ai Enterprise AI, open-source policy, and cybersecurity evals
Samsung Electronics brings ChatGPT and Codex to employees
Samsung Electronics deployed ChatGPT Enterprise and Codex to employees worldwide, making it one of OpenAI's largest enterprise rollouts to date. The deployment gives Samsung staff access to both the conversational assistant and the code-generation tool across its global workforce. OpenAI has been targeting large enterprises as a revenue stabilizer alongside its consumer and API businesses.
Patterns for Building Cybersecurity Evals
Eugene Yan described a pattern for building cybersecurity evals built on four components: a sandboxed target, inputs that control task difficulty, tools available to the model, and a grader that scores outcomes. The approach treats security evaluation as a structured benchmark problem rather than an ad hoc red-teaming exercise. Yan's framework gives teams a repeatable way to measure whether a model can complete offensive or defensive security tasks at defined difficulty levels.
Banning Open Source AI Would Be A Mistake
Nathan Lambert and Kevin Xu co-authored an op-ed arguing that banning open-source AI would be a strategic mistake for the United States. They contend that open weights create compounding public benefits: fine-tunes, evals, and tooling accumulate across the research community, lowering the marginal cost of each subsequent improvement. They cited China's high-participation release culture as evidence that restricting open weights would disadvantage the US without meaningfully slowing adversaries.
Identity verification on Claude
Anthropic introduced identity verification on Claude, requiring users to confirm their identity in certain contexts. The move generated 803 points and 669 comments on Hacker News, making it the most-discussed item in the current window. The policy change raises questions about how verification interacts with Claude's existing usage policies and what it signals about the platform's direction on accountability.
There is minimal downside to switching to open models
A post titled "There is minimal downside to switching to open models" argued that closed frontier models no longer justify their cost premium for most production workloads. The author contended that open-weights alternatives have reached a quality threshold where switching costs are low and the risks of vendor dependency outweigh performance differences. The post reached 275 points on Hacker News with 227 comments.
Temporary Cloudflare Accounts for AI agents
Cloudflare launched Temporary Accounts for Workers, letting any AI agent run wrangler deploy and get a live deployment in seconds without a persistent account. The feature removes a friction point that had blocked agents from autonomous deployment: they previously needed human-controlled credentials to interact with Cloudflare's infrastructure. Simon Willison noted the capability is broadly useful beyond the AI agent framing.
[AINews] GLM > GPT? GLM-5.2 passes vibe check; Z.ai forecasts Open Fable by December
Latent Space reported that GLM-5.2 passed community vibe checks as a genuine frontier open model, and that Z.ai forecast an open equivalent of Claude Fable by December. The coverage framed open-model quality as finally reaching a level where it competes with proprietary frontier releases rather than lagging behind them. The MIT-licensed GLM-5.2 was the primary evidence cited.
Patterns for Building Cybersecurity Evals
Eugene Yan documented patterns for building cybersecurity evals, covering sandboxed targets, variable task difficulty, tool use, and grading mechanisms. The framework addresses how to measure AI system performance on adversarial security tasks where the difficulty varies by scenario.
sqlite-utils 4.0rc1 adds migrations and nested transactions
sqlite-utils 4.0rc1 introduces migrations and nested transactions, expanding the library's capabilities for database schema management and transaction control. The release brings higher-level operations that developers previously had to implement manually on top of the Python sqlite3 module.
Temporary Cloudflare Accounts for AI agents
Cloudflare launched Temporary Accounts for AI agents, allowing agents to call wrangler deploy and get live Workers in seconds without human credential setup. The feature solves a deployment barrier for agentic workflows by creating ephemeral, time-limited access credentials.
Chinese OPEN SOURCE model good as FABLE
Mo Bitar reported that a Chinese open-source model outperforms Claude Fable on several benchmarks. The comparison underscores rapid progress in open-weights development outside the US and competitive pressure from non-US labs.
Quoting Sean Lynch
Sean Lynch argued that Model Context Protocol's core value is isolating auth flows outside an agent's context window, potentially outside the harness altogether. MCP's idealized form may be primarily an auth gateway for APIs rather than a skills or CLI abstraction layer.
Claude Fable Blocked - 11 Quiet Details on What's Next
An AI Explained video documented 11 details about the Fable 5 blocking incident and what comes next for Claude deployment. The piece examined policy implications and how the export control dispute shapes future AI release strategy.
Scientists Found A Better Language For AI Agents
Two Minute Papers covered research on agent languages, examining work that identifies better ways for systems to communicate task plans. The research suggests that how agents represent intention to themselves affects their ability to execute reliably.
Elon's AI won.
Mo Bitar examined strategic moves by Elon Musk's AI efforts, covering xAI's positioning and competitive dynamics against OpenAI and other labs. The video assessed recent announcements and what they signal about resource allocation and market ambitions.
Datasette Apps: Host custom HTML applications inside Datasette
Simon Willison documented Datasette Apps, a plugin allowing developers to host custom HTML applications directly inside Datasette. The feature lets teams embed interactive interfaces and dashboards without a separate server, reducing deployment overhead.
datasette-acl 0.6a0
datasette-acl 0.6a0 expanded the access control plugin toward a general resource-sharing system. The release fleshes out multi-user data sharing scenarios and permission delegation, moving beyond table-level controls.
New usage analytics and updated spend controls for enterprises
OpenAI announced new spend controls and usage analytics for ChatGPT Enterprise, helping organizations monitor costs and understand consumption patterns. The tools address a key pain point for enterprises scaling AI deployments across teams.
[ECCV 2026] Paper Decision Appeals Discussion [D]
ECCV 2026 opened a formal appeals process for dissatisfied authors, allowing them to challenge decisions on grounds including policy errors or reviewer misconduct. The meta-review stage created a new checkpoint in peer review governance.
An Update on Matrix Recurrent Units, an Attention Alternative [R]
A researcher updated the Matrix Recurrent Units algorithm as an alternative to attention, documenting improvements and new results. The MRU architecture operates in linear time on sequence length, addressing the quadratic scaling of standard transformers.
Data-centric debugging for teams training neural nets [P]
WeightsLab launched a major revamp focused on data-centric debugging for neural network training. The tool lets teams pause training mid-run to inspect live data, identify distribution shift, and catch bugs early without rebuilding from scratch.
Best current methods for finetuning whisper on domain specific vocabulary? [P]
A researcher asked about best practices for fine-tuning Whisper on domain-specific speech with technical terminology. The thread surfaced current approaches to vocabulary adaptation and data efficiency in speech model customization.
EMA on LoRA ? [R]
A researcher asked whether exponential moving average applied to LoRA adapters has been explored for self-teaching, where an EMA adapter generates soft labels for a trainable adapter. The thread examined whether on-policy distillation patterns have precedent in LoRA literature.
A slightly improved DVD-JEPA demo [P]
A developer published a minimal DVD-JEPA demo with improvements including environment noise simulation and fairer comparison to pixel-based baselines. The work demonstrated joint embedding predictive architectures in a clearer setting than prior implementations.
[Exclusive] $250 off AI Engineer tix til Monday
Latent Space offered a subscriber-only discount on AI Engineer conference tickets, dropping the price by $250 through Monday. The event covers infrastructure, agents, and AI system design.
I released a softmax-free attention model at GPT-2 Medium scale (~354M params, 11.5B tokens): structural sparsity + tile-skipping kernels for long-context VRAM savings. Open weights + custom Triton kernels [R]
A developer released a softmax-free attention model at GPT-2 Medium scale using structural sparsity and tile-skipping kernels. The work demonstrates VRAM and compute savings for long-context inference through custom Triton kernels and open weights.
Feature Pyramid Networks - Explained!
CodeEmporium explained Feature Pyramid Networks, covering how the architecture enables multi-scale feature extraction. The video traced FPN's application across detection, segmentation, and object recognition tasks.
Voice for AI Agents and Applications
DeepLearningAI covered voice capabilities for AI agents and applications, examining how speech interfaces are changing agent interaction patterns. The video documented recent advances in voice quality and latency.
Python packages for particle swarms, genetic algorithms. Scikit-opt maybe? [D]
A practitioner asked about Python packages for particle swarms and genetic algorithms to improve curve-fitting optimization. The thread compared scikit-opt and other metaheuristic libraries against gradient-based alternatives.
Deno Desktop
Deno launched Deno Desktop, expanding the runtime beyond server environments into desktop application development. The launch reached 595 points on Hacker News with 219 comments, signaling developer interest in cross-platform JavaScript runtimes.
Did my old job only exist because of fraud?
A developer reflected on discovering their previous job may have existed primarily to sustain internal fraud, raising questions about organizational waste and accountability. The post reached 647 points on Hacker News and prompted widespread discussion of corporate dysfunction.
Apertus; Open Foundation Model for Sovereign AI
Apertus introduced an open foundation model for sovereign AI, positioning itself as an alternative to US-controlled or Chinese-controlled frontier models. The project reached 446 points on Hacker News, reflecting interest in geopolitically independent AI development.
Everything is logarithms
Alex Kritchevsky argued that logarithmic scaling relationships underlie many phenomena in machine learning and beyond. The post reached 245 points on Hacker News and examined how log relationships simplify seemingly unrelated observations across fields.
JSON-LD explained for personal websites
A developer explained JSON-LD for personal websites, covering how structured data markup improves SEO and machine readability. The post reached 234 points on Hacker News with 75 comments.
The most trusted code on Earth is being rewritten in Rust
Fireship covered the major Rust rewrite efforts in critical infrastructure software, examining why memory safety has become a priority for systems that run the internet. The video documented concrete projects and their security outcomes.
Help I accidentally a wigglegram
A developer accidentally created a wigglegram and documented the process. The post reached 331 points on Hacker News, sparking discussion of stereoscopic image techniques and creative accidents in image manipulation.
Temporary Cloudflare Accounts for AI agents
Cloudflare rolled out Temporary Accounts on Workers, allowing agents to deploy code without human credential provisioning. The feature addresses a deployment friction point by creating short-lived, agent-specific access tokens.
Build your own vulnerability harness
Cloudflare documented the technical architecture of a multi-stage vulnerability discovery harness and automated triage system. The post explained state management, false positive squashing through adversarial review, and routing around LLM context limits.
Celebrating 12 years of Project Galileo
Cloudflare released the first comprehensive report analyzing cyberattacks against civil society, marking the 12th anniversary of Project Galileo. The report covers attack patterns, threat actors, and trends in targeting vulnerable organizations.
Building Agents that Don't Break Themselves
Fly.io documented patterns for building agents that isolate risky operations in Sprite sandboxes. The approach prevents agents from corrupting their own state or the host environment when they attempt destructive or self-modifying tasks.
Sakana Fugu Ultra now available on AI Gateway
Sakana Fugu Ultra is now available on Vercel's AI Gateway, expanding routing options for models deployed through the platform. The addition broadens the gateway's coverage of open-weights and proprietary frontier models.
software Deno goes to the desktop, Cloudflare opens agent deployment
Deno Desktop
Deno released Deno Desktop, a native desktop runtime built on the same JavaScript and TypeScript foundation as the server-side runtime. The project reached 595 points on Hacker News with 219 comments. The release extends Deno's target beyond server and edge workloads into local application development, putting it in competition with Electron and Tauri for cross-platform desktop apps built with web technologies.
The Pulse: Big implications of US banning Anthropic's new model, Fable
The Pragmatic Engineer covered the US government's blocking of Anthropic's Fable model and its implications for software engineering teams, noting the episode forces organizations to evaluate dependency on any single frontier model provider. The issue also covered the SpaceX IPO, SpaceX's acquisition of Cursor, and Cursor's GitHub competitor. The combination of export control risk and consolidation activity in developer tools is reshaping how engineering teams think about their AI stack.
Datasette Apps: Host custom HTML applications inside Datasette
Simon Willison launched Datasette Apps, a plugin that lets developers host custom HTML applications directly inside Datasette instances. The feature allows single-file HTML apps to be embedded and served within the Datasette interface, treating the database tool as a lightweight application host. Willison described it as expanding Datasette from a data-exploration tool toward a platform for building data-driven interfaces.
sqlite-utils 4.0rc1 adds migrations and nested transactions
Simon Willison released sqlite-utils 4.0rc1, adding schema migrations and nested transactions to the combined Python library and CLI tool for SQLite. Migrations allow users to evolve database schemas without manual SQL; nested transactions enable finer-grained rollback control within complex operations. The release candidate marks the first major version change since the tool's initial release and brings it closer to production-grade database management functionality.
Build your own vulnerability harness
Cloudflare published the architecture behind its multi-stage vulnerability discovery harness and automated triage loop. The system manages state across multiple LLM calls, uses adversarial review to reduce false positives, and routes around LLM context limits by chunking the analysis pipeline. The post gives practitioners a concrete reference for building AI-assisted security tooling on top of existing infrastructure.
Bridge Demolition is Complicated
Practical Engineering covered the complexity of controlled bridge demolition, examining engineering considerations in deconstructing large structures safely. The video detailed planning, sequencing, and risk management in urban demolition projects.
pharma AbbVie's $11B deal, an LSD Phase 3 win, and an FDA reversal
STAT+: AbbVie to buy Apogee Therapeutics in nearly $11B deal
AbbVie agreed to acquire Apogee Therapeutics for nearly $11 billion, adding Apogee's experimental atopic dermatitis drug zumilokibart to its immunology pipeline. The deal continues a pharma M&A run that STAT reported has totaled $123 billion in deals in 2026 so far. AbbVie has been aggressive in dealmaking as it seeks to offset revenue exposure from Humira biosimilar competition.
STAT+: Pharma goes on a spending spree, snapping up biotechs in a hurry
Pharma companies have announced $123 billion in biotech acquisitions in 2026, more than in any comparable period in recent years. STAT attributed the surge to companies facing patent cliffs, flush cash balances from GLP-1 revenue, and a buyer's market in biotech valuations following a prolonged downturn. The AbbVie-Apogee deal is the largest single transaction of the current wave.
STAT+: Definium LSD therapy helped patients with major depression in late-stage trial
Definium Therapeutics reported that its LSD-based therapy DT-120 met its primary endpoint in a Phase 3 trial for major depression, the first late-stage psychedelic trial to succeed in this indication. The result moves the drug closer to an FDA submission. Definium's approach differs from psilocybin programs in molecular target and dosing regimen; the company has not yet disclosed the full dataset.
STAT+: FDA reverses course on Regenxbio's childhood gene therapy after rejection
The FDA agreed to reconsider Regenxbio's gene therapy navsunli for Hunter syndrome, a rare and fatal childhood brain disorder, four months after rejecting the application. The agency's reversal is unusual; it typically requires a formal appeal process or new data submission before reconsidering a rejection. Regenxbio did not disclose what new information or procedural argument prompted the FDA to reopen the review.
Health equity researchers fear unseen level of scrutiny under White House proposal
Health equity researchers said a White House proposal that would tighten scrutiny of federally funded research could disqualify much of their work from NIH grants. The proposal would apply new DEI-related review criteria to grant applications; researchers in the field described it as the most serious threat to health disparities research funding they have seen. NIH has not yet implemented the rule.
STAT+: AbbVie to buy Apogee Therapeutics in nearly $11B deal
AbbVie announced an acquisition of Apogee Therapeutics for nearly $11 billion, targeting the developer's experimental atopic dermatitis drug. The deal continues a run of pharma acquisitions as companies seek to expand immunology portfolios.
STAT+: Definium LSD therapy helped patients with major depression in late-stage trial
Definium Therapeutics reported that its LSD therapy helped patients with major depression in a first Phase 3 trial, advancing the psychedelic drug toward potential FDA approval. The result is the first positive late-stage trial for an LSD-based therapy in decades.
STAT+: FDA reverses course on Regenxbio's childhood gene therapy after rejection
The FDA announced it will reconsider approving Regenxbio's experimental gene therapy for Hunter syndrome, reversing a rejection it issued just four months earlier. The decision opens the door to another review of the therapy targeting a rare childhood brain disorder.
Health equity researchers fear unseen level of scrutiny under White House proposal
Health disparities researchers said a new White House proposal could disqualify much of their work from federal funding, representing the most serious threat yet to the future of NIH diversity-oriented science. The rule would restrict funding to narrowly defined health research, affecting programs focused on equity.
STAT+: Pharma goes on a spending spree, snapping up biotechs in a hurry
STAT reported that pharmaceutical companies are acquiring biotechs at an accelerated pace, with 2026 deal volume approaching $123 billion. The spending spree reflects hunger from large pharma to shore up pipelines and acquire emerging technologies before competitors.
Opinion: I work in a psychiatric ER. I'm watching the system fail people in real time
A psychiatrist described the psychiatric emergency room as a last-resort detox facility and respite center for caregivers at their breaking point. The opinion piece documented systemic failure in mental health infrastructure and the patient volume crisis.
Opinion: STAT+: Enough, already: the problem with clinical trial data collection
A researcher noted that nearly 30% of clinical trial data collected never directly inform key trial decisions, yet patients are still asked to provide it. The opinion challenged data collection practices that add patient burden without proportional value.
STAT+: Sanofi names new R&D head as it tries to jump-start pipeline
Sanofi appointed Paulo Fontoura, the chief medical officer at Xaira, as its new head of R&D. The move signals the company's effort to jumpstart its drug pipeline through fresh leadership and integration of AI-driven research approaches.
Opinion: STAT readers debate blue zones, open-access publishing fees, and more
STAT readers debated blue zones, open-access publishing fees, and other topics in a letters-to-editor roundup. The correspondence reflects ongoing discussion about health claims, scientific publishing economics, and reader concerns.
Opinion: Why the space in 'health care' matters
STAT explained why spacing matters in the healthcare terminology debate, with 60% of readers preferring 'health care' over 'healthcare.' The distinction carries implications for how the field understands its scope and mission.
STAT+: Up and down the ladder: The latest comings and goings
STAT covered personnel moves and reorganizations across major pharmaceutical companies, documenting shifts in leadership and organizational structure. The updates included hires, departures, and promotions at Moderna, Pfizer, Novartis, and others.
Federal grant delays could jeopardize essential disability services, research
Disability services researchers worry about job security and institutional futures as federal grant delays continue. STAT documented concerns that prolonged funding uncertainty undermines research continuity and drives talent away from the field.
healthtech Psychiatric ERs under strain, disability grants stalled
Opinion: I work in a psychiatric ER. I'm watching the system fail people in real time
A psychiatrist working in an emergency room described the psychiatric ER as a detox center for patients with nowhere safe to withdraw and a respite center for caregivers at breaking point. The piece documented how the facility absorbs patients who fall through gaps in the outpatient mental health system, housing, and addiction treatment networks. The author argued the current model is unsustainable and shifts costs onto the most acute and expensive care setting.
Federal grant delays could jeopardize essential disability services, research
Disability researchers said federal grant delays stretching into months are putting essential disability services and research programs at risk of collapse. Organizations that rely on federal funding to deliver direct services have been drawing down reserves; some told STAT they will have to shut programs if the backlog is not resolved within weeks. The delays affect grants that were already awarded and approved, not new applications.
STAT+: Sanofi names new R&D head as it tries to jump-start pipeline
Sanofi named Paulo Fontoura, the chief medical officer at AI drug-discovery startup Xaira, as its new head of research and development. Fontoura replaces Houman Ashrafian. The hire signals Sanofi's intent to bring AI-native drug discovery thinking into its R&D leadership at a moment when the company has been trying to accelerate its pipeline after a period of relative underperformance.
SpaceX: The Biggest Money Loser in IPO History
Patrick Boyle examined SpaceX's historical financial losses relative to other IPO debuts, contextualizing the company's billion-dollar burn rate within the aerospace industry. The video examined what makes SpaceX's losses different from typical money-losing IPO candidates.
economy SpaceX humiliates Wall Street, Cuba opens its economy
How SpaceX Humiliated Wall Street
Patrick Boyle examined how SpaceX has consistently outperformed Wall Street's financial models while keeping the company private, denying analysts the data needed to build accurate forecasts. He traced how SpaceX's cost structure, launch cadence, and Starlink revenue trajectory repeatedly invalidated consensus estimates. The video argues that SpaceX's decision to stay private was itself a strategic move that preserved operational flexibility Wall Street scrutiny would have constrained.
How to Lose a Global AI Monopoly in One Afternoon
Patrick Boyle examined how the US lost its position as the dominant force in AI in a single afternoon, tracing the sequence of export control decisions, model releases from Chinese labs, and policy missteps that eroded American advantage. He argued the period between the Fable export restriction and the GLM-5.2 MIT release crystallized a shift that had been building for months.
Cuba
Cuba's Prime Minister Manuel Marrero announced the country's most significant economic reforms in decades, including allowing private and foreign capital to purchase and sell fuel, the creation of private corporate banking, and the end of longstanding subsidies. Tyler Cowen called the measures a genuine opening toward capitalism. The Miami Herald reported the changes; Cowen flagged them as potentially consequential for the Cuban economy's trajectory.
The Transfer Market
Marc Rubinstein examined Wise, the cross-border payments company, tracing how it built a business by attacking the margin embedded in correspondent banking. He described the structural advantage Wise accumulated by building its own network of local accounts in multiple countries, which lets it settle most transfers internally rather than routing through the traditional correspondent chain. The piece ran as a paid issue of Net Interest.
Labor market effects of the Tax Cuts and Jobs Act
New research estimated the labor market effects of the Tax Cuts and Jobs Act of 2017 using post-enactment data rather than pre-reform projections. Tyler Cowen summarized the findings; the paper found measurable employment and wage effects, particularly in sectors most exposed to the corporate rate cut. Earlier analyses based on pre-reform estimates had produced mixed or inconclusive results.
California's Gay Certification Program
Chris Rufo and Austen Hufford examined California's Gay Certification Program, a utility oversight requirement from 1986 mandating CPUC-regulated companies submit annual plans on recruiting and retaining LGBTQ employees. The program reflects how social policy gets embedded in utility regulation.
Scotland facts of the day
Between 2012 and 2022, Catholics in Scotland outnumbered Protestants for the first time since the Reformation. The demographic shift stems from differential birth rates and immigration patterns, reflecting long-term religious change in Europe.
What I've been reading
Tyler Cowen recommended Allison Schrager's book 'Worth the Risk' and a new translation of the Confucian Dialogues. His reading list spans economics, philosophy, and recent scholarship on classical texts.
Pizza wheels are bad, Japanese toilets are great
Noah Smith examined why Americans prefer pizza wheels over Japanese toilets despite the latter being objectively superior. The piece explores signaling, status, and why consumption choices diverge from pure functionality.
Bastiat's telephone?
Oakland saw a 37% decrease in car break-ins over the past year, but the decline has hurt repair shops specializing in window and windshield replacements. The outcome illustrates Bastiat's observation about unseen consequences of policy changes.
Sunday assorted links
Cuba is expanding private banking, creating a foreign exchange market, and ending subsidies in its largest economic reforms in decades. The moves signal a shift toward market mechanisms after decades of centralized control.
Music markets remain deglobalized
Despite being fluent in English and living in a globalized world, Danes listen mainly to homegrown music. In 2019, only five songs in Denmark's top 20 were in Danish; by 2024, that figure had reversed, reflecting deglobalization in cultural consumption.
My aesthetics podcast with Benjamin Lima
Tyler Cowen discussed aesthetics with Benjamin Lima in a podcast conversation, exploring how aesthetic preferences shape consumption, art, and culture. The conversation covered taste formation and the economics of beauty.
Saturday assorted links
Tyler Cowen's Saturday links covered corporate governance structures, elite economist compensation, and AI's role in a Los Angeles museum. The roundup examined governance tradeoffs and incentive structures in organizations.
What defines Japan's national identity?
Hiroko Yoda examined what defines Japan's national identity in a guest post, exploring the intersection of tradition, technology, and global influence. The piece reflects on cultural continuity and change in a major Asian economy.
The Free Press summer reading list
Tyler Cowen nominated a biography of Paul Celan for the Free Press summer reading list, highlighting one of the finest poets of the 20th century. The reading recommendation reflects scholarly interest in Celan's life and work.
Adrian Wooldridge on Sweden and liberalism
Sweden continues to reap rewards from fiscal rectitude and pro-market reforms, with 2026 GDP growth projected at 1.8-1.9%, headline inflation at 1.5%, and debt-to-GDP among the lowest in the world. The model offers lessons in policy coherence and long-term commitment.
Engel's Pause
Kyla Scanlon explained Engel's Law, which describes how the share of income spent on food declines as income rises. The law offers insight into consumer spending patterns and economic development.
Does anything I write matter anymore?
Noah Smith reflected on whether his writing matters in an age of populism, AI, and monetized media. The piece explores the role of analysis and expertise when partisan incentives dominate and AI generates competing content at scale.
Slow Growth: Mexico Edition
Timothy Taylor examined Mexico's slow growth problem, noting that for eight years the economy has expanded at 1% annually or less, barely faster than population growth. The stagnation raises questions about structural constraints and policy reform.
When travel restrictions became trade frictions: evidence from Covid-era border closures
Bank of England researchers examined how COVID travel restrictions affected cross-border goods trade using a structural gravity model. The work quantifies the friction costs imposed by border closures and documents the supply-chain consequences.
Why Elon Musk Is Really Building SpaceX
Maxinomics examined Elon Musk's true motivations for building SpaceX, separating stated rationales from actual business logic. The video explored Mars colonization rhetoric versus terrestrial revenue drivers.
Space has the properties of things that become bubbles
Maxinomics observed that space companies exhibit properties that make them vulnerable to bubble dynamics. The segment examined valuation cycles and sustainability questions for space-focused ventures.