https://fanout.sh/daily Daily technical papers with concise reading notes across AI research, systems, inference, and ML mathematics. en-US Tue, 25 Aug 2026 12:00:00 GMT https://fanout.sh/daily/2026-08-25-segment-anything https://fanout.sh/daily/2026-08-25-segment-anything Tue, 25 Aug 2026 12:00:00 GMT AI Research Kirillov et al. · ICCV 2023 · 35 min. Segment Anything asks whether one vision model can turn a point, box, or rough mask into the pixels of a reasonable object, even on unfamiliar images. Read it to see how one reusable image embedding, a lightweight prompt decoder, ambiguity-aware mask candidates, and a model-in-the-loop data engine make segmentation feel like an interface rather than a fixed label set. https://fanout.sh/daily/2026-08-24-google-borg-cluster-management https://fanout.sh/daily/2026-08-24-google-borg-cluster-management Mon, 24 Aug 2026 12:00:00 GMT Systems Verma et al. · EuroSys 2015 · 40 min. Borg asks how one fleet can run latency-sensitive services and opportunistic batch work without wasting the space between their peaks. Read it to see how cells, jobs, tasks, priorities, limits, reservations, placement scoring, and a continuously repaired control plane turn heterogeneous machines into shared infrastructure. https://fanout.sh/daily/2026-08-23-clip-natural-language-supervision https://fanout.sh/daily/2026-08-23-clip-natural-language-supervision Sun, 23 Aug 2026 12:00:00 GMT AI Research Radford et al. · ICML 2021 · 35 min. CLIP asks how an image model can recognize categories described in words without training a new classifier for every task. Read it to see how two encoders turn a batch into an all-pairs similarity matrix, how the true image-text pairs teach a shared embedding space, and why prompts make zero-shot recognition flexible but not automatically reliable. https://fanout.sh/daily/2026-08-22-development-domain-name-system https://fanout.sh/daily/2026-08-22-development-domain-name-system Sat, 22 Aug 2026 12:00:00 GMT Systems Mockapetris and Dunlap · SIGCOMM 1988 · 30 min. This paper asks how the Internet can keep one usable naming system without making every machine download one global host file. Read it to see how DNS cuts a name tree into delegated zones, follows referrals toward the server responsible for one branch, and uses time-limited caching to make repeated lookups local. https://fanout.sh/daily/2026-08-21-end-to-end-arguments https://fanout.sh/daily/2026-08-21-end-to-end-arguments Fri, 21 Aug 2026 12:00:00 GMT Systems Saltzer, Reed, and Clark · ACM TOCS, 1984 · 30 min. This paper asks where a system should place reliability, delivery, security, and duplicate-suppression guarantees. Read it to see why a lower layer can make failures rarer yet still cannot certify the application result, and why the final check belongs at the endpoints that understand what success means. https://fanout.sh/daily/2026-08-20-toolformer https://fanout.sh/daily/2026-08-20-toolformer Thu, 20 Aug 2026 12:00:00 GMT AI Research Schick et al. · NeurIPS 2023 · 40 min. Toolformer asks how a language model can learn when and how to call external tools without a large hand-labeled tool-use dataset. Read it to see how the model proposes calls inside ordinary text, executes them, and keeps only those whose returned result lowers continuation loss enough to become a training example. https://fanout.sh/daily/2026-08-19-shazam-audio-search https://fanout.sh/daily/2026-08-19-shazam-audio-search Wed, 19 Aug 2026 12:00:00 GMT Systems Avery Li-Chun Wang · ISMIR 2003 · 30 min. This is the paper behind Shazam: how a phone can name a song from ten noisy seconds in a pub. Read it to see how a track is reduced to a constellation of spectrogram peaks, how pairing peaks into 32-bit hashes makes lookup about ten thousand times faster, and why a match is declared only when the surviving hashes agree on one time offset. https://fanout.sh/daily/2026-08-18-chain-of-thought https://fanout.sh/daily/2026-08-18-chain-of-thought Tue, 18 Aug 2026 12:00:00 GMT AI Research Wei, Wang, Schuurmans, Bosma, Ichter, Xia, Chi, Le, and Zhou · NeurIPS 2022 · 30 min. This paper asks why showing a language model a few worked examples of step-by-step reasoning unlocks problems that direct prompting cannot touch. Read it to see how eight hand-written exemplars lifted PaLM 540B from 17.9% to 56.9% on grade-school math, why meaningless filler tokens explain none of the gain, and why the same prompt actively hurts smaller models. https://fanout.sh/daily/2026-08-17-bloom-filter https://fanout.sh/daily/2026-08-17-bloom-filter Mon, 17 Aug 2026 12:00:00 GMT Systems Burton H. Bloom · Communications of the ACM, 1970 · 30 min. Bloom's paper asks how a compact in-memory screen can reject missing keys before an expensive exact lookup. Read it to see why one zero bit proves absence, why all-one evidence means only possibly present, and how shared hash positions trade memory for a controlled false-positive path. https://fanout.sh/daily/2026-08-16-tor-onion-router https://fanout.sh/daily/2026-08-16-tor-onion-router Sun, 16 Aug 2026 12:00:00 GMT Systems Dingledine, Mathewson, and Syverson · USENIX Security 2004 · 35 min. This paper asks how a browser can cross several volunteer relays without handing any one relay the complete route. Read it to see how the client extends a circuit one hop at a time, wraps each fixed-size cell in a different session key for every relay, and why those layers still cannot defeat an observer who can correlate traffic at both ends. https://fanout.sh/daily/2026-08-15-alphago-neural-tree-search https://fanout.sh/daily/2026-08-15-alphago-neural-tree-search Sat, 15 Aug 2026 12:00:00 GMT AI Research Silver et al. · Nature 2016 · 35 min. This paper asks how a computer can choose a strong Go move when the game tree is far too large to enumerate. Read it to see how a learned policy concentrates search on promising moves, how a value network estimates who is ahead before a game ends, and how Monte Carlo tree search turns those estimates into professional-level play under a large compute budget. https://fanout.sh/daily/2026-08-14-bitcoin-proof-of-work https://fanout.sh/daily/2026-08-14-bitcoin-proof-of-work Fri, 14 Aug 2026 12:00:00 GMT Systems Satoshi Nakamoto · Bitcoin white paper, 2008 · 35 min. This paper asks how two people can agree that a digital payment happened first without asking a central mint to keep the ledger. Read it to see why signatures prove ownership but not uniqueness, how hash-linked proof of work makes one public history costly to rewrite, and why every confirmation still depends on an honest-compute majority assumption. https://fanout.sh/daily/2026-08-13-congestion-avoidance-control https://fanout.sh/daily/2026-08-13-congestion-avoidance-control Thu, 13 Aug 2026 12:00:00 GMT Systems Van Jacobson · SIGCOMM 1988 · 35 min. This paper asks how TCP can keep a busy network useful when delay, packet loss, and retransmission tempt every sender to add even more traffic. Read it to see how acknowledgements become a clock, how slow start fills an empty path without a full-window burst, and why a congestion signal should make the sender retreat quickly before probing capacity again. https://fanout.sh/daily/2026-08-12-stolen-reasoning-traces https://fanout.sh/daily/2026-08-12-stolen-reasoning-traces Wed, 12 Aug 2026 12:00:00 GMT AI Research Panfilov, Schmotz, Shumailov, Beurer-Kellner, Schaeffer, Prabhu, Geiping, and Andriushchenko · arXiv, 2026 · 35 min. This paper asks what happens when an API gives encrypted reasoning state to the client and later accepts that opaque state again. Read it to see why confidentiality and integrity do not automatically bind a reasoning block to one user, session, or model, how the authors tested compatible weaker decoders, and what the resulting privacy evidence means after responsible disclosure. https://fanout.sh/daily/2026-08-11-pagerank https://fanout.sh/daily/2026-08-11-pagerank Tue, 11 Aug 2026 12:00:00 GMT Systems Page, Brin, Motwani, and Winograd · Stanford InfoLab, 1999 · 35 min. PageRank asks how a search engine can estimate which web pages matter before it even sees a query. Read it to see how every page divides its authority among outgoing links, how repeated link-weight transfer settles into a global ranking, and why teleportation, incomplete crawls, and manipulation keep that score from being a synonym for truth. https://fanout.sh/daily/2026-08-10-react-reasoning-acting https://fanout.sh/daily/2026-08-10-react-reasoning-acting Mon, 10 Aug 2026 12:00:00 GMT AI Research Yao, Zhao, Yu, Du, Shafran, Narasimhan, and Cao · ICLR 2023 · 35 min. ReAct gives language-model agents a simple working rhythm: reason about the current situation, take a bounded action, observe what the environment returns, and revise the next move. Read it for the durable idea behind many tool-using agent harnesses and for an honest look at why better grounding can still fail when tools return weak evidence or the agent gets stuck in a loop. https://fanout.sh/daily/2026-08-09-ring-attention https://fanout.sh/daily/2026-08-09-ring-attention Sun, 09 Aug 2026 12:00:00 GMT Systems Liu, Zaharia, and Abbeel · ICLR 2024 · 50 min. Ring Attention asks how exact Transformer context can grow across devices without gathering the whole sequence on each one. Read it to see how stationary query blocks, circulating key-value blocks, and compute-transfer overlap turn a device ring into aggregate context capacity while keeping per-host activation memory independent of total sequence length. https://fanout.sh/daily/2026-08-08-flashattention-2 https://fanout.sh/daily/2026-08-08-flashattention-2 Sat, 08 Aug 2026 12:00:00 GMT Inference Tri Dao · ICLR 2024 · 50 min. FlashAttention-2 asks why an exact, I/O-aware attention kernel can still leave much of an A100 idle. Read it to see how reducing non-matmul work, parallelizing along sequence tiles, and giving each warp disjoint query and output rows improves GPU utilization without changing the attention result. https://fanout.sh/daily/2026-08-07-mamba-selective-state-spaces https://fanout.sh/daily/2026-08-07-mamba-selective-state-spaces Fri, 07 Aug 2026 12:00:00 GMT AI Research Gu and Dao · COLM 2024 · 55 min. Mamba asks how a recurrent model with a fixed-size state can decide what information deserves to survive. Read it to see how input-dependent state-space parameters create content-aware persistence and reset, and how a fused parallel scan makes the resulting time-varying recurrence practical on GPUs. https://fanout.sh/daily/2026-08-06-rmsnorm https://fanout.sh/daily/2026-08-06-rmsnorm Thu, 06 Aug 2026 12:00:00 GMT AI Research Zhang and Sennrich · NeurIPS 2019 · 35 min. RMSNorm asks whether LayerNorm needs to subtract the mean as well as control activation scale. Read it to see how dividing by a vector's root mean square preserves global rescaling invariance with less normalization work, while deliberately giving up re-centering invariance. https://fanout.sh/daily/2026-08-05-zero-memory-optimization https://fanout.sh/daily/2026-08-05-zero-memory-optimization Wed, 05 Aug 2026 12:00:00 GMT Systems Rajbhandari et al. · SC 2020 · 50 min. ZeRO asks why adding data-parallel workers does not make a model easier to fit when every worker keeps the same training state. Read it to see how partitioning optimizer states, gradients, and parameters turns aggregate cluster memory into usable model capacity while preserving the data-parallel computation. https://fanout.sh/daily/2026-08-04-orca-continuous-batching https://fanout.sh/daily/2026-08-04-orca-continuous-batching Tue, 04 Aug 2026 12:00:00 GMT Inference Yu et al. · OSDI 2022 · 45 min. Orca moves LLM serving from fixed request batches to iteration-level scheduling. Read it to see how a scheduler can reconsider the active batch after every generated token, and how selective batching preserves efficient shared GPU work even when requests have different sequence lengths and lifetimes. https://fanout.sh/daily/2026-08-03-grouped-query-attention https://fanout.sh/daily/2026-08-03-grouped-query-attention Mon, 03 Aug 2026 12:00:00 GMT Inference Ainslie et al. · EMNLP 2023 · 35 min. Grouped-query attention finds a practical middle ground between multi-head attention's quality and multi-query attention's decoding speed. Read it to see how sharing key-value heads within groups shrinks the KV cache, and how an existing multi-head checkpoint can be converted with only a small amount of additional pretraining. https://fanout.sh/daily/2026-08-02-deepseekmath-grpo https://fanout.sh/daily/2026-08-02-deepseekmath-grpo Sun, 02 Aug 2026 12:00:00 GMT ML Math Shao et al. · arXiv 2024 · 55 min. DeepSeekMath shows how targeted data and reinforcement learning can turn a compact open model into a strong mathematical reasoner. Read it for the data-selection pipeline behind its 120 billion math tokens and for GRPO, a PPO variant that removes the separate critic model by learning from relative rewards within groups of sampled answers. https://fanout.sh/daily/2026-08-01-proximal-policy-optimization https://fanout.sh/daily/2026-08-01-proximal-policy-optimization Sat, 01 Aug 2026 12:00:00 GMT ML Math Schulman et al. · arXiv 2017 · 55 min. Proximal Policy Optimization shows how to reuse on-policy experience for several minibatch updates without letting the policy move too far in one step. Read it to understand the clipped surrogate objective behind a practical reinforcement-learning algorithm that later became central to RLHF pipelines. https://fanout.sh/daily/2026-07-31-instructgpt https://fanout.sh/daily/2026-07-31-instructgpt Fri, 31 Jul 2026 12:00:00 GMT AI Research Ouyang et al. · NeurIPS 2022 · 45 min. InstructGPT shows why a larger language model is not automatically better at following a user's intent. Read it to see how demonstrations, preference rankings, a learned reward model, and reinforcement learning turn human judgments into a practical post-training pipeline for more helpful instruction following. https://fanout.sh/daily/2026-07-30-switch-transformers https://fanout.sh/daily/2026-07-30-switch-transformers Thu, 30 Jul 2026 12:00:00 GMT AI Research Fedus, Zoph, and Shazeer · JMLR 2022 · 50 min. Switch Transformers shows how to grow a model's capacity without activating every parameter for every token. Read it to see how top-1 expert routing, load balancing, and expert capacity turn mixture-of-experts into a simpler scaling recipe while keeping per-token computation roughly constant. https://fanout.sh/daily/2026-07-29-knowledge-distillation https://fanout.sh/daily/2026-07-29-knowledge-distillation Wed, 29 Jul 2026 12:00:00 GMT AI Research Hinton, Vinyals, and Dean · NIPS 2014 Deep Learning Workshop · 30 min. Knowledge distillation turns an expensive ensemble or large teacher into a smaller, deployable student. Read it to see why the full pattern of class probabilities carries more useful structure than hard labels alone, and how temperature-scaled soft targets transfer that structure during training. https://fanout.sh/daily/2026-07-28-tail-at-scale https://fanout.sh/daily/2026-07-28-tail-at-scale Tue, 28 Jul 2026 12:00:00 GMT Systems Dean and Barroso · Communications of the ACM 2013 · 30 min. The Tail at Scale explains why rare slowdowns become routine user-visible delays when one request fans out across many machines. Read it for a practical systems lesson: large services need techniques that tolerate latency variability, not only faster average components. https://fanout.sh/daily/2026-07-27-speculative-decoding https://fanout.sh/daily/2026-07-27-speculative-decoding Mon, 27 Jul 2026 12:00:00 GMT Inference Leviathan, Kalman, and Matias · ICML 2023 · 40 min. Speculative decoding makes autoregressive generation faster without changing the target model's output distribution. Read it to see how a small draft model, parallel verification, and a corrected sampling rule turn extra compute into lower decoding latency without retraining the models. https://fanout.sh/daily/2026-07-26-retrieval-augmented-generation https://fanout.sh/daily/2026-07-26-retrieval-augmented-generation Sun, 26 Jul 2026 12:00:00 GMT AI Research Lewis et al. · NeurIPS 2020 · 40 min. Retrieval-Augmented Generation gives a language model an explicit, updateable memory outside its parameters. Read it to see how a neural retriever and a sequence generator work together, and why grounding generation in retrieved evidence can improve factual, knowledge-intensive answers. https://fanout.sh/daily/2026-07-25-denoising-diffusion-probabilistic-models https://fanout.sh/daily/2026-07-25-denoising-diffusion-probabilistic-models Sat, 25 Jul 2026 12:00:00 GMT ML Math Ho, Jain, and Abbeel · NeurIPS 2020 · 50 min. Denoising Diffusion Probabilistic Models turns generation into a sequence of small denoising steps. Read it to connect variational learning, Gaussian noise, and score matching to the practical recipe that became the foundation of modern diffusion image models. https://fanout.sh/daily/2026-07-24-auto-encoding-variational-bayes https://fanout.sh/daily/2026-07-24-auto-encoding-variational-bayes Fri, 24 Jul 2026 12:00:00 GMT ML Math Kingma and Welling · ICLR 2014 · 55 min. Auto-Encoding Variational Bayes makes latent-variable models trainable with ordinary backpropagation. Read it to see how the evidence lower bound, amortized inference, and the reparameterization trick turn an intractable posterior problem into a practical stochastic-optimization recipe. https://fanout.sh/daily/2026-07-23-dropout https://fanout.sh/daily/2026-07-23-dropout Thu, 23 Jul 2026 12:00:00 GMT ML Math Srivastava et al. · JMLR 2014 · 35 min. Dropout is a compact lesson in turning randomness into regularization. Read it to see how masking neurons during training discourages brittle co-adaptation and approximates an ensemble without paying the cost of running many networks at inference time. https://fanout.sh/daily/2026-07-22-direct-preference-optimization https://fanout.sh/daily/2026-07-22-direct-preference-optimization Wed, 22 Jul 2026 12:00:00 GMT AI Research Rafailov et al. · NeurIPS 2023 · 40 min. DPO turns language-model alignment from a multi-stage reinforcement-learning pipeline into a direct preference-learning objective. Read it to see how one mathematical reparameterization can remove the separate reward model and make preference tuning much simpler to train. https://fanout.sh/daily/2026-07-21-pagedattention https://fanout.sh/daily/2026-07-21-pagedattention Tue, 21 Jul 2026 12:00:00 GMT Inference Kwon et al. · SOSP 2023 · 55 min. PagedAttention connects a familiar systems idea—virtual-memory paging—to one of LLM serving's hardest constraints: a large, dynamic KV cache. Read it to see how memory layout, sharing, and allocation policy turn directly into higher batching capacity and serving throughput. https://fanout.sh/daily/2026-07-20-kv-cache-management-survey https://fanout.sh/daily/2026-07-20-kv-cache-management-survey Mon, 20 Jul 2026 12:00:00 GMT Inference Li, Wang, and Chen · arXiv 2026 · 50 min. This survey treats the KV cache as a distributed memory-management problem, not just a temporary tensor. It is a useful July 20 pick for understanding why LLM serving depends on locality, lifetime, ownership, substrate, and the tradeoffs behind paged, shared, disaggregated, and tiered cache designs. https://fanout.sh/daily/2026-07-19-attention-is-all-you-need https://fanout.sh/daily/2026-07-19-attention-is-all-you-need Sun, 19 Jul 2026 12:00:00 GMT AI Research Vaswani et al. · NeurIPS 2017 · 45 min. The Transformer paper replaced recurrence with attention and made sequence modeling easier to parallelize. Read it for the architectural move that turned token-to-token routing into the core primitive behind modern language models. https://fanout.sh/daily/2026-07-18-flashattention https://fanout.sh/daily/2026-07-18-flashattention Sat, 18 Jul 2026 12:00:00 GMT Inference Dao et al. · NeurIPS 2022 · 55 min. FlashAttention keeps exact attention but changes the memory traffic pattern. It is a clean example of why practical model speed often comes from respecting hardware limits rather than changing the math. https://fanout.sh/daily/2026-07-17-scaling-laws https://fanout.sh/daily/2026-07-17-scaling-laws Fri, 17 Jul 2026 12:00:00 GMT ML Math Kaplan et al. · arXiv 2020 · 50 min. The scaling-laws paper turns model size, data, and compute into a quantitative tradeoff. It is useful when you want to reason about progress curves instead of treating bigger training runs as folklore. https://fanout.sh/daily/2026-07-16-chinchilla https://fanout.sh/daily/2026-07-16-chinchilla Thu, 16 Jul 2026 12:00:00 GMT AI Research Hoffmann et al. · arXiv 2022 · 45 min. Chinchilla is the scaling paper that made data budget feel as important as parameter count. It is a good daily pick for learning why a smaller model trained on more tokens can beat a larger undertrained one. https://fanout.sh/daily/2026-07-15-dynamo https://fanout.sh/daily/2026-07-15-dynamo Wed, 15 Jul 2026 12:00:00 GMT Systems DeCandia et al. · SOSP 2007 · 60 min. Dynamo is the classic availability-first storage paper. It is worth reading for how partitioning, quorums, hinted handoff, vector clocks, and application-level conflict resolution combine into a product-shaped system. https://fanout.sh/daily/2026-07-14-mapreduce https://fanout.sh/daily/2026-07-14-mapreduce Tue, 14 Jul 2026 12:00:00 GMT Systems Dean and Ghemawat · OSDI 2004 · 40 min. MapReduce shows how a constrained programming model can hide distribution, retries, shuffling, and locality. Read it for the system-design lesson that an API can be an operational strategy. https://fanout.sh/daily/2026-07-13-raft https://fanout.sh/daily/2026-07-13-raft Mon, 13 Jul 2026 12:00:00 GMT Systems Ongaro and Ousterhout · USENIX ATC 2014 · 45 min. Raft is the consensus paper to read before discussing leader election, replicated logs, or metadata coordination. It is intentionally structured for understandability, which makes it a strong beginner systems paper. https://fanout.sh/daily/2026-07-12-lora https://fanout.sh/daily/2026-07-12-lora Sun, 12 Jul 2026 12:00:00 GMT Inference Hu et al. · ICLR 2022 · 35 min. LoRA is the cleanest first paper for understanding efficient model adaptation. It turns fine-tuning into small low-rank update matrices, which connects directly to Fanout's model training and serving material. https://fanout.sh/daily/2026-07-11-bert https://fanout.sh/daily/2026-07-11-bert Sat, 11 Jul 2026 12:00:00 GMT AI Research Devlin et al. · NAACL 2019 · 45 min. BERT is the pretrain-then-fine-tune paper that made bidirectional Transformer encoders the default for many language understanding tasks. It is useful for separating encoder intuition from decoder-only LLM intuition. https://fanout.sh/daily/2026-07-10-batch-normalization https://fanout.sh/daily/2026-07-10-batch-normalization Fri, 10 Jul 2026 12:00:00 GMT ML Math Ioffe and Szegedy · ICML 2015 · 35 min. BatchNorm is a compact paper for learning how activation statistics, optimization stability, and architecture design interact. It is a good bridge between practical deep learning and the math of normalization. https://fanout.sh/daily/2026-07-09-resnet https://fanout.sh/daily/2026-07-09-resnet Thu, 09 Jul 2026 12:00:00 GMT AI Research He et al. · CVPR 2016 · 35 min. ResNet is the skip-connection paper that makes depth trainable in practice. It is beginner-friendly because the core idea is simple: learn a residual change instead of relearning the whole representation. https://fanout.sh/daily/2026-07-08-word2vec https://fanout.sh/daily/2026-07-08-word2vec Wed, 08 Jul 2026 12:00:00 GMT AI Research Mikolov et al. · ICLR Workshop 2013 · 30 min. The Word2Vec paper is still one of the best first reads for representation learning. It shows how simple prediction tasks can turn words into vectors with useful geometric structure. https://fanout.sh/daily/2026-07-07-sequence-to-sequence https://fanout.sh/daily/2026-07-07-sequence-to-sequence Tue, 07 Jul 2026 12:00:00 GMT AI Research Sutskever et al. · NeurIPS 2014 · 35 min. Seq2Seq is the clean predecessor to attention and Transformers. It is a good beginner paper because the encoder-decoder contract is easy to visualize and still appears in translation, summarization, and agents. https://fanout.sh/daily/2026-07-06-adam https://fanout.sh/daily/2026-07-06-adam Mon, 06 Jul 2026 12:00:00 GMT ML Math Kingma and Ba · ICLR 2015 · 30 min. Adam is the optimizer paper learners meet constantly in code before they understand it. The first pass is approachable: it combines momentum-like gradient averages with squared-gradient adaptation. https://fanout.sh/daily/2026-07-05-dremel https://fanout.sh/daily/2026-07-05-dremel Sun, 05 Jul 2026 12:00:00 GMT Systems Melnik et al. · VLDB 2010 · 45 min. Dremel is a strong systems pick for understanding columnar analytics, nested data, and query serving at scale. It connects directly to Fanout's storage and analytics material without becoming a database-product tour. https://fanout.sh/daily/2026-07-04-bigtable https://fanout.sh/daily/2026-07-04-bigtable Sat, 04 Jul 2026 12:00:00 GMT Systems Chang et al. · OSDI 2006 · 45 min. Bigtable is the beginner-friendly bridge from simple key-value stores to distributed sorted maps. It teaches rows, tablets, locality, metadata, and why storage shape follows access patterns. https://fanout.sh/daily/2026-07-03-google-file-system https://fanout.sh/daily/2026-07-03-google-file-system Fri, 03 Jul 2026 12:00:00 GMT Systems Ghemawat et al. · SOSP 2003 · 40 min. GFS is a useful first distributed storage paper because the assumptions are explicit: large files, append-heavy workloads, commodity failures, and a master that keeps metadata tractable. https://fanout.sh/daily/2026-07-02-mathematical-theory-of-communication https://fanout.sh/daily/2026-07-02-mathematical-theory-of-communication Thu, 02 Jul 2026 12:00:00 GMT ML Math Claude Shannon · Bell System Technical Journal 1948 · 65 min. Shannon's information theory paper is hard but worth anchoring early in the calendar. Entropy, bits, uncertainty, and channel capacity sit underneath compression, coding, language modeling, and evaluation intuition. https://fanout.sh/daily/2026-07-01-perceptron https://fanout.sh/daily/2026-07-01-perceptron Wed, 01 Jul 2026 12:00:00 GMT AI Research Frank Rosenblatt · Psychological Review 1958 · 30 min. The Perceptron is a good July 1 starting point because it gives learners the original linear-classifier idea before modern depth, attention, or scale enter the story.