PDFtoLink Library
Foundational papers, whitepapers, and ideas behind human progress.
Important technologies often begin with specific problems. Shannon studied communication, Berners-Lee wanted to connect information inside a research organization, and the Transformer paper reconsidered how to model sequences. Reading the originals lets you see familiar tools while they were still proposals, with open questions and choices yet to be tested.
The collection also includes business memos, early presentations, and opinion essays. Funding, organizations, and user behavior help determine which technologies reach people. Each entry has an introduction and a reading perspective. Hosted PDFs can be opened or downloaded; other entries link to the original source. Agreement with the author is optional: the questions and tradeoffs are often the most interesting part.
89 documents · 43 hosted PDFs
89 documents
Claude Shannon
Shannon separates the engineering problem of transmitting messages from their meaning, then develops a mathematical language for information and channel limits. It is a foundational text for understanding reliable communication.
Vaswani et al. (Google)
The Transformer replaces recurrence with attention-based components for sequence modeling. The paper lays out the encoder, decoder, positional information, and training choices behind that architecture.
Satoshi Nakamoto
Nakamoto combines digital signatures, proof of work, and a shared transaction history to address double spending without a central clearing authority. The paper is remarkably compact for the system it proposes.
He, Zhang, Ren, Sun (Microsoft)
ResNet introduces residual connections that let a network learn a change to its input representation. The paper studies why making a network deeper can otherwise make training more difficult.
DeepSeek
DeepSeek-R1 examines reinforcement learning as a route to stronger reasoning behavior in language models. It discusses a training path without initial supervised fine-tuning and a more developed multi-stage pipeline.
Mikolov et al. (Google)
Mikolov and colleagues describe efficient ways to learn vector representations of words from large text collections. The paper connects training objectives with representations that encode useful linguistic relationships.
Vitalik Buterin
Buterin proposes a blockchain that can run general-purpose programs rather than only record payments. The whitepaper introduces smart contracts as building blocks for applications with shared state.
Krizhevsky, Sutskever, Hinton
AlexNet shows how a deep convolutional network, trained with GPUs and practical regularization choices, can perform strongly on large-scale image classification. Architecture and engineering matter together here.
Tim Berners-Lee
Tim Berners-Lee's proposal begins with the difficulty of keeping information connected inside a changing research organization. Hypertext is presented as a practical response to scattered knowledge and fragile links between people and projects.
Brown et al. (OpenAI)
The GPT-3 paper studies whether a sufficiently large language model can perform tasks from instructions and examples in its context, without task-specific parameter updates. It compares several prompting settings.
Kaplan et al. (OpenAI)
Kaplan and colleagues study how language-model loss varies with model size, data, and compute. They turn scaling from an intuition into a set of empirical relationships within their experiments.
Anatoly Yakovenko
Yakovenko proposes Proof of History as a way to establish an ordering of events in a distributed system. The whitepaper connects a cryptographic sequence with the coordination demands of a blockchain.
Sergey Brin, Lawrence Page
Brin and Page describe the architecture of a large-scale web search engine, including crawling, indexing, ranking, and the use of links. The paper explains search as an engineered system with several interacting parts.
Fischer Black, Myron Scholes
Black and Scholes derive an option-pricing framework through a hedging argument and a set of market assumptions. The paper connects a contingent payoff with the behavior of the underlying asset.
Hoffmann et al. (DeepMind)
Chinchilla revisits the balance between model size and training data under a fixed compute budget. The central argument is that many large models were trained on too few tokens.
Hayden Adams
Hayden Adams describes an exchange built around liquidity pools and an automated pricing rule. The appeal is that participants can trade against a pool rather than wait for a matching order.
Ken Olsen / Harlan Anderson
Olsen and Anderson's 1957 proposal to American Research and Development sets out an early plan for Digital Equipment Corporation. It records a computer business while its products and organization were still taking shape.
Sequoia Capital
Written for a harsher funding environment in 2022, this Sequoia presentation examines inflation, changing monetary conditions, and the pressure those changes put on startup spending and growth.
ARK Invest (Cathie Wood)
ARK's Big Ideas research connects technological progress with potential changes in markets and industries. The hosted 2024 edition ranges across areas such as AI, robotics, energy, and digital infrastructure.
Zoltan Pozsar
Pozsar links monetary arrangements with commodities, collateral, and geopolitical conflict. The note argues that the material inputs to an economy can matter as much as financial claims on those inputs.
Jim Chanos / Kynikos
The hosted PDF is a 2002 congressional hearing on Enron's collapse and the accounting industry, rather than a standalone Chanos slide deck. It provides a broader record of the scrutiny surrounding the company.
Arthur Rock / Eugene Kleiner
The hosted Computer History Museum archive describes Fairchild's formation and semiconductor history. It connects technical developments with the people, financing, and organizational changes behind the company.
W. Brian Arthur
W. Brian Arthur explains how positive feedback can reinforce an early advantage in technology markets. Learning, network effects, and related mechanisms can make outcomes depend on the path taken.
Mary Meeker
Mary Meeker's Internet Trends reports assemble charts about adoption, usage, commerce, and the changing economics of digital services. They offer a wide-angle snapshot of how the internet entered everyday activity.
Muddy Waters / Carson Block
This hosted report was distributed by Muddy Waters and presents allegations about Luckin Coffee using observational research and financial analysis. Its argument relies heavily on the connection between store activity and reported results.
Benedict Evans
Benedict Evans examines smartphones as a new center of computing, distribution, and consumer behavior. The linked presentations show the mobile transition through charts and comparisons with earlier technology platforms.
Ray Dalio
Dalio compares debt crises through a recurring sequence of borrowing, rising obligations, stress, and policy responses. Historical cases show how the same pressures can play out under different monetary conditions.
Jeremy Grantham
Written during the aftermath of the financial crisis, Grantham's note examines the difficulty of acting when fear dominates market sentiment. It places valuation judgments beside the emotional experience of uncertainty.
Sequoia Capital
Sequoia's crisis presentation turns a deteriorating economy into practical questions about cash, spending, and survival. It captures the abrupt change in startup priorities during the 2008 financial crisis.
Howard Marks
Howard Marks reflects on how the shift away from a long period of falling interest rates changes investment conditions. The memo contrasts habits formed in one environment with decisions in another.
Carson Block / Muddy Waters
Muddy Waters presents its allegations about Sino-Forest through company disclosures and investigative claims about its operations. The report shows how a short seller tries to challenge an established corporate narrative.
Leopold Aschenbrenner
Aschenbrenner connects projected AI progress with compute, industrial capacity, security, and geopolitics. The collection tries to follow the consequences of increasingly capable systems beyond the model laboratory.
Ray Dalio
Dalio organizes economic and political history around recurring pressures: debt, internal conflict, and competition between powers. The work offers a framework for comparing different periods of national strength and decline.
Bill Gates
Bill Gates's 1995 memo treats the internet as a strategic shift Microsoft must respond to across its products. It shows a large company trying to reorient itself around a new platform.
Holden Karnofsky
Holden Karnofsky explores why advances in AI could make this century unusually consequential for the long-term future. The essays connect possible technological transitions with decisions made before their effects are clear.
Howard Marks
Marks describes how competition among providers of capital can weaken lending standards. The memo examines the concessions people make when they fear missing an attractive opportunity.
Warren Buffett
Buffett compares several investors who share a value-oriented intellectual background but make different portfolio choices. He uses their records to challenge an explanation based purely on luck.
Mary Meeker / BOND
Meeker and the BOND team bring their chart-heavy approach to AI in this 2025 report. Adoption, infrastructure, investment, and the economics of deploying models appear in the same landscape.
Andrew Left / Citron Research
Citron's report questions the relationship between Valeant and its distribution arrangements. It uses accounting concerns and comparisons with earlier corporate failures to push readers to reconsider the company's reported business.
Jeremy Grantham
Grantham examines speculative behavior, valuations, and the psychology of a late-stage market boom. His argument joins financial measures with observations about how people behave around rising prices.
Bill Ackman / Pershing Square
Ackman's presentation lays out his criticism of Herbalife's business model using sales practices, participant economics, and corporate disclosures. It is an example of a highly developed activist short-selling argument.
Roelof Botha / Sequoia Capital
Roelof Botha's YouTube investment memo turns a young video service into a set of questions about users, distribution, growth, and risks. It records a decision before the outcome was known.
Kevin Kelly
Kevin Kelly suggests that a creator can build a livelihood around a relatively small group of committed supporters. The essay shifts attention from mass popularity to a direct relationship with an audience.
Hindenburg Research
Hindenburg organizes its criticism of the Adani group around ownership structures, related-party relationships, and accounting concerns. The report is a forceful example of investigative financial argument.
Ben Thompson
Ben Thompson explains how internet distribution changes the relationship between suppliers, intermediaries, and customers. Platforms that own the user relationship can reorganize whole markets around themselves.
Eliezer Yudkowsky
Yudkowsky lists reasons he believes very capable AI could be difficult to control. The essay emphasizes failures of alignment, incentives, and the limits of current approaches to managing them.
Airbnb (Brian Chesky et al.)
Airbnb's early pitch distills a travel accommodation idea into a problem, a proposed service, and a business opportunity. Its brevity makes the underlying presentation structure easy to inspect.
Bill Gurley
Bill Gurley examines the characteristics that make digital marketplaces easier or harder to build: transaction frequency, fragmented supply, market size, and the value an intermediary can add.
Reid Hoffman & Chris Yeh
Hoffman and Yeh describe prioritizing speed over efficiency when a market offers the possibility of unusually large scale. The framework connects rapid growth with organizational strain and uncertainty.
Jack Dorsey
Jack Dorsey's 2026 statement connects AI tools with a proposed change in Block's organization and operating model. It presents intelligence-native work as a management direction rather than just a product feature.
Warren Buffett
Writing during the 2008 crisis, Buffett separates frightening economic news from the question of long-term business ownership. The essay explains his reasoning in unusually accessible language.
Peter Thiel
Thiel challenges the assumption that intense competition is always a sign of a good business. He argues that distinctive capabilities can give a company room to build something lasting.
Paul Graham
Paul Graham makes a case for recruiting early users personally and giving them unusually attentive service. Before a startup can automate a process, it often has to learn that process by hand.
The early Facebook media kit presents the service through its audience and the opportunity to reach that audience. It is a window into how a young social network explained its commercial appeal.
Joel Monegro / USV
Joel Monegro argues that blockchain networks may capture value differently from earlier internet architectures. Shared data and tokens are used to explain why value might accrue at the protocol layer.
David Einhorn / Greenlight
Einhorn's account of Allied Capital follows a long dispute over valuation, accounting, and corporate responses to criticism. It combines investment research with the difficulties of getting contested evidence heard.
Paul Graham
Graham distinguishes ideas that sound plausible from problems people actually have. He suggests noticing missing tools in environments you understand deeply, rather than brainstorming startups in the abstract.
Eugene Wei
Eugene Wei explores the hidden ceilings that limit a product's growth even when its early trajectory looks promising. Price, selection, convenience, and audience needs can each create a different ceiling.
Chamath Palihapitiya
The IPO 2.0 materials promote SPACs as an alternative route for private companies to enter public markets. They frame the process through financing, negotiation, and the interests of founders and investors.
David Einhorn / Greenlight
This entry concerns Einhorn's criticism of Lehman's accounting and balance-sheet presentation before its collapse. The catalog's linked page, however, covers a different Ira Sohn speech from 2010.
Dario Amodei
Dario Amodei explores what powerful AI might contribute to biology, health, economic development, and human institutions. The essay focuses on a desirable future rather than only cataloging risks.
Scott Alexander
Scott Alexander uses coordination failures to explain why individually sensible choices can produce outcomes nobody wants. The essay moves between social systems and the incentives that keep them running.
Sam Altman
Sam Altman imagines AI driving down the cost of important goods and services, then asks how society might distribute the resulting gains. Technology and economic institutions are treated together.
Hindenburg Research
Hindenburg's Nikola report challenges claims about technology and commercial progress using public statements and investigative material. It illustrates how a technology story can become a target for detailed verification.
Battery Ventures
Battery's OpenCloud research surveys enterprise software, cloud infrastructure, and the companies building around them. It connects technical shifts with adoption and the changing shape of software businesses.
David Skok
David Skok connects recurring revenue with customer acquisition costs, churn, and the time needed to recover spending. The article makes SaaS growth easier to understand as a cash-flow problem.
Michael Burry
Burry's Scion correspondence is associated with his analysis of mortgage lending, securitization, and the risks within apparently safe credit products. The letters also offer a view into how he explains decisions to investors.
Andrej Karpathy
Karpathy describes neural-network weights as a different way to express software behavior. Instead of specifying every rule directly, developers shape datasets, objectives, and the process that learns the rules.
Nick Bostrom
Bostrom examines possible routes to machine intelligence beyond human capabilities and the challenge of steering such systems. The book gives a vocabulary to problems involving goals, control, and strategic advantage.
Rich Sutton
Rich Sutton argues that methods which make effective use of growing computation tend to outperform approaches built around human-crafted domain knowledge. Search and learning are his recurring examples.
Vijay Boyapati
Vijay Boyapati builds an argument for Bitcoin as a monetary asset by comparing properties such as scarcity, portability, and resistance to interference. The essay connects those properties with a theory of adoption.
Lyn Alden
Lyn Alden examines the pressures on a global monetary system centered on the US dollar. Trade, reserves, energy, and the incentives of different countries form the backbone of her explanation.
Eric Ries
Eric Ries presents the minimum viable product as a way to test assumptions and learn with less wasted effort. The broader Lean Startup framework links experiments, measurement, and changes of direction.
Jerry Chen
Jerry Chen examines how software businesses can defend their position through data, workflows, and systems of intelligence. The linked Greylock page also revisits the framework in light of newer AI tools.
Chris Dixon
Chris Dixon argues that technologies dismissed as toys can improve until they threaten established products. Early limitations may conceal a different audience, cost structure, or path of development.
Nikhil Basu Trivedi
Nikhil Basu Trivedi describes investors who operate with the identity and decision-making focus of an individual while managing meaningful pools of capital. The essay examines a changing shape of venture investing.
Gwern Branwen
Gwern examines the possibility that larger training runs can unlock broader language-model capabilities. The essay collects observations and arguments about what scaling might explain and where it might lead.
Elon Musk
Musk's early Tesla plan links a high-priced initial vehicle with the aim of funding more accessible models. It explains a sequence for entering an expensive industry while pursuing a broader energy ambition.
James Dale Davidson & Lord William Rees-Mogg
Davidson and Rees-Mogg connect information technology with changes in state power, taxation, and the mobility of individuals. The book treats digital networks as a force that could reshape political institutions.
Marc Andreessen
Andreessen makes a philosophical case for technological growth, abundance, and the people who build new tools. The manifesto presents progress as something worth actively pursuing.
Harry Markopolos
Harry Markopolos's memo challenges the plausibility of Madoff's reported investment results. It is known for approaching a persuasive performance story through consistency checks and questions about how the strategy could work.
Uber (Garrett Camp / Travis Kalanick)
The early UberCab pitch starts with an urban transportation problem and proposes a different experience for booking a ride. It captures a narrower starting point than the business Uber later became.
Patrick Collison & Tyler Cowen
Collison and Cowen propose studying progress itself: how scientific, technological, and organizational advances happen, and how societies might make them more frequent. The essay calls for work across disciplinary boundaries.
Founders Fund (Bruce Gibney)
Founders Fund questions whether the technology industry has become too comfortable with incremental digital products. It makes a case for ambitious work on problems outside the familiar software playbook.
William MacAskill
William MacAskill asks how present decisions should account for people who will live far into the future. The book connects moral responsibility with risks and choices whose consequences may last a long time.
Marc Andreessen
Andreessen presents an optimistic argument for AI as a tool that can improve education, productivity, and scientific work. The essay also contests several common arguments against AI development.
Chris Dixon
Dixon contrasts centrally controlled networks with networks whose rules and incentives are distributed among participants. He connects architecture choices with the treatment of developers and users over time.
Marc Andreessen
Andreessen argues that software is becoming the operating layer of industries that once looked unrelated to computing, from retail to entertainment. The essay asks readers to see businesses through their software capabilities.
Peter Thiel & Blake Masters
Thiel and Masters argue for building distinctive businesses rather than repeating an established formula. Technology, differentiation, and the choices that shape an early company run through the book.
Directory source: Billion Dollar PDFs