<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"><channel><title>Gregory Friedman · The Atlas</title><description>Essays and lectures charting the ground between computation, language, &amp; history, surveyed &amp; drawn in Storrs, Connecticut. New ones reach you here on their own, with no platform in between.</description><link>https://gregoryfriedman.com/</link><language>en-us</language><copyright>© 2026 Gregory Friedman · content licensed CC BY-NC 4.0</copyright><item><title>Clustering without an answer key</title><link>https://gregoryfriedman.com/lectures/machine-learning/clustering-without-an-answer-key/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/machine-learning/clustering-without-an-answer-key/</guid><description>k-means and a Gaussian mixture on the same points: two objectives, two alternating algorithms, and a dataset where the correct answer scores worse than the wrong one.</description><pubDate>Mon, 03 Aug 2026 00:00:00 GMT</pubDate></item><item><title>Decision trees, and a model with no gradient</title><link>https://gregoryfriedman.com/lectures/machine-learning/decision-trees/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/machine-learning/decision-trees/</guid><description>A learner with no loss to differentiate and no learning rate: greedy search over thresholds, impurity as the thing being reduced, and the axis-aligned bias that comes free with the choice of question.</description><pubDate>Mon, 03 Aug 2026 00:00:00 GMT</pubDate></item><item><title>Disjoint sets and Kruskal</title><link>https://gregoryfriedman.com/lectures/data-structures/disjoint-sets-kruskal/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/disjoint-sets-kruskal/</guid><description>Union-find, path compression, union by size, and a second spanning-tree algorithm to close the loop: the structure whose running time is so nearly constant that saying so precisely requires the inverse Ackermann function.</description><pubDate>Mon, 27 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Graphs: search, shortest paths, spanning trees</title><link>https://gregoryfriedman.com/lectures/data-structures/graphs/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/graphs/</guid><description>Graph ADTs, DFS and BFS, Dijkstra and Prim, and the observation that ties them together: all four are one algorithm, and the only difference is which container holds the frontier.</description><pubDate>Mon, 27 Jul 2026 00:00:00 GMT</pubDate></item><item><title>A neural network from scratch</title><link>https://gregoryfriedman.com/lectures/machine-learning/a-neural-network-from-scratch/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/machine-learning/a-neural-network-from-scratch/</guid><description>Backpropagation derived honestly as the chain rule run backwards, then implemented raw in plain Python, with every gradient checked against finite differences.</description><pubDate>Mon, 27 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Linear Regression &amp; Gradient Descent</title><link>https://gregoryfriedman.com/lectures/machine-learning/linear-regression-gradient-descent/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/machine-learning/linear-regression-gradient-descent/</guid><description>The canonical opener; least squares derived rather than asserted, coded from scratch in plain Python, and a loss surface you can descend yourself.</description><pubDate>Mon, 27 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Contract Without Equals: Hierarchy and the Limits of Redistribution</title><link>https://gregoryfriedman.com/essays/contract-without-equals/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/contract-without-equals/</guid><description>Rousseau&apos;s social contract can license redistributive coercion only among civic equals. Where racial or gendered hierarchy survives, contract language stops naming self-rule and starts covering domination.</description><pubDate>Fri, 24 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Hiroshima and Nagasaki: Was Annihilation Justified?</title><link>https://gregoryfriedman.com/essays/hiroshima-and-nagasaki/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/hiroshima-and-nagasaki/</guid><description>Three lenses on the same decision, humanitarian, utilitarian, and political, and the case that the utilitarian defense was assembled after the fact.</description><pubDate>Fri, 24 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Law-As-A-Service: AI adoption in the legal industry</title><link>https://gregoryfriedman.com/essays/lawyers-as-ai-supervisors/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/lawyers-as-ai-supervisors/</guid><description>Legal work is written, research-heavy, and repetitive, which is exactly the profile generative models handle. The damage lands on the training pipeline before it lands on the partners.</description><pubDate>Fri, 24 Jul 2026 00:00:00 GMT</pubDate></item><item><title>When Period Data Becomes Legal Evidence</title><link>https://gregoryfriedman.com/essays/when-period-data-becomes-legal-evidence/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/when-period-data-becomes-legal-evidence/</guid><description>Menstrual tracking apps sit outside HIPAA, so the data they retain can cross from the health context into the legal one. Nothing in the information lifecycle stops it.</description><pubDate>Fri, 24 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Hash tables, or: it’s too slow!</title><link>https://gregoryfriedman.com/lectures/data-structures/hash-tables/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/hash-tables/</guid><description>The Mapping ADT, a linear scan that will not do, and the trick that fixes it: let the key compute its own address, then keep the table roomy enough that it stays true.</description><pubDate>Wed, 22 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Heaps and priority queues</title><link>https://gregoryfriedman.com/lectures/data-structures/heaps-priority-queues/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/heaps-priority-queues/</guid><description>Serve the most urgent item first: the heap keeps just enough order to do that, hides a tree inside a flat array, and builds itself from scratch in linear time.</description><pubDate>Wed, 22 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Trees, BSTs, and balance</title><link>https://gregoryfriedman.com/lectures/data-structures/trees-bsts-balance/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/trees-bsts-balance/</guid><description>A tree is a list that branched: the search tree property, the four walks, and the rotations that keep the height logarithmic when the insertion order is out to get you.</description><pubDate>Wed, 22 Jul 2026 00:00:00 GMT</pubDate></item><item><title>A mental model for Python</title><link>https://gregoryfriedman.com/lectures/data-structures/a-mental-model-for-python/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/a-mental-model-for-python/</guid><description>Names are labels tied to objects, not boxes that hold them, and getting that picture right is the whole game.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Binary search and the quadratic sorts</title><link>https://gregoryfriedman.com/lectures/data-structures/binary-search-and-quadratic-sorts/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/binary-search-and-quadratic-sorts/</guid><description>Keeping a list sorted so search can be fast: binary search kept honest by its loop invariant, and insertion and selection sort as baselines.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Divide and conquer: mergesort, quicksort, quickselect</title><link>https://gregoryfriedman.com/lectures/data-structures/divide-and-conquer/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/divide-and-conquer/</guid><description>Split the problem in half, solve the pieces, then combine: how mergesort, quicksort, and quickselect turn sorting into O(n log n).</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Dynamic programming</title><link>https://gregoryfriedman.com/lectures/data-structures/dynamic-programming/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/dynamic-programming/</guid><description>Greedy, recursive, memoized, tabulated: the same problem solved four ways, and why remembering what you already computed turns an exponential tree into a straight line.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>How fast is it, really?</title><link>https://gregoryfriedman.com/lectures/data-structures/how-fast-is-it-really/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/how-fast-is-it-really/</guid><description>Timing programs, counting the dominant operation, and reading complexity classes off a log-log plot: big-O without hand-waving.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Linked lists and the wrapper pattern</title><link>https://gregoryfriedman.com/lectures/data-structures/linked-lists/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/linked-lists/</guid><description>Give up random access, buy cheap splicing, and let a wrapper class store the length and the tail so the ends stay O(1).</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Recursion and the call stack</title><link>https://gregoryfriedman.com/lectures/data-structures/recursion-and-the-call-stack/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/recursion-and-the-call-stack/</guid><description>Induction you can run: the recursive leap of faith, the call stack it rides on, and Fibonacci and Euclid as the two traces to know by heart.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Stacks, queues, and the idea of an ADT</title><link>https://gregoryfriedman.com/lectures/data-structures/stacks-queues-adts/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/stacks-queues-adts/</guid><description>One interface, two backings: a stack and a queue built from scratch, and why a list makes a slow queue but a deque does not.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Testing first</title><link>https://gregoryfriedman.com/lectures/data-structures/testing-first/</link><guid isPermaLink="true">https://gregoryfriedman.com/lectures/data-structures/testing-first/</guid><description>Write the test before the code: unittest, the red–green loop, and why you test the contract an ADT promises rather than its plumbing.</description><pubDate>Tue, 21 Jul 2026 00:00:00 GMT</pubDate></item><item><title>Debt and Dessert: Economic Exploitation in the Caribbean</title><link>https://gregoryfriedman.com/essays/debt-and-dessert/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/debt-and-dessert/</guid><description>Two documentaries about the same wound. &quot;Life and Debt&quot; names what neoliberalism did to Jamaica; &quot;Nothing Like Chocolate&quot; builds an alternative in Grenada. Both leave the geopolitics out.</description><pubDate>Wed, 19 Jun 2024 00:00:00 GMT</pubDate></item><item><title>Toxic Stress and the Body</title><link>https://gregoryfriedman.com/essays/toxic-stress-and-the-body/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/toxic-stress-and-the-body/</guid><description>A literature review in psychoneuroimmunology. Short-term stress sharpens the immune response; chronic stress degrades it, and the duration of the exposure turns out to be the variable that decides which.</description><pubDate>Tue, 11 Jun 2024 00:00:00 GMT</pubDate></item><item><title>iPhorgot the Consequences: Defamiliarization of the iPhone</title><link>https://gregoryfriedman.com/essays/iphorgot-the-consequences/</link><guid isPermaLink="true">https://gregoryfriedman.com/essays/iphorgot-the-consequences/</guid><description>One device, three populations: lithium miners in the Atacama, assembly workers in Shenzhen, and consumers in the United States. The same two costs recur at every stage, and they are not evenly distributed.</description><pubDate>Tue, 04 Jun 2024 00:00:00 GMT</pubDate></item></channel></rss>