - suzetrigine (AKA Journavx).  This news is a big deal - we medical people only have 4-5 classes of pain medications, all of which have some trade-offs and limitations due to efficacy, complications, or dosing limits.  The new medication has been years in the making and is a good example of how benchtop research and hypotheses can translate to clinical medicine.  In this instance, researchers investigating how information propagates along peripheral nerves (nerves not in the spinal cord or brain) yielded an understanding of how to block the signals and thus control some types of pain.  And, in the spirit of a good story, there is a seemingly bizarre array of converging topics, including man-on-fire syndrome, a family of Pakistani fire walkers, and puffer fish poison. 

https://www.nytimes.com/2025/01/30/health/fda-journavx-suzetrigine-vertex-opioids.html?smid=nytcore-ios-share&referringSource=articleShare

and

https://www.fda.gov/news-events/press-announcements/fda-approves-novel-non-opioid-treatment-moderate-severe-acute-pain

 

I did not realize there was a commercial/industrial CT scanning market.  However, using CT scans to evaluate everything from welds to candy is hitting a speed and price point such that (it appears) to be a viable means of manufacturing quality control.  One of the companies offering this service, Lumafield, provides some interesting articles on CT-scanned, 3D-reconstructions of water filters, candy, and other various manufactured goods.  The before and after images of new and used water filters were fascinating.

https://www.lumafield.com/article/whats-inside-your-water-filter-a-ct-scan-comparison

and

https://x.com/JonBruner/status/1885020930533187893

and

https://www.lumafield.com/article/sweet-precision-ct-and-food-safety-candy

 

The article "Increasing rat numbers in cities are linked to climate warming, urbanization, and human population" amused me.  I was particularly interested in the "trend in rat numbers vs. temperature change" or "trend in rat numbers vs. available city green space" graphs.  While I would not want to conflate causation and association, my first thought was that increasing the temperature in a city makes rats replicate (the visual in my head was like getting a mogwai wet in the 1980s Gremlins movie).  Besides that, the article is a good environmental biology review reinforcing that rats are among the many species that win when humans make a mess.

https://www.science.org/doi/10.1126/sciadv.ads6782

and

What went through my mind in reading this article - Getting Gizmo Wet scene: https://www.youtube.com/watch?v=NLZP4QA4mU4

and

I am not the only person who has pondered the biology of Gremlins:

https://medium.com/@Naturalish/gremlins-three-rules-an-evolutionary-analysis-de4c4fae2785

 

 

Anti- Anti-Science Articles of Note

 

I went down the rabbit hole of the anti-vaccine movement this week - a cesspool of biased and logically fallacious ideas.  The impetus of my exploration was an article that popped up on social media - a retrospective review associating vaccine administration with various illnesses (seizures, autism, etc.) using billing claims from the Florida Medicaid database.  The "journal article" [the journal is an online blog that pretends to be a peer-reviewed journal] offers a lot of data.  However, there are so many methodologic errors and biases it is laughable.   Science writer Dr. Jess Steier published a detailed analysis of the vaccine article, highlighting all the flaws.  However, as I have said before, refuting pseudoscientific writing is far more challenging than generating doubt through broad assertions and misused data.  (And to be abundantly clear - the overwhelming evidence demonstrates the safety and efficacy of vaccines far outweigh the risks of side effects.)

Dr. Steier's review of the paper (including a link to the fake journal article):

https://theunbiasedscipod.substack.com/p/anatomy-of-a-failure-why-this-latest

The kinds of social media discussion that these anti-science articles generate:

https://www.reddit.com/r/ScienceBasedParenting/comments/1icudwi/help_analyzing_these_antivax_studies/?rdt=38576

The Unbiased Science Podcast

https://www.unbiasedscipod.com/

And if all of this is too much reading, you can watch Dr. Noc's Instagram anti-antiscience videos:

https://www.instagram.com/dr.noc?igsh=aThleDdkaDk5ZnZl

 

 

Living with A.I.

 

AI-focused blogger Trung Phan offered the best long-form discussion of Deepseek I found this week.  He breaks down the knowns and unknowns from the LLMS training, the intellectual property issues, the politics of Chinese A.I., how to build a nimble A.I. team, and (the much-touted) costs of running the model.

https://www.readtrung.com/p/deepseek-links-and-memes-so-many?hashed_user=ca3e215fc660247aa63f3956718db782&utm_source=Sailthru&utm_medium=email&utm_campaign=02%2F01&utm_term=Trung

And, for the TLDR crowd, here is a Claude AI Summary of the blog post:

https://claude.site/artifacts/1a14350f-5301-4386-a3e6-c41bf826070d

 

Professor Ethan Mollick's latest blog addresses a good question, "Today I'm tackling the one question I get asked most: what A.I. should you actually use?  Not five years from now.  Not in some hypothetical future.  Today."  He offers an excellent compare and contrast of the different models from OpenAI, Anthropic, and Google.

https://www.oneusefulthing.org/p/which-ai-to-use-now-an-updated-opinionated

 

Infographics

 

We have driven one of our cars, an Infiniti SUV, more than 230,000 miles.  I did not appreciate how rare that was.

https://alansfactoryoutlet.com/the-20-longest-lasting-vehicles/

 

 

Things I learned this week

 

Adding to the list of animals we should not resurrect; I offer the 300-million-year-old Tully Monster.  It is a 12-inch slug-like water-borne creature with "a segmented torpedo of a body and two rippling tail fins that look a little like the wings of squid at the rear.  At the front there is a long, thin hose-like snout with a tiny, tooth-filled grabbing claw.  In addition,  there is a solid bar running from side to side across the top of the creature, horizontal stalks on which are set bulbous organs of some kind, which are generally assumed to be its eyes." You need to see this and remind yourself, once upon a geologic time, you could have been "bitten" by one.

https://www.smithsonianmag.com/science-nature/what-tully-monster-scientists-finally-think-they-know-180958422/

and

https://en.wikipedia.org/wiki/Tullimonstrum

 

A global cultural exchange that makes me smile.  "How Danish Cookies Became Hong Kong's Favorite Lunar New Year Snack."

https://www.atlasobscura.com/articles/kjeldsens-danish-cookies-hong-kong

 

My favorite headline from last week:  "Scientists Use Grapes to Juice Quantum Sensor Performance.  "Macquarie University researchers have discovered that due to their water content, supermarket grapes can enhance microwave magnetic fields, potentially advancing the development of compact and cost-effective quantum sensors."  I had so many questions about how they knew to do this - but it appears social media posts played a role, "The research builds on viral social media videos showing grapes creating plasma – glowing balls of electrically charged particles – in microwave ovens." 

https://thequantuminsider.com/2024/12/24/scientists-use-grapes-to-juice-quantum-sensor-performance/

and

https://journals.aps.org/prapplied/abstract/10.1103/PhysRevApplied.22.064078

Plasma grapes: https://www.youtube.com/watch?v=wCrtk-pyP0I

 

 

A.I. art of the week (A visual mashup of topics from the newsletter, now using ChatGPT to summarize the newsletter, suggest prompts, and make the images).

 

This week, Adam and DALL-E wrestled to describe and generate an anatomically correct Tully Monster.  After numerous prompts and asking other engines to help with prompt writing, I ran out of patience.   Here is DALLE's Disney-fied result of a "Digital painting of a whimsical underwater scene. Two friendly, prehistoric creatures called 'Tully Monsters' are having a picnic on a sandy lake bottom. The Tully Monsters are pale blue and light pink, with slug-like bodies, long proboscises ending in claws, and eyes on stalks.  They are eating green grapes and Danish butter cookies.  Soft light filters through the blue-green water, illuminating aquatic plants and creating dappled shadows.  A tin of butter cookies sits nearby.”

https://drive.google.com/file/d/1RL0gRvN7_P0rqDC4B9BacTZLOeBc7YK1/view

 

---

1 in 108 Americans were infected with COVID last week.  The PMC team projects COVID rates will start rising again shortly.

 

The Pandemic Mitigation Collaborative (PMC) website uses wastewater levels to forecast 4-week predictions of COVID rates.

https://pmc19.com/data/

based upon https://biobot.io/data/

 

Wastewater Scan offers a multi-organism wastewater dashboard with an excellent visual display of individual treatment plant-level data.

https://data.wastewaterscan.org/

----

 

 

Clean hands and sharp minds,

 

Adam

Comments