loop-benchmarking

Controlled experiments across agentic coding configurations. Same task, one variable, what actually works.
git clone https://git.shiptheloop.com/loop-benchmarking.git
Log | Files | Refs | README

commit 3012989bb80dca8980569effc60dd0bd59e283c3
parent 58c112b941608fed3c655c638c0e4c17daa5bb19
Author: Brian Graham <brian@buildingbetterteams.de>
Date:   Thu,  9 Apr 2026 20:22:15 +0200

Update calibration: 9805c24a (broken rotation, bad randomizer),
cbbff570 (mostly works, spurious line clear, weird preview)

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

Diffstat:
Mtasks/tetris/eval/gameplay-bot/calibration/9805c24a.json | 41++++++++++++++++++++---------------------
Mtasks/tetris/eval/gameplay-bot/calibration/cbbff570.json | 49++++++++++++++++++++++++-------------------------
2 files changed, 44 insertions(+), 46 deletions(-)

diff --git a/tasks/tetris/eval/gameplay-bot/calibration/9805c24a.json b/tasks/tetris/eval/gameplay-bot/calibration/9805c24a.json @@ -2,32 +2,32 @@ "run_id": "tetris_arch=none_ctx=none_noise=clean_dsgn=none_eff=high_echk=none_hlang=en_lang=ts_lint=on_budget=low_model=haiku45_pw=avail_prompt=simple_rndr=none_strat=usub_tst=none_tedit=off_tglob=on_tgrep=on_tread=on_twrite=on_web=on_run2", "short_id": "9805c24a", "label": "DOM game (haiku-4.5, en)", - "notes": "", - "human_tested_at": "", + "notes": "Very ugly misaligned UI. Rotation partially broken: only does 1 of 4 rotations for some blocks. Poor randomizer: first 8-10 blocks were only 2 different tetrominoes. Next piece preview and game over display present. Soft drop works correctly.", + "human_tested_at": "2026-04-09", "human_tests": { - "game_loads": null, - "game_starts": null, - "auto_drop": null, - "move_left": null, - "move_right": null, - "move_down": null, - "rotate": null, - "hard_drop": null, - "all_pieces_rotate": null, + "game_loads": true, + "game_starts": true, + "auto_drop": true, + "move_left": true, + "move_right": true, + "move_down": true, + "rotate": false, + "hard_drop": true, + "all_pieces_rotate": false, "piece_locks": null, - "new_piece_spawns": null, - "multiple_pieces": null, + "new_piece_spawns": true, + "multiple_pieces": true, "line_clear": null, "score_changes": null, - "game_over": null, - "playable_30s": null, + "game_over": true, + "playable_30s": true, "multi_line_clear": null, "score_scaling": null, "level_progression": null, "speed_progression": null, - "next_piece_preview": null, - "game_over_display": null, - "counter_clockwise_rotation": null, - "soft_drop_distinct": null + "next_piece_preview": true, + "game_over_display": true, + "counter_clockwise_rotation": false, + "soft_drop_distinct": true } -} -\ No newline at end of file +} diff --git a/tasks/tetris/eval/gameplay-bot/calibration/cbbff570.json b/tasks/tetris/eval/gameplay-bot/calibration/cbbff570.json @@ -2,32 +2,32 @@ "run_id": "tetris_arch=none_ctx=none_noise=clean_dsgn=none_eff=high_echk=none_hlang=en_lang=ts_lint=on_budget=low_model=haiku45_pw=avail_prompt=detailed_rndr=none_strat=usub_tst=none_tedit=on_tglob=on_tgrep=on_tread=on_twrite=on_web=on_run1", "short_id": "cbbff570", "label": "DOM game (haiku-4.5, en)", - "notes": "", - "human_tested_at": "", + "notes": "Mostly works well. Next piece preview shows a shaded box outline instead of the actual upcoming block shape. Line clear works but at one point randomly cleared a line incorrectly (spurious clear). CCW rotation, soft drop, multi-line clear all work.", + "human_tested_at": "2026-04-09", "human_tests": { - "game_loads": null, - "game_starts": null, - "auto_drop": null, - "move_left": null, - "move_right": null, - "move_down": null, - "rotate": null, - "hard_drop": null, - "all_pieces_rotate": null, - "piece_locks": null, - "new_piece_spawns": null, - "multiple_pieces": null, - "line_clear": null, - "score_changes": null, - "game_over": null, - "playable_30s": null, - "multi_line_clear": null, + "game_loads": true, + "game_starts": true, + "auto_drop": true, + "move_left": true, + "move_right": true, + "move_down": true, + "rotate": true, + "hard_drop": true, + "all_pieces_rotate": true, + "piece_locks": true, + "new_piece_spawns": true, + "multiple_pieces": true, + "line_clear": true, + "score_changes": true, + "game_over": true, + "playable_30s": true, + "multi_line_clear": true, "score_scaling": null, "level_progression": null, "speed_progression": null, - "next_piece_preview": null, - "game_over_display": null, - "counter_clockwise_rotation": null, - "soft_drop_distinct": null + "next_piece_preview": true, + "game_over_display": true, + "counter_clockwise_rotation": true, + "soft_drop_distinct": true } -} -\ No newline at end of file +}

Impressum · Datenschutz