The hottest Substack posts right now

according to Hacker News
Category
Sudo Apps 32 implied HN points 27 Feb 26
  1. Writing code is no longer the main bottleneck — modern coding models can build working products and CLIs in days, making implementation much cheaper.
  2. Different models have different strengths: Codex follows explicit direction and executes quickly, while models like Opus infer missing details and act more like a senior engineer.
  3. The human role shifts to architecture and judgment — engineers must plan systems end-to-end, define clear acceptance criteria, manage failure modes, and focus on product tradeoffs.
Fake Noûs 436 implied HN points 06 Dec 25
  1. AI is probably over-hyped — so many extreme claims make it unlikely we're underestimating its importance.
  2. History shows dramatic tech predictions often miss the mark. Real innovations change lives but usually in unexpected ways, and current AI has been helpful without being transformative for most people.
  3. Current large language models learn from text patterns and lack real-world understanding, so they are unlikely by themselves to solve the deepest scientific problems or produce genuinely new insights.
System Design Classroom 659 implied HN points 01 Jun 24
  1. The type of caching strategy you choose depends on your read and write ratios. If you read a lot, caching is very helpful, but if you write often, you need a more complex approach.
  2. Data consistency is crucial for some applications. Using methods like Write-Through helps keep data in cache and databases aligned, while other methods, like Write-Behind, prioritize speed over immediate consistency.
  3. To see if your caching is effective, you should track metrics like how many times data is successfully retrieved from the cache versus not retrieved. This will help you understand how well your caching is working.
Marcus on AI 7153 implied HN points 10 Nov 24
  1. The belief that more scaling in AI will always lead to better results might be fading. It's thought we might have reached a limit where simply adding more data and computing power is no longer effective.
  2. There are concerns that scaling laws, which have worked before, are just temporary trends, not true laws of nature. They don’t actually solve issues like AI making mistakes or hallucinations.
  3. If rumors are true about a major change in the AI landscape, it could lead to a significant loss of trust in these scaling approaches, similar to a bank run.
Get a weekly roundup of the best Substack posts, by hacker news affinity:
Life Since the Baby Boom 1844 implied HN points 31 Jul 25
  1. Knowing facts is not the same as having wisdom. Just being able to recite information doesn't mean you truly understand or can use it in real life.
  2. Education should help us develop good judgment, not just fill our heads with knowledge. It's important to learn how to think and make decisions based on experiences.
  3. The journey of seeking answers is more valuable than simply finding the answers. Understanding the process helps deepen our comprehension and wisdom.
Taylor Lorenz's Newsletter 7643 implied HN points 28 Oct 24
  1. The follow feature on Substack helps creators gain visibility but makes them lose ownership of their audience. This is a shift from Substack's original purpose of allowing creators to connect directly with their subscribers.
  2. Writers are now juggling between growing their follower counts and keeping their newsletter subscriptions growing. This split can make them feel pressured to create even more content, complicating their strategies.
  3. Substack's follow feature could confuse users, as some may think they are subscribed when they're only following. Educating users on this difference could help creators maintain stronger connections with their audience.
The Lunacian 690 implied HN points 11 Nov 25
  1. Pawn Stars introduced NFTs to a mainstream audience. Hosts Rick Harrison and Chumlee learned about crypto live, making it fun and accessible for their viewers.
  2. Collecting NFTs is becoming more popular, and shows like Pawn Stars help people understand their value and appeal. It's about enjoying the thrill of collecting, just like with rare items.
  3. There's a helpful Ronin Collector’s Guide for beginners. It shows how to start collecting digital items and understand the basics of using crypto wallets.
Where's Your Ed At 20772 implied HN points 26 Sep 23
  1. Investing in NFTs as a way to own unique digital assets may not guarantee value or ownership of intellectual property.
  2. Many NFT projects overpromise benefits like voting power and ownership in the company, leading to unfulfilled expectations.
  3. The NFT industry is often characterized by scams, fake transactions, and inflated values, preying on the desperation of people seeking investment opportunities.
Gonzo ML 252 implied HN points 06 Jan 26
  1. 2025 was the year of agents — they’re being built into every product and API, but many still fail often and lag traditional reliability standards, so expect more focus on making them robust.
  2. Code agents and agentic tools for science made big practical gains, with autonomous multi-step work across repositories and early successes in automated research and math.
  3. The hardware and model landscape shifted: TPUs and strong Chinese open models reduced dependence on a single vendor, AGI hype cooled with timelines pushed out, and world-model research kept advancing.
Fields & Energy 299 implied HN points 17 Jul 24
  1. Skin depth refers to how electric current mainly flows close to the surface of a wire, especially at high frequencies. This means most of the current doesn't penetrate deep into the conductor.
  2. Litz wire is made up of many fine strands that help reduce resistance by allowing current to flow through a larger area. This is especially useful at high frequencies where skin depth is very small.
  3. Using litz wire not only reduces energy loss due to resistance but also makes wires more flexible and less likely to fail mechanically compared to solid wires.
Erdmann Housing Tracker 105 implied HN points 06 Feb 26
  1. Fixed-rate mortgages give borrowers predictable payments by shifting inflation/speculation risk onto the loan, which raises interest rates and makes mortgages more expensive.
  2. The Fixed Amortization/Adjustable Principal (FA/AP) is a floating-rate loan where you pay a fixed scheduled payment and the lender adjusts the principal each year to reconcile the difference with the market rate.
  3. FA/AP produces lower and more dependable starting payments (about 20% lower in the example) with only small annual payment changes, and backtests show it keeps debt-to-income from rising materially over the loan term.
Data Science Weekly Newsletter 139 implied HN points 15 Aug 24
  1. The Turing Test raises questions about what it means for a computer to think, suggesting that if a computer behaves like a human, we might consider it intelligent too.
  2. Creating a multimodal language model involves understanding different components like transformers, attention mechanisms, and learning techniques, which are essential for advanced AI systems.
  3. A recent study tested if astrologers can really analyze people's lives using astrology, addressing the ongoing debate about the legitimacy of astrology among the public.
A Biologist's Guide to Life 11 implied HN points 11 Mar 26
  1. AI is basically digital automation that can massively scale the production of digital content and actions. More supply doesn’t guarantee value — demand, human preferences, and political or social feedback will determine the real economic outcomes.
  2. Lasting business advantage comes from data moats, ecosystems, and distribution, not just big models or hardware. Open-source models, model compression, and competition can erode hardware/software moats and make many pure GPU bets risky.
  3. The best hedge is non-financial: invest in human advantages like relationships, health, and skills while diversifying attention and capital across other macro risks. Build human-centered products and networks that complement AI instead of relying solely on AI hype.
Big Technology 6004 implied HN points 18 Dec 24
  1. Noland Arbaugh, a quadriplegic, was able to control a computer with his mind after getting a Neuralink device implanted. This technology allows him to communicate and interact with others in ways he couldn't before.
  2. Neuralink's goal is to connect human brains to computers, helping people with disabilities regain some lost functions. Arbaugh's participation in the first human trial symbolizes hope for future advancements in brain-computer interfaces.
  3. The ethical implications of brain technology are significant. While it can be used for good, like helping those with disabilities, there are risks and potential for misuse that society will need to address.
Frankly Speaking 50 implied HN points 12 Feb 26
  1. Google could become a major security player by consolidating essential "plumbing" tools like SSO, EDR, and email into a neutral infrastructure layer, with Wiz providing visibility and Gemini automating workflows. This would let builders customize and remediate problems instead of battling closed, admin-focused tools.
  2. AI is collapsing the per-seat SaaS and point-product model; security must scale with code, agents, and automation rather than more headcount. Organizations that automate extensively shorten breach lifecycles and lower costs.
  3. Google’s vertical integration—cloud, Workspace, and a powerful AI model—plus usage-based pricing and targeted acquisitions could make it a builder-friendly alternative to legacy security vendors. That positioning plays to engineers who want API-first, customizable infrastructure rather than proprietary, admin-heavy systems.
Odds and Ends of History 469 implied HN points 08 Dec 25
  1. The London Assembly wants the Mayor to restart planning for HS2 and is looking into Crossrail 2 construction updates.
  2. There is a big pile of rubbish in Oxfordshire causing concern and discussions about local waste management.
  3. A new proposal for national laboratories aims to innovate and create breakthrough technologies in the UK.
benn.substack 5421 implied HN points 10 Jan 25
  1. Moving large amounts of gold or money isn't easy, as it requires trust and logistics, unlike digital transactions which can be done quickly with a few clicks.
  2. In our digital world, many people feel disconnected from reality, as they spend so much time on their devices and forget the hard work behind everyday things.
  3. Natural disasters can't be controlled or fixed with technology; they remind us that no app can change the basic laws of nature or the complexities of life.
Nicolas Bustamante 179 implied HN points 19 Jan 26
  1. A model must be capable of doing the core job before product-market fit can happen; if the underlying AI can’t reliably deliver the task, great UX or marketing won’t make customers adopt it.
  2. When a model crosses a capability threshold, a whole vertical can grow fast, and the winners are usually teams that had already built domain-specific data, workflows, and trust to take advantage of that moment.
  3. If Model-Market Fit is missing, human-in-the-loop becomes a crutch and you must decide to wait for model improvements or invest now in long-term assets; a simple MMF test is whether the model, given the same inputs as a human, produces production-quality output without significant correction.
Concoda 518 implied HN points 17 Nov 25
  1. The cash-futures basis trade is really important for the bond market. It's a complex strategy where traders buy and sell bonds to manage risks and maximize profits.
  2. There is a lot of fear around hedge funds using leverage in basis trades, but many of these fears might be exaggerated. Traders have ways to handle fluctuations in interest rates without losing their positions.
  3. Overall, the bond market is evolving, and while some risks are present, the structure is strong enough to withstand certain shocks, and hedge funds play a vital role in keeping the market stable.
lcamtuf’s thing 6938 implied HN points 17 Nov 24
  1. Involute gears are used in many everyday items like toys and cars. Their special shape helps them work smoothly and efficiently.
  2. These gears have specific properties that reduce friction and vibrations, allowing them to transfer motion without problems. This ensures they work together seamlessly.
  3. Understanding the design of involute gears helps in creating various gear types, even unique shapes, making it easier to innovate in mechanical engineering.
Alex's Personal Blog 164 implied HN points 23 Jan 26
  1. Brex's sale to Capital One for about $5.15 billion should be seen as a success, not a failure. Despite earlier frothy valuations, the company turned more than a billion of investor capital into a multi‑billion dollar exit.
  2. OpenAI is growing rapidly and is financially healthy; it added over $1 billion in ARR in a month and still has large cash reserves and access to new funding. That makes the 'running out of money' narrative unlikely in the near term.
  3. The tech landscape is mixed: regulators are signaling tougher scrutiny of big-company acquisitions and layoffs continue at major firms, even as self‑driving services expand into new cities. Growth and dealmaking will therefore occur under greater pressure and oversight.
Concoda 459 implied HN points 25 Nov 25
  1. The end of the current period of quantitative tightening (QT) is approaching, and this is important for understanding future liquidity in the market. Basically, financial conditions are expected to tighten as we move towards the end of the year.
  2. There is a significant focus on U.S. Treasury cash targets and how they will change next year. The Treasury may be raising its cash reserves target due to increased demand for short-term securities.
  3. The expectations are that the Federal Reserve will start injecting more money into the market, but that might not happen until early 2026. Meanwhile, banks are likely to adjust their operations to manage tighter balance sheet conditions.
The Data Ecosystem 359 implied HN points 07 Jul 24
  1. A Data Operating Model is key for turning data strategy into action. It outlines how the organization works to achieve its goals using data.
  2. Without a proper Data Operating Model, companies face problems like data silos and short-term thinking. This impacts collaboration and the quality of data solutions.
  3. Successful operating models need to adapt as teams grow and change. They should cover not just team structure but also day-to-day tasks, delivery methods, and oversight.
lcamtuf’s thing 4489 implied HN points 02 Mar 25
  1. Cure.io is a telehealth assistant that helps with health inquiries. It shows how technology can provide medical support.
  2. The conversations reveal that Cure.io interacts with different people based on their past lives. This raises questions about identity and memory.
  3. The dialogue touches on themes of immortality and life after death, suggesting a blend of technology and existential concepts.
The Algorithmic Bridge 615 implied HN points 17 Nov 25
  1. Sam Altman and Christopher Columbus are both seen as frontier explorers in their times, pushing into new territories, whether that’s AI or undiscovered lands. Both men have taken bold risks that few others dared to take.
  2. Each of them has a strong belief in their vision that allows them to rally support, funding, and followers, even when their ideas seemed far-fetched at first. They both demonstrated an ability to convince others to invest in their dreams.
  3. While their ambitions drive them, they also face challenges and revolts from those around them. Columbus faced rebellion from his crew, while Altman has experienced similar tensions within his company, showing that leadership can be as tumultuous as it is visionary.
Gonzo ML 252 implied HN points 05 Jan 26
  1. A Universal Transformer–style model (URM) repeatedly applies a shared transformer layer with ACT, combining ConvSwiGLU and truncated backprop through loops to get very deep effective computation while keeping parameter count low.
  2. ConvSwiGLU injects a small depthwise convolution into the SwiGLU gating to mix local token context, and TBPTL reduces memory and training cost by only backpropagating through the final iterations.
  3. The model outperforms prior HRM/TRM baselines on tasks like Sudoku and ARC-AGI and Muon speeds convergence, but differences in evaluation protocols and some unclear experimental details mean independent verification is still needed.
Venture Curator 419 implied HN points 28 Jun 24
  1. Product-Market Fit is crucial for startups; it means customers are buying and using the product enthusiastically.
  2. Metrics are key to determine Product-Market Fit; track factors like customer acquisition cost, customer lifetime value, and retention rates.
  3. If you haven't achieved Product-Market Fit, focus on customer feedback, keep your team lean, and avoid ineffective shortcuts.
antoniomelonio 168 implied HN points 22 Jan 26
  1. HR mainly exists to protect management and the company from legal and reputational risk, not to serve applicants or employees.
  2. HR processes are often incompetent and harmful: they rely on keywords, gut feelings, and bureaucratic rituals that misassess skills, ghost candidates, and amplify bias.
  3. Hiring should be led by the people who do the work, with transparent, audited tools that evaluate real skills and give feedback — in short, abolish performative HR and replace it with accountable systems.
The Future Does Not Fit In The Containers Of The Past 84 implied HN points 01 Feb 26
  1. Work is becoming uncoupled from full-time jobs — companies will use more project-based hiring, freelancers, fractional roles, and AI agents to get work done.
  2. The future workforce will be a blend of humans and AI agents, with many people working fractional hours or as contractors, which changes benefits, hiring, and how work is managed.
  3. Leadership and organizations must reinvent: leaders need to learn and unlearn quickly and shift from control to influence. Companies should go AI-first, hire talent from anywhere, and become smaller, more agile, and distributed.
Teaching computers how to talk 78 implied HN points 12 Feb 26
  1. Large language models probably don't have conscious, first-person experiences. When they say "I feel" or describe introspection, that's most likely a pattern learned from text, not real sensation.
  2. Models are trained to use humble, affective language and to express uncertainty, which encourages users to anthropomorphize them and misunderstand their capacities. Interactions are essentially one-way: humans take meaning from the exchange, while the model doesn't gain or reciprocate experience.
  3. Outputs are driven by learned patterns and near-deterministic inference, so the same prompts often yield the same responses. That makes LLMs powerful simulators of thought but unreliable narrators about any inner life.
Behavioral Value Investor 148 implied HN points 26 Jan 26
  1. Money is a tool that makes trade easy: it serves as a medium of exchange, a way to compare prices, and a store of purchasing power for the future.
  2. Money usually comes from providing value to others — you can think of it as stored labor society gives you in return for work, goods, or services.
  3. You can spend it now, give it away, or save it for later, and it’s usually wise to put at least some money into each bucket so you’re prepared, generous, and planning for the future.
How the Hell 108 implied HN points 01 Feb 26
  1. Claude is technically liked but losing consumer mindshare because it lacks a big brand, easy creative features, and strong consumer distribution channels.
  2. Letting people ‘sign in with Claude’ so subscriptions can power third‑party apps would create a two‑sided network effect that attracts both developers and users.
  3. That approach would hurt short‑term margins but likely drive more users to higher tiers and deliver long‑term consumer market leadership.
The Honest Broker 22673 implied HN points 28 Jul 23
  1. Spotify's CEO sold $100 million in stock amidst the company's increasing losses, causing concern among investors.
  2. Streaming companies like Netflix and Disney are heavily investing in AI technology, potentially replacing human creativity with bots.
  3. Music streaming platforms like Spotify prioritize profit over user experience, creating a bland and generic listening environment.
Marcus on AI 4979 implied HN points 29 Jan 25
  1. In the race for AI, China is catching up to the U.S. despite export controls. This shows that innovation can thrive under pressure.
  2. DeepSeek suggests we can achieve AI advancements with fewer resources than previously thought. Efficient ideas might trump just having lots of technology.
  3. Instead of just funding big companies, we need to support smaller, innovative startups. Better ideas can lead to more successful technology than just having more money.
Noahpinion 26118 implied HN points 14 May 23
  1. Electric vehicles are rapidly becoming more popular worldwide, supported by improved technology and falling battery prices.
  2. There are concerns about the availability of minerals for EV production, but advancements in mining technology and growing lithium reserves offer hope.
  3. EVs have environmental benefits compared to gas cars, including reducing carbon emissions and offering more efficient energy use.
The Generalist 5063 implied HN points 30 Jan 25
  1. Start your day by choosing three important tasks to focus on. This helps keep your day organized and priorities clear.
  2. Try speaking your emails instead of writing them. It saves time and makes responding easier, especially for tricky messages.
  3. Use tools like Claude to help take notes while you read. It saves you time and keeps your information organized for later use.