Farm Fresh Insights

Sign in Subscribe

Latest — 18 Jun 2025

Optimizing Early Trajectories in K-Means Clustering: Lookahead Initialization for K-Means

K-Means performance depends heavily on how clusters are initialized. While k-means++ improves over random starts by spreading centroids apart, it’s still greedy and can lock into suboptimal configurations—especially in noisy or high-dimensional data. This post explores a simple tweak: lookahead initialization. For each candidate seed, we simulate a

More issues

Greedy Over Longer Horizons: Look Ahead CART

Some machine learning algorithms, particularly tree-based and sequential selection methods, make locally optimal decisions at each step. Classification and Regression Trees (CART) exemplify this approach: at each node, the algorithm evaluates all possible splits and chooses the one that provides the best immediate improvement in some criterion like Gini impurity

The Invisible Hand Needs a Hand: Two Short Vignettes on Market Design

Top Choice On LinkedIn (Premium?), users can mark up to three jobs per month as a “Top Choice,” making them 43% more likely to hear back. (Note: Given that the relevant base rate is likely close to 1%, a 43% relative increase means going from 1 to 1.43 per

Credit Where It Is Due: Two Facts About Credit Cards

The Credit Card Market Has Become More Concentrated Over Time In 1987, the top 10 credit card issuers accounted for just about two-fifths of the market share based on outstanding balances. In 2024, "A relatively small group of card issuers holds most of the outstanding credit card balances, with

Problem Solving

The first rule of problem-solving is to diagnose before looking for solutions. Though like all rules, the exception proves it. Sometimes it is cheaper to try out solutions to diagnose and solve a problem. For instance, to diagnose allergies, you could give an antihistamine. If the symptoms improve, the likely

The Modal Social Science Dataset is Under-Analyzed

When estimates can be combined to yield a useful, consistent, asymptotically unbiased estimate, should theory limit which analyses we perform?

Internet Search is Not a Naive Information Retrieval Problem

"During RL training, we employ a curriculum-based rollout strategy that incrementally degrades the quality of generated documents, progressively eliciting the model’s reasoning ability by exposing it to increasingly challenging retrieval scenarios. Extensive experiments demonstrate that ZEROSEARCH effectively incentivizes the search capabilities of LLMs using a 3B LLM as

Cell Mates: Extracting Useful Information From Tables for LLMs

We seem to have cracked the art of distilling information in words and images using large machine learning models. However, our ability to exploit useful information in tabular data using large models is mostly missing. The upshot is that LLMs don't largely encode the knowledge from these tabular