Cross-Context Review (CCR) improves LLM verification by separating production and review into independent sessions. A natural extension is multi-turn review: letting the reviewer ask follow-up questions, receive author responses, and review again. We call this Dynamic Cross-Context Review (D-CCR). In a controlled experiment with 30 artifacts and 150 injected errors, we tested four D-CCR variants against the single-pass CCR baseline. Single-pass CCR (F1 = 0.376) significantly outperformed all multi-turn variants, including D-CCR-2b with question-and-answer exchange (F1 = 0.303, p<0.001, d=−0.59). Multi-turn review increased recall (+0.08) but generated 62% more false positives (8.5 vs. 5.2), collapsing precision from 0.30 to 0.20. Two mechanisms drive this degradation: (1) false positive pressure -- reviewers in later rounds fabricate findings when the artifact's real errors have been exhausted, and (2) Review Target Drift -- reviewers provided with prior Q&A exchanges shift from reviewing the artifact to critiquing the conversation itself. Independent re-review without prior context (D-CCR-2c) performed worst (F1 = 0.263), confirming that mere repetition degrades rather than helps. The degradation stems from false positive pressure in additional rounds, not from information amount -- within multi-turn conditions, more information actually helps (D-CCR-2b > D-CCR-2a). The problem is not what the reviewer sees, but that reviewing again invites noise.
Figures & tables
Condition
Round 1
Round 2
Purpose
CCR-1
Artifact only
—
Baseline
D-CCR-2a
Artifact only
Artifact + questions
Question-only
D-CCR-2b
Artifact only
Artifact + Q&A
Full D-CCR
D-CCR-2c
Artifact only
Artifact only (fresh)
Independence
Table 1
Condition
Findings
TP
FP
Dup
Precision
Recall
F1
F1 SD
CCR-1
9.3
2.64
5.23
1.43
0.297
0.529
0.376
0.136
D-CCR-2a
15.4
2.96
9.17
3.27
0.197
0.591
0.293
0.102
D-CCR-2b
15.2
3.03
8.47
3.70
0.204
0.607
0.303
0.110
D-CCR-2c
18.4
3.10
9.70
5.60
0.168
0.620
0.263
0.091
Table 1: Main experiment results ( N=30 artifacts). Findings = TP + FP + Dup, where Dup counts findings matching a ground-truth error already identified by another finding (redundant true positives). Duplicate rates increase with rounds (15% for CCR-1 vs. 30% for D-CCR-2c).
Figure 1: F1, Precision, and Recall by condition. CCR-1 achieves the highest F1 and Precision. Multi-turn conditions increase Recall but collapse Precision, resulting in lower F1.
Table 3: Round 2 discovery metrics. R1 TP values vary slightly across conditions (2.61–2.67) due to probabilistic model behavior. N(R2 new > 0) shows how many artifacts had any new true positive in round 2.
Category
CCR-1
2a
2b
2c
Code
0.354
0.308
0.284
0.255
Document
0.446
0.334
0.361
0.318
Script
0.328
0.238
0.265
0.217
Table 4: F1 by artifact category.
Figure 2: False positive accumulation across conditions. The dashed line marks CCR-1’s FP level. All multi-turn conditions add substantial Round 2 FPs (red), with D-CCR-2c generating the most noise.
Automated peer review uses computational systems, especially language models, to assess research papers, generate feedback, and support editorial decisions.
Large language models struggle to catch errors in their own outputs when the review happens in the same session that produced them. This paper introduces Cross-Context Review (CCR), a straightforward method where the review is conducted in a fresh session with no access to the production conversation history. We ran a controlled experiment: 30 artifacts (code, technical documents, presentation scripts) with 150 injected errors, tested under four review conditions -- same-session Self-Review (SR), repeated Self-Review (SR2), context-aware Subagent Review (SA), and Cross-Context Review (CCR). The central result is that a second review helps only when it happens in a fresh session: CCR (F1 28.6%) outperforms a second review in the same session (SR2, 21.7%) robustly, both in the first run (paired t, p<0.001) and in the three-run average (Holm-adjusted p=0.004). This version updates the broader comparisons. Averaged across runs, and excluding one SR run whose records could not be verified, CCR is not significantly ahead of context-aware subagent review (SA, 23.8%; p=0.057) or of a single same-session review (SR, 27.1%; p=0.26); the first version's advantages over these two baselines came from run 1. CCR needs no infrastructure and costs one extra session.
Large language models now generate code, documentation, and analyses, and are increasingly used to review such output. We ask when a second review by a different model helps. Building on the author's earlier preprints, which varied context, repetition, and role structure within one model, we test model independence in a controlled experiment: 30 artifacts with 150 planted errors, 10 review conditions, and 900 review sessions with three reviewer models from two developers. In this experiment, (1) a top-tier cross-model reviewer is not significantly different in F1 from same-model review in a fresh session (CCR), which does not establish equivalence; (2) the two find partly different errors (Jaccard 41.2%); and (3) at two review calls, one CCR plus one cross-model review matches more planted errors than two CCR reviews (56.7% vs. 42.7%; Holm-adjusted p=.006), but not significantly more than two reviews by the top-tier cross-model reviewer, so model difference and reviewer capability are not separated. A lightweight cross-model reviewer scores no higher than same-model review. Withholding requirements from the reviewer raises F1 for the two lower tiers but not the top tier, in untested point estimates whose pattern depends on how failed sessions are scored. Before analysis we audited all session records, excluding one baseline run of uncertain provenance and 14 failed calls; results with all sessions are also reported. A partial check on public detector outputs from another benchmark neither replicates nor contradicts the main comparison. Records, artifacts, and scripts are available from the author on request.
LLM coding benchmarks face a credibility crisis: widespread solution leakage and test quality issues undermine SWE-bench Verified, while existing detection methods--paraphrase consistency, n-gram overlap, perplexity analysis--never directly observe whether a model reasons or recalls. Meanwhile, simply repeating verification degrades accuracy: multi-turn review generates false positives faster than it discovers true errors, suggesting that structural approaches are needed. We introduce Cross-Context Verification (CCV), a black-box method that solves the same benchmark problem in N independent sessions and measures solution diversity, combined with the Hierarchical Cross-Context Architecture (HCCA), a multi-agent analysis framework that prevents confirmation bias through intentional information restriction across specialized analytical roles. On 9 SWE-bench Verified problems (45 trials, Claude Opus 4.6, temperature 0), CCV achieves perfect separation between contaminated and genuine reasoning (Mann-Whitney U=0, p approx 0.012, r = 1.0). Key findings: (1) contamination is binary--models either recall perfectly or not at all; (2) reasoning absence is a perfect discriminator; (3) 33% of prior contamination labels are false positives; (4) HCCA's independent analysis structure discovers contamination-flaw composite cases that single-analyst approaches miss. A pilot experiment extending HCCA to multi-stage verification (Worker to Verifier to Director) yields a negative result--100% sycophantic confirmation--providing further evidence that information restriction, not structural complexity, is the key mechanism. We release all code and data.