The hottest Research Substack posts right now

And their main takeaways
Category
Top Health Politics Topics

Getting the Healthcare That Thin People Get

Weight and Healthcare • 179 implied HN points • 30 Apr 22
🏥 Health & Wellness Research
  1. Question healthcare practitioners about treatment in thin individuals to receive equal, evidence-based care as a higher weight patient.
  2. Healthcare system needs a major paradigm shift to accommodate higher weight patients in terms of research, spaces, and training.
  3. Higher weight patients deserve research inclusive of all body sizes, accommodations in healthcare spaces, and doctors educated to work with diverse body types.

Kerala Cancer Data, 2021 Update

Gordian Knot News • 205 implied HN points • 09 Jan 24
🔬 Science Research
  1. The Karunagappally cohort study in Kerala compared cancer rates in high dose villages
  2. Data from the study challenges the Linear No-Threshold model for radiation risk
  3. The updated study suggests low dose radiation exposure may have lower cancer risk than acute exposure

Results from rapid onset autism survey show a significant temporal association with vaccines

Steve Kirsch's newsletter • 6 implied HN points • 20 Nov 25
🏥 Health Politics Research
  1. A survey found a strong link between the timing of vaccines and the onset of autism. This connection should not be ignored.
  2. No previous studies have looked at how vaccine timing relates to autism. This lack of research raises questions about what is being investigated.
  3. The author plans to fund a new, unbiased survey to gather more data on this topic and ensure the questions are reviewed by an impartial expert.

Genetics can predict height, cancer risks, neuropsychiatric diseases and more — but not IQ?

Nucleus Genomics • 4 HN points • 11 Jul 24
🔬 Science Research
  1. Genetics can help us understand traits like height and disease risks, but predicting intelligence (IQ) is much more complex and not fully possible yet.
  2. Current genetic models can give partial insights about how genes may influence intelligence but cannot tell you your exact IQ.
  3. As technology improves, genetic predictions for traits, including intelligence, may get better, providing more useful information about how genetics interacts with environment.

The Long Game 171: Fatigue, Unfair Advantage, Vertical vs. General AI, Sales Learnings

The Long Game by Mehdi Yacoubi • 2 implied HN points • 16 Jan 26
🕹 Technology Research
  1. Being physically tired from real work or regular exercise quiets overthinking and forces action, so most people would actually feel and perform better with more controlled physical fatigue.
  2. Identify the unfair advantages you already have — such as time, connections, money, or skills — and use them relentlessly instead of pretending everyone starts on an even playing field.
  3. General coding or program‑synthesis AI is becoming a powerful horizontal tool that can automate many tasks, but domain‑specific work still matters for enterprise data access, compliance, integration, and closing the last 5–10% of value.
Get a weekly roundup of the best Substack posts, by hacker news affinity:

How Taiwan became Arrakis

Deep Into The Mundane • 17 HN points • 20 Mar 24
🕹 Technology Research
  1. Taiwan's TSMC is crucial to the global economy, producing chips for major tech companies like Apple, Nvidia, and Tesla.
  2. ASML, a Dutch company, has a monopoly on EUV lithography machines, a key technology for chip manufacturing, with machines costing hundreds of millions of dollars.
  3. TSMC's innovative 'pure-play foundry' model, led by visionary Morris Chang, revolutionized chip manufacturing, emphasizing collaboration with designers and avoiding competition with customers.

Teacher Influence and Innovation

New Things Under the Sun • 208 implied HN points • 15 Dec 23
🚌 Education Research
  1. Great teachers can influence their students' interests and innovative career paths.
  2. Studies show correlations between teachers and students in music, science, and research commercialization.
  3. Teacher influence on student achievements and research topics is significant, even in cases of random student-teacher matching.

Missing Fixed Effects Don't Justify Segregation

Cremieux Recueil • 199 implied HN points • 06 Jan 24
🇺🇸 U.S. Politics Research
  1. Justice Jackson's statements about segregation and survival rates of Black newborns were not strongly supported by a 2020 study.
  2. The exclusion of important fixed effects in studies can lead to incorrect or misleading conclusions.
  3. The issue of missing fixed effects is common in research and can impact policy decisions and research priorities.

Talking research integrity at Stanford

Science Forever • 19 implied HN points • 04 Mar 24
🔬 Science Research
  1. Research integrity issues are systemic, not just due to individual mistakes. Institutions must take responsibility for addressing these problems.
  2. Tools like Proofig help catch errors in papers, emphasizing the importance of correcting mistakes promptly.
  3. Reducing stigma around correcting papers, proactive responses from institutions, and encouraging a culture of self-correction are crucial for improving research integrity.

More info, more problems

inexactscience • 39 implied HN points • 16 Nov 23
🔬 Science Research
  1. When people get more information, they often underreact instead of overreact. This means they might ignore new data instead of properly adjusting their predictions.
  2. Experiments showed that when faced with two variables, people made less accurate forecasts. Adding complexity actually made their predictions worse.
  3. Having clear instructions and understanding of the information really helps improve decision-making. If people are confused, they tend to ignore important details.

RLHF progress: Scaling DPO to 70B, DPO vs PPO update, Tülu 2, Zephyr-β, meaningful evaluation, data contamination

Democratizing Automation • 213 implied HN points • 22 Nov 23
🕹 Technology Research
  1. Reinforcement learning from human feedback (RLHF) is a technology that is still unknown and undocumented.
  2. Scaling DPO to 70B parameters showed strong performance by directly integrating the data and using lower learning rates.
  3. DPO and PPO have differences in their approaches, with DPO showing potential for enhancing chat evaluations and happy users of Tulu and Zephyr models.

Lateral thinking

Logging the World • 79 implied HN points • 12 Nov 22
🔬 Science Research
  1. Lateral flow tests had a much lower false positive rate than many initially assumed, around 0.03%, showing their effectiveness.
  2. Data on PCR retests of positive lateral flow tests revealed a positive predictive value of 82% even at low prevalence, supporting the reliability of lateral flow tests.
  3. A rise in prevalence due to variants like delta and omicron, as well as ease in lockdown restrictions, contributed to the wider acceptance of lateral flow tests for controlling the pandemic.

Writing With Your Second Brain

Adjacent Possible • 174 implied HN points • 23 Feb 24
🕹 Technology Research
  1. Deep dive conversations about craft in writing and research are now more accessible through platforms like podcasts and YouTube.
  2. Software tools like NotebookLM and techniques shared by authors like Tiago Forte can revolutionize the way we organize research material and notes.
  3. Integration of tools like ReadWise's 'Export to Docs' feature can enhance the ability to work with research material and create a 'second brain' for storing important ideas.

How to cultivate a high-signal AI feed

Democratizing Automation • 166 implied HN points • 28 Feb 24
🕹 Technology Research
  1. Be intentional about your media diet in the ML space, curate and focus your energy to save time and avoid misleading content.
  2. When evaluating ML content, focus on model access, credibility, and demos; choosing between depth or breadth in your feed; and checking for reproducibility and verifiability.
  3. Ensure to socialize your information, build relationships in the community, and consider different sources and content types for a well-rounded perspective.

Debunking the SCMP report that shaved 36 bln HKD off Baidu

Pekingnology • 188 implied HN points • 15 Jan 24
🕹 Technology Research
  1. The SCMP report falsely claimed a link between Baidu and the Chinese military, resulting in a significant financial impact on Baidu.
  2. The Chinese journal paper discussed theoretical ideas, not real 'military AI' experiments, and lacked academic rigor in its approach.
  3. The paper's experiments were basic simulated scenarios, not real tests, and did not provide actionable insights or findings for military application.

What's Quantum Computing

Technically • 59 implied HN points • 28 Jan 25
🕹 Technology Research
  1. Quantum computing uses qubits instead of bits. While bits can be either 0 or 1, qubits can be both at the same time, allowing for much faster problem-solving.
  2. Qubits can work together in a unique way, using superposition and interference to find answers much faster than traditional computers. This makes them great for complex problems like drug discovery.
  3. Quantum computers are still in the experimental stage and have challenges like needing very cold temperatures and handling errors, but they hold great promise for the future.

AI Roundup 105: AI Action Summit

Artificial Ignorance • 54 implied HN points • 14 Feb 25
🕹 Technology Research
  1. AI regulation is slowing down as countries disagree on how to move forward. Some leaders are critical of existing acts, leading to a lack of international agreement.
  2. China is pushing ahead in an AI arms race, pushing other countries to provide more resources for AI development. Leaders in the industry are predicting rapid advancements in AI, suggesting it might drastically change society soon.
  3. Big tech companies are making strategic partnerships and adjustments to survive in the competitive AI landscape. For example, Apple plans to work with Alibaba for AI in China while other firms are focusing on custom AI designs to reduce dependency on major chip manufacturers.

mathiness instills confidence

bad cattitude • 165 implied HN points • 22 Feb 24
🔬 Science Research
  1. Mathiness can make people feel more confident, especially if they aren't familiar with math.
  2. Adding complex math or 'mathiness' to information can influence how people perceive its quality, especially if they lack knowledge in math and models.
  3. It's important to be cautious of trusting information just because it includes numbers or complex equations; don't assume accuracy or rigor without verifying.

Roche Nanopore: First Thoughts (Pt 1)

ASeq Newsletter • 51 implied HN points • 20 Feb 25
🔬 Science Research
  1. Roche Nanopore's new sequencing method shows really impressive early results, which builds confidence in its technology.
  2. The accuracy of their duplex sequencing is reportedly better than Illumina's, suggesting it could be useful for many different applications.
  3. More information will be shared soon, as there's ongoing discussion and a second post in the works.

To the Reader

Dada Drummer Almanach • 163 implied HN points • 27 Feb 24
📚 Literature Research
  1. Working with what's available can lead to innovative approaches, like practicing 'deconstruction' even in a small college far from big libraries.
  2. Access to resources can change drastically over time, as seen in the shift from paid alumni access to free borrowing privileges at Harvard Library.
  3. The shift to online resources has transformed research practices, offering a wealth of materials instantly but altering the traditional experience of library exploration.

Anger, sadness, guilt, hope: on the complex emotions of climate change

Sustainability by numbers • 238 implied HN points • 28 Aug 23
🌞 Climate & Environment Research
  1. Researchers are uncertain about which emotion is most effective in driving climate action.
  2. Emotions in response to climate change vary and are likely a combination of concern, anger, sadness, and hope.
  3. Pairing emotions like concern, anger, and sadness with hope can be vital for motivating action towards addressing climate change.

Edge 440: Interested in AI Evaluation? Meet Microsoft's EUREKA

TheSequence • 84 implied HN points • 17 Oct 24
🕹 Technology Research
  1. Microsoft's EUREKA is a new framework for evaluating AI models. It helps in analyzing and measuring the abilities of large foundation models more effectively.
  2. The framework goes beyond just giving one score. It provides a detailed understanding of how well AI models perform across different tasks.
  3. EUREKA aims to address the need for better evaluation tools in the industry as current benchmarks are becoming outdated.

Education: How Can It Cost So Much?

Fake Noûs • 294 implied HN points • 15 Apr 23
🚌 Education Research
  1. College education has become extremely costly over the last 50 years, with average tuition soaring to high levels.
  2. A significant portion of the high cost of college goes to faculty salaries, research, administrative work, and not necessarily to benefit students directly.
  3. Academia's administrative procedures can be overly redundant, biased, and counterproductive, leading to high costs without much added value.

Do Higher-Weight People Need Longer Needles for Vaccines?

Weight and Healthcare • 219 implied HN points • 17 Nov 21
🏥 Health & Wellness Research
  1. Intramuscular vaccines need to be injected properly to get the desired immune response. Injecting into fat tissue can lead to less effectiveness and potential adverse effects.
  2. Research on needle length for vaccines often excludes representation of higher-weight individuals, impacting vaccine efficacy and blaming fat bodies for lack of results.
  3. Healthcare practitioners sometimes do not use adequate needle lengths for higher-weight individuals, leading to potential suboptimal vaccinations and highlighting healthcare inequalities.

Heroic Appearances

Painful Signs, Or, Joel's Substack • 39 implied HN points • 26 Oct 23
🚌 Education Research
  1. The ethnicity and appearance of ancient heroes, like Helen, may not align with modern concepts of race and beauty. Ancient depictions and concepts of color and appearance were likely different.
  2. Descriptions of heroes in ancient Greek literature, such as Helen, were often open to interpretation and projection by the audience, allowing for individualized ideals of beauty.
  3. Homeric poetry intentionally left physical descriptions of figures like Helen vague, encouraging audience engagement and diverse interpretations rather than prescribing fixed appearances.

🌻 E46: Next-Gen Agent or a True Leap in AI Consciousness?

Musings on AI • 72 implied HN points • 11 Nov 24
🕹 Technology Research
  1. AI agents are still developing but show promise for the near future. They're getting better at aligning with human values and being more useful.
  2. Stanford's new method using Information-Directed Sampling helps AI learn more efficiently while keeping human preferences in mind. It can adapt well in changing environments.
  3. As AI becomes more common, we might see a mix of human-friendly websites and those that cater directly to AI agents. This means both types of users can interact effectively.

Discussion with Steve Stewart-Williams on evolutionary psychology

Optimally Irrational • 63 implied HN points • 18 Dec 24
🚌 Education Research
  1. Evolutionary psychology studies behavior based on human evolution. It looks for reasons why we think and act the way we do by using ideas from evolutionary biology.
  2. While some critics argue that evolutionary psychology can lead to over-simplified stories about behavior, it still provides useful insights into human nature and basic drives.
  3. Understanding how evolution shapes our thoughts and actions can help us improve mental health and make better decisions in life and relationships.

ChatGPT is Quietly Rewiring Your Brain

Perspective Agents • 21 implied HN points • 08 Jul 25
🕹 Technology Research
  1. ChatGPT and AI can change the way we think, sometimes making it harder for us to form our own ideas. We have to be aware of how they're framing the information we use.
  2. Using AI can either make us smarter or dumber, depending on how we interact with it. If we treat AI like a partner and question its suggestions, we can boost our own thinking.
  3. It's important to be intentional when using AI tools. Instead of just accepting the first answer, we should challenge AI and think deeply about the results to keep our skills sharp.

Do we need RL for RLHF?

Democratizing Automation • 182 implied HN points • 06 Dec 23
🕹 Technology Research
  1. The debate around integrating human preferences into large language models using RL methods like DPO is ongoing.
  2. There is a need for high-quality datasets and tools to definitively answer questions about the alignment of language models with RLHF.
  3. DPO can be a strong optimizer, but the key challenge lies in limitations with data, tooling, and evaluation rather than the choice of optimizer.

Further Reading

Material World • 164 implied HN points • 30 Jan 24
🚌 Education Research
  1. There is a wide range of reading materials available to delve deeper into the topics discussed in the book.
  2. Various books and resources are recommended for exploring different aspects of material science, such as sand, salt, iron, copper, oil, and lithium.
  3. Authors and specific books are highlighted as excellent starting points for those interested in learning more about specific material-related subjects.

Batch Calibration for LLMs

MLOps Newsletter • 39 implied HN points • 21 Oct 23
🕹 Technology Research
  1. Flash-Decoding optimizes attention to speed up decoding of Large Language Models (LLMs).
  2. Batch Calibration (BC) is a new zero-shot calibration method for LLMs, improving accuracy without labeled data.
  3. MiniGPT-v2 introduces unique identifiers for tasks, enhancing performance on vision-language tasks.