Archive - Error Bars

Your Vendor's Benchmark Is Not Your Workload

Two points on SWE-bench is not a decision.

Jul 27 • Buschleague

The Battery, Drop 1: Four Fixtures You Can Run on Your Own Agent Right Now

Four fixtures, four difficulty tiers, fully inline. Run them against your own agent this afternoon.

Jul 27 • Buschleague

The Argument Was Never About the Model. It Was About the Sample Size.

Three timepoints, nine runs, and what happens to a launch-day argument once you put error bars on both sides of it.

Jul 27 • Buschleague

Welcome to Error Bars

Receipts, not vibes.

Jul 27 • Buschleague

#nojs-banner { position: fixed; bottom: 0; left: 0; padding: 16px 16px 16px 32px; width: 100%; box-sizing: border-box; background: red; color: white; font-family: -apple-system, "Segoe UI", Roboto, Helvetica, Arial, sans-serif, "Apple Color Emoji", "Segoe UI Emoji", "Segoe UI Symbol"; font-size: 13px; line-height: 13px; } #nojs-banner a { color: inherit; text-decoration: underline; } This site requires JavaScript to run correctly. Please turn on JavaScript or unblock scripts