Sign in Subscribe

Gaurav

https://gsood.com

PageRank Measures the Wrong Fixed Point Under Polarization

Directly evaluating information quality at web scale is intractable. Early web search adopted PageRank¹ as a scalable surrogate: rather than grade content, infer a site's credibility; rather than measure credibility directly, use inter-site links as votes. The chain—quality ← credibility ← links—stacks two proxies. It works when the

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

The Undersupply Theorem Meets the Extraction Equilibrium

Meta Perspective "If you think about the pieces of advertising, there’s content creation, the creative, there’s the targeting, ... and there’s the measurement and probably the first pieces that we started building were the measurement..." https://stratechery.com/2025/an-interview-with-meta-ceo-mark-zuckerberg-about-ai-and-the-evolution-of-social-media/ Targeting: "Over the last 5

Bias as a Fix for Congestion

Economists describe overloaded labor and admissions markets as congested. Yet the term is usually deployed to mean “too many applicants,” glossing over the deeper constraint: decision‑makers cannot extract enough reliable information from superficially similar files without incurring prohibitive cost. I call this bottleneck evaluation congestion. The concept explains why

A Textbook Paradox: The Puzzle of Price Dispersion in E-Books

The same e-book sells at very different prices on different platforms. For instance, the e-book version of The Supreme Court and the Attitudinal Model Revisited sells at \$25.57 on Amazon and \$39.99 on Google. (You can also buy a DRM-Free version for $39.99 on CUP). We can&

Scan to Save: Arbitrage Between Paper and Digital

Remember when e‑books were supposed to undercut paper editions—no printing, warehousing, shipping, or returns to pay for? Fast‑forward ten years, and you’ll often pay more for a DRM‑locked Kindle file than for a used paperback plus scanning. Case in point: * The Supreme Court and the

Resolution Lost: The Deadweight Costs of Strict Isotonicity

Calibration aligns model scores with event frequencies. For a binary outcome $Y\in{0,1}$ and a score $S$, the calibration function is $g(s)=\mathbb{E}[Y\mid S=s]$. Post hoc calibration estimates $g$ on a holdout set and applies the estimate to future scores. A common difficulty

Uncorrelated Tariffs

We know the means look a little off (not to say anything about the ends): The EU, like the United States, has generally low tariffs; the average tariff it charges on US goods is less than 3 percent. So where does this 39 percent number come from? I have no

(Don't) Feed the Robots

"My site [with] modest traffic (~1,000 visitors per day) went from serving ~1GB/hr (to tons of AI crawlers!) to ~50MB/hr in traffic" Can you tell what was the point where I 1. Manually blocked a few AI crawlers 2. Moved my nameserver to Cloudflare plus