Card: name all four pooled searches, describe the samples exactly, compute every number

README.md

@@ -30,7 +30,8 @@ The [CodeSearchNet Challenge](https://github.com/github/CodeSearchNet) (Husain e
 dataset treats every rated function of a language as one codebase and searches all of it: for each
 search, every function in its language is a candidate. The experts' ratings are kept, and the
 pairs they never rated but a search tool returned were rated on the same scale by a judge model,
-validated against the experts first. Random samples of the pairs no tool returned were rated too.
+validated against the experts first. Random samples of the pairs only a loose any-keyword grep
+returned, and of the pairs nothing returned, were rated too.
 
 | Language | Searches | Functions | Rated by experts | Rated by the judge | Relevant |
 |---|---:|---:|---:|---:|---:|
@@ -50,21 +51,21 @@ which describes the method, the judge's validation and the results in full.
   searches used to tune the benchmark's tools, `test` for all others).
 - `corpus`: one row per function: `id`, `language`, `code` (the rated lines at the rated commit),
   `url`, `repository`, `commit`, `path`, `first_line`, `last_line`, `license` (the repository's
-  SPDX license id as GitHub reports it in September 2026, `NOASSERTION` or `null`) and `sha256`
+  SPDX license id as GitHub reported it in September 2026, `NOASSERTION` or `null`) and `sha256`
   of `code`.
 - `qrels`: one row per rated search-function pair: `relevance` (the experts' mean rating, else the
   judge's rating, 0 to 3), `relevant` (`relevance` of 2 or more), `source` (`expert` or `judge`),
   `expert_ratings`, `judge_rating` (also given for expert-rated pairs, to check the judge), and
-  `rated_because` (`expert`, `pooled`: a search tool returned it, or `sample`: a random sample of
-  what the tools did not return).
+  `rated_because` (`expert`, `pooled`: a search tool returned it, or `sample`: one of the random
+  samples above).
 
 ## How the judge rated
 
-GPT-6 Astra through the Codex CLI, low reasoning effort, no tools, in batches of 40 pairs mixed
-across searches, seeing an opaque id, the search and the code (cut at 20,000 characters), never
-which tool returned the pair. The prompt gave the Challenge's 0-3 scale and its description of each
-level. Before any of its ratings were used, it rated every expert-rated pair of the language; the
-gate, set in advance, was an F1 of 0.67 on "relevant" against the experts:
+GPT-6 Astra through the Codex CLI, low reasoning effort, no tools, in batches of 40 pairs
+mixed across searches, seeing an opaque id, the search and the code (cut at 20,000 characters),
+never which tool returned the pair. The prompt gave the Challenge's 0-3 scale and its description of
+each level. Before any of its ratings were used, it rated every expert-rated pair of the language;
+the gate, set in advance, was an F1 of 0.67 on "relevant" against the experts:
 
 | Language | Expert-rated pairs | Same verdict as the experts | F1 | Gate |
 |---|---:|---:|---:|---|
@@ -75,12 +76,11 @@ gate, set in advance, was an F1 of 0.67 on "relevant" against the experts:
 | Ruby | 300 | 83% | 0.709 | passed |
 | Go | 162 | 80% | 0.629 | **missed** |
 
-Where two experts rated the same pair (848 Python pairs), they agree on 70% of verdicts, F1 0.711.
-In Go the judge agrees on most verdicts but calls more pairs relevant than Go's experts did, and it
-missed the gate on only 40 relevant expert pairs; use Go's judge ratings with that in mind.
+Where two experts rated the same pair (848 Python pairs), they agree on 70% of verdicts, F1 0.711. In Go the judge missed the gate on only 40 relevant expert pairs, and calls more pairs relevant (49) than its experts did (40); use its ratings there with that in mind.
 
-Pooled pairs came from three tools: grep with a pattern a coding agent wrote per search, jevpipe and
-DeepSeek V4.1 Flash.
+Pooled pairs came from four searches: grep with a pattern a coding agent wrote per search, grep
+for all of the search's keywords, jevpipe with a probability of 0.3 or more, and DeepSeek V4.1
+Flash answering yes or with a probability of 0.3 or more.
 
 ## Use it with care