The hottest Substack posts right now

according to Hacker News
Category
Mindful Modeler • 898 implied HN points • 07 Feb 23
  1. It's important to avoid assuming one method is always the best for all interpretation contexts when working with machine learning interpretability tools like SHAP.
  2. Different interpretability methods like SHAP and permutation feature importance (PFI) have unique goals and can provide different insights, so it's crucial to choose the method that aligns with the specific question you want to answer.
  3. Research on interpretability should be more driven by questions rather than methods, to ensure that the tools used provide meaningful insights based on the context.
Technically • 14 implied HN points • 05 Feb 26
  1. Modern generative models mirror pathways in the human brain, and many researchers believe leveraging that similarity could be key to much stronger AI.
  2. Real cloud-spend data shows the fastest-growing AI use cases are coding agents, low-latency LLM inference, and computational biology, while AI art and video generation have plateaued as the market professionalizes.
  3. Models overuse em dashes mainly because of their training data and tokenization quirks—older texts and auto-converted punctuation make the em dash common—and this highlights how dataset quality and representativeness drive model behavior.
In My Tribe • 698 implied HN points • 02 Jan 25
  1. Many companies are reducing their number of middle managers to trim costs and cut bureaucracy. This means fewer people are overseeing employees, which can simplify decision-making.
  2. Firms are focusing on their core business and letting go of complex managerial structures that don’t add immediate value. They might be cutting back on certain initiatives to streamline operations.
  3. Cost-cutting has become a priority for many senior executives, especially when revenue growth slows. This focus on reducing expenses can lead to a significant reshaping of company structures.
Concepts of Finance 🧠 • 199 implied HN points • 07 Mar 24
  1. Commodity traders buy and sell things like oil, gold, and wheat. They try to predict price changes based on global events to make profits.
  2. Their work impacts everyday prices for many products we use, helping producers manage risks and securing stable prices for the future.
  3. Traders pay attention to weather, politics, and market feelings to make informed decisions, using tools like futures contracts and diversification to manage risks.
Astral Codex Ten • 1858 implied HN points • 07 Mar 24
  1. ACX Grants involves an impact market where investors crowdfund grant proposals with potential for big profits if successful.
  2. Only 44 out of approximately 140 qualified participants have signed up for the impact market so far, with a different approach allowing non-accredited investors to participate.
  3. Investing in impact certificates through this system has risks like potential project failure, but also benefits such as tax deductibility.
Get a weekly roundup of the best Substack posts, by hacker news affinity:
Quant Trading Rules • 79 implied HN points • 02 Jun 24
  1. Peter Thiel emphasizes the importance of future cash flows in valuing technology companies, projecting significant value from profits generated well into the future.
  2. Interest rates, linked to Treasury bond yields, affect stock valuations by altering discount rates for future cash flows, particularly impacting tech stocks with high growth expectations.
  3. Implementing trading strategies based on spreads and RSI signals between tech stocks (like QQQ) and Treasury bonds (like TLT) can show variations in returns, with opportunities for leveraging and rule adjustments to potentially improve performance.
Read Max • 2186 implied HN points • 21 Dec 23
  1. Many feel the internet is less fun now because of changing demographics and shifting cultural preferences.
  2. The internet may not be objectively worse, but it is used differently by various audiences.
  3. The embrace of AI in tech culture is influenced by historical themes of hierarchical society and authoritarian politics.
The Joyous Struggle • 395 implied HN points • 27 Nov 23
  1. Many people have mixed feelings about technology, especially artificial intelligence, due to fear of missing out, lack of understanding, and a sense of exclusion from the tech world.
  2. The author shares a sense of 'tech incredulity' toward AI, questioning its potential impact, limitations, and whether it truly warrants the level of concern it receives.
  3. Despite not having expert knowledge, the author acknowledges a responsibility to learn more about AI, to demystify the complexities surrounding it, and to understand the risks, potential, and ethical implications better.
PETITION • 275 implied HN points • 24 Jan 24
  1. A medical apparel seller has filed for bankruptcy due to various financial challenges and changes in the industry.
  2. The restructuring process involves equitizing first lien debt and securing a DIP financing with interesting features.
  3. The bankruptcy filing of Terraform Labs Pte. Ltd., known for the failed stablecoin Terra USD, has led to ongoing legal proceedings and the company's intent to continue its business plan.
Substack • 635 implied HN points • 28 Jan 25
  1. Substack now offers translations in six new languages, making it easier for more people to read posts. This means readers can enjoy content in French, German, Spanish, Italian, Portuguese, and Brazilian Portuguese.
  2. Live video previews are now part of the feed. This helps users discover and join livestreams that are open to everyone, making it easier to connect and engage.
  3. New community moderation tools allow admins to manage comments more effectively. Users can now ban commenters right from the app, helping maintain a positive discussion environment.
ChinaTalk • 622 implied HN points • 01 Feb 25
  1. DeepSeek is a unique AI research lab that has no pressure to make money. This allows them to focus on innovation and open-source work without the typical commercial constraints most tech companies face.
  2. They prioritize hiring young, talented engineers who are passionate about technology. This approach leads to fresh ideas and creativity, breaking from traditional hiring practices in other companies.
  3. DeepSeek's relationship with the Chinese government is evolving, with potential benefits and challenges. As they gain more attention, there are questions about how much freedom they'll have in their open-source projects.
Import AI • 459 implied HN points • 30 Oct 23
  1. UK's intelligence services are slightly worried about the safety implications of generative AI technologies, particularly in amplifying existing risks like cyber-attacks and digital vulnerabilities
  2. Research shows that a basic Transformer neural net architecture can meta-learn and match human performance in inferring complex rules from small data, hinting at AI systems increasingly displaying human-like qualities
  3. Facebook's Habitat 3.0 software enables training and testing agents to collaborate with humans by simulating realistic 3D environments with humanoid avatars, human-in-the-loop interactions, and benchmark tasks for human-robot interaction
UX Psychology • 297 implied HN points • 12 Jan 24
  1. Increased automation can lead to unexpected complications for human tasks, creating a paradox where reliance on technology may actually hinder human performance.
  2. The 'Irony of Automation' highlights unintended consequences like automation not reducing human workload, requiring more complex skills for operators, and leading to decreased vigilance.
  3. Strategies like enhancing monitoring systems, maintaining manual and cognitive skills, and thoughtful interface design are crucial for addressing the challenges posed by automation and keeping human factors in focus.
Abstraction • 29 implied HN points • 09 Jan 26
  1. A single probability for a time window needs a decay model because where the probability mass sits across the window determines how much chance remains as time passes.
  2. Probability can follow different hazard patterns—constant (linear decay), increasing (back-loaded, like last‑minute negotiations), decreasing (front‑loaded, like ceasefires), or event‑driven—and each pattern changes how fast the cumulative probability is consumed over time.
  3. The forecasting bot classifies which hazard applies (defaulting to constant when unsure) and uses that to update remaining probability as time elapses, but this is a refinement that can be misclassified and matters most for long‑horizon questions.
davidj.substack • 47 implied HN points • 13 Dec 25
  1. Routine, language-driven legal tasks are likely to be automated, so junior and mid-level lawyer roles will shrink while partners and senior lawyers who provide judgment, sales, accountability and human interaction stay essential.
  2. Firms will become more top-heavy and need far fewer junior hires, which will reduce demand for law graduates—especially from second- and third-tier programs—and increase competition for the remaining positions.
  3. This is part of a wider knowledge revolution: AI will replace much routine knowledge work across industries, reshaping labour markets and the economy in a way comparable to the industrial revolution.
Bite code! • 978 implied HN points • 13 Oct 24
  1. Always have your business logic on both the frontend and the server. If you only trust the client side, you risk getting incorrect data.
  2. Your server needs to handle requests from various sources, including non-standard browsers and bots. These can bypass your frontend checks if they're not replicated on the server.
  3. Any important checks for security and data integrity should happen on the server to prevent unexpected issues. This means you'll often have to duplicate checks that you already do on the frontend.
AI Research & Strategy • 237 implied HN points • 07 Mar 24
  1. A Google engineer was arrested for leaking important AI designs, which could have serious effects on the company's competitive edge. It's alarming that such sensitive information can be accessed so easily.
  2. Once Google discovered the employee's suspicious activity, they didn't act quickly enough. Instead of launching a serious investigation, they let him continue working for a bit, which gave him a chance to escape.
  3. This situation raises concerns about how other companies might handle security. If a major firm like Google has weaknesses, it makes you wonder about the safety of information in smaller firms and universities.
In My Tribe • 303 implied HN points • 11 Jun 25
  1. A conversation with AI is different from simply asking a question. You can explore topics more deeply and learn from the back-and-forth interaction.
  2. Using AI for projects is essential to becoming skilled with it. It’s like doing a group assignment, where you can create something together.
  3. Providing clear instructions and materials to AI helps it assist you better. Treating it like a partner, rather than just a tool, can lead to better results.
NEUROTECH FUTURES • 99 implied HN points • 15 May 24
  1. Commercializing Brain Computer Interfaces (BCI) is a long-term endeavor that requires patience and perseverance
  2. Educating the public about neurotechnology involves balancing technical details with ethical considerations
  3. Collaboration and communication among experts in neurotechnology is crucial to advance innovation and ensure human dignity and safety
Conspirador Norteño • 24 implied HN points • 17 Jan 26
  1. A coordinated spam network of at least 31 newly created Bluesky accounts posted and amplified identical T‑shirt ads.
  2. The accounts all use female first/last name formats, follow users in large batches, and mostly rally around one low‑activity account that also promotes the same shirts.
  3. They recycle just two plagiarized profile photos and run on an external host (myatproto.social), making them clearly inauthentic astroturfing rather than real users.
Musings on Markets • 839 implied HN points • 06 May 23
  1. The recent banking crisis in the US started with the sudden collapse of Silicon Valley Bank, which was unexpected and quick, leading to a series of other bank failures. This raises concerns about whether the crisis will impact the larger economy like the 2008 crisis did.
  2. Understanding what makes a bank strong or weak is essential. Key factors include how sticky their deposits are, the amount of equity they have to absorb losses, and the quality of their loan portfolios.
  3. The market's reaction to the crisis has been mixed, with larger banks holding up better than regional ones. However, the crisis may push more consolidation within the banking industry, affecting competition and overall bank profitability.
David Friedman’s Substack • 278 implied HN points • 05 Jul 25
  1. Corporations are supposed to benefit stockholders but often face challenges in making sure that happens. Stockholders can vote out management if they're unhappy, but individual votes often feel insignificant to them.
  2. Stakeholders like customers and employees are already protected by market forces. If they don't like the products or the job conditions, they can easily switch to alternatives without needing special voting rights.
  3. Stockholders face unique risks because their investments can be tied up even if the company isn't performing well. Strengthening stockholder control and removing barriers to takeover bids could help ensure companies focus more on maximizing stockholder value.
Dev Interrupted • 28 implied HN points • 06 Jan 26
  1. Standardizing build and deployment pipelines and automating SRE tasks removes repetitive work so large engineering teams can move like startups and focus on high‑value problems.
  2. AI in 2026 shifts from demos to real procurement: organizations will budget heavily for AI and should prioritize applying models to new workflows while enforcing strong security and governance.
  3. Pausing deploys (like Friday freezes) often increases risk by accumulating untested changes; regular, practiced deployments build resilience and reduce surprise failures.
OSS.fund Newsletter • 37 implied HN points • 01 Jan 26
  1. Human agents are still essential as the safety and empathy layer alongside AI, so companies must design and budget for hybrid human+AI workflows with clear escalation and QA paths.
  2. Enterprise buying now demands predictable, governable pricing and clear unit economics, pushing vendors toward outcome- or unit-based costing and hybrid seat/credit models that finance can forecast and control.
  3. The real enterprise risk and competitive moat is in orchestration, connectors, and governance — permissions, logging, and blast-radius controls (plus compliance posture and multi-model routing) are becoming hard buying criteria.
Erifili • 219 implied HN points • 22 Feb 24
  1. Yutori is a concept of intentional slowing down in life to appreciate the present moments and find contentment.
  2. Having a bias for action means favoring action over inaction, which can be a key trait for success in entrepreneurship.
  3. Learning to make decisions intuitively based on the heart rather than the ego can lead to a more fulfilling life.
DeFi Education • 659 implied HN points • 26 Jul 23
  1. Crypto faces a big problem with not having enough users. Many people find it really complex to use, which makes them give up before they even start.
  2. There's also the issue of not having enough attractive things to do with crypto. Many projects don't solve real problems, leading to confusion for new users about what to do with it.
  3. Finally, scalability and regulation are major hurdles. Current blockchain networks have trouble handling lots of users without high fees, and regulations are changing, creating a lot of uncertainty.
DeFi Education • 559 implied HN points • 01 Sep 23
  1. Grayscale recently won a court case that could lead to the approval of a Bitcoin ETF. This is important because it could help investors get better prices for their Bitcoin investments.
  2. A court ruled that Uniswap, a decentralized exchange, is not responsible for scams involving tokens traded on its platform. This is good news for developers in the DeFi space, as it shows they can’t be held liable for the actions of others.
  3. The SEC has confirmed that some NFTs can be considered securities, as seen in a case where a company settled for not registering its NFT offering. This emphasizes that just because something is digital doesn't mean it escapes regulation.
Artificial Ignorance • 58 implied HN points • 05 Dec 25
  1. OpenAI is changing its focus back to improving ChatGPT, stepping away from other projects like ads and personal assistants due to rising competition with companies like Google.
  2. Anthropic is planning to go public and has made significant strides in revenue and product offerings despite facing substantial losses, aiming to challenge the big tech firms.
  3. Three years after the launch of ChatGPT, American opinions about AI are mixed, with some people excited and others fearful, even as AI continues to change industries like education and finance.
The Future Does Not Fit In The Containers Of The Past • 60 implied HN points • 23 Nov 25
  1. Change in companies happens through people, not just technology. It’s important to focus on growing and developing the individuals within an organization.
  2. To transform a company, you can either help current employees change or bring in new ones with fresh perspectives. Success often comes from a mix of both.
  3. Six key skills are important in today's work: Think critically, be creative, stay curious, collaborate well, communicate clearly, and convince others. These are essential for thriving in a fast-changing environment.
Import AI • 499 implied HN points • 18 Sep 23
  1. Adept has released an impressive small AI model that performs exceptionally well and is optimized for accessibility on various devices.
  2. AI pioneer Richard Sutton suggests the idea of 'AI Succession', where machines could surpass humans in driving progress forward, emphasizing the need for careful navigation of AI development.
  3. A drone controlled by an autonomous AI system defeated human pilots in a challenging race, showcasing advancements in real-world reinforcement learning capabilities.
The Bear Cave • 676 implied HN points • 05 Jan 25
  1. Hims & Hers Health could be in trouble if the FDA removes its shortage label on semaglutide, which would stop them from selling compounded versions of the drug.
  2. Recent reports revealed Carvana's questionable dealings related to $800 million in loan sales and a possible SEC investigation, leading to a significant drop in their stock.
  3. There have been several high-profile executive resignations recently, indicating potential instability in companies like Integral Ad Science and Aurora Innovation.
Comment is Freed • 60 implied HN points • 01 Dec 25
  1. Designing choices and defaults works big time: changing systems like automatic enrollment in pensions can produce huge, lasting effects, while simple wording changes and social comparisons give smaller but very cost-effective boosts.
  2. There are big practical and political limits to nudging: nudge teams often can only persuade rather than redesign systems, and deliberate "sludge" or gamified interfaces can harm people while current political trends make evidence-based reforms harder.
  3. Behavioural economics has uncovered many real anomalies and useful tools (like mental accounting), but it hasn’t replaced standard economic theory or textbooks and probably won’t offer a single grand theory; its strength is adding realistic, descriptive insights to existing models.
Venture Prose • 738 implied HN points • 15 Jan 23
  1. Investors in venture capital should not shy away from addressing sensitive topics with the founders they support, to avoid miscommunication and potential issues.
  2. Communication is key in maintaining productive relationships between investors and entrepreneurs. Creating an environment where feedback is welcomed and honest conversations can take place is crucial.
  3. Investors need to share observations directly with entrepreneurs, rather than gossiping or keeping concerns to themselves. Open dialogue and constructive conflict can lead to better outcomes for all parties involved.