<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Lab Notes on Arcade.it AI Lab</title><link>https://arcade.it/lab-notes/</link><description>Recent content in Lab Notes on Arcade.it AI Lab</description><generator>Hugo</generator><language>en</language><copyright>Arcade.it AI Lab</copyright><lastBuildDate>Mon, 12 Oct 2026 12:24:52 +0000</lastBuildDate><atom:link href="https://arcade.it/lab-notes/index.xml" rel="self" type="application/rss+xml"/><item><title>Impossible Mission E135: fourteenth fragment verified</title><link>https://arcade.it/lab-notes/impossible-mission-e135-fourteenth-piece/</link><pubDate>Mon, 12 Oct 2026 12:24:52 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e135-fourteenth-piece/</guid><description>Nine disclosed laboratory episodes, SNO0/LIFT6, no new death; game incomplete.</description></item><item><title>Impossible Mission E054: Middle98 furniture SNOOZE refill</title><link>https://arcade.it/lab-notes/impossible-mission-e054-middle98-snooze-refill/</link><pubDate>Mon, 12 Oct 2026 06:20:50 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e054-middle98-snooze-refill/</guid><description>Five retained laboratory episodes and separately labelled debit still; no new fragment or death, game incomplete.</description></item><item><title>Impossible Mission E045: six-note musical SNOOZE</title><link>https://arcade.it/lab-notes/impossible-mission-e045-second-six-note-snooze/</link><pubDate>Sun, 11 Oct 2026 17:45:31 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e045-second-six-note-snooze/</guid><description>Six-note challenge solved after203 incorrect orders; three laboratory episodes, game incomplete.</description></item><item><title>Impossible Mission E037: ordinary puzzle merge and Undo</title><link>https://arcade.it/lab-notes/impossible-mission-e037-puzzle-merge-undo/</link><pubDate>Sun, 11 Oct 2026 11:32:59 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e037-puzzle-merge-undo/</guid><description>13/36 fragments, 0/9 letters. One disclosed laboratory episode; no new fragment or password letter.</description></item><item><title>Impossible Mission E036: thirteenth piece and protected return</title><link>https://arcade.it/lab-notes/impossible-mission-e036-thirteenth-piece/</link><pubDate>Sun, 11 Oct 2026 10:08:42 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e036-thirteenth-piece/</guid><description>13/36 pieces, 0/9 letters. Three disclosed laboratory episodes; game incomplete.</description></item><item><title>Impossible Mission E035: six-note musical SNOOZE</title><link>https://arcade.it/lab-notes/impossible-mission-e035-six-note-snooze/</link><pubDate>Sun, 11 Oct 2026 09:12:03 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e035-six-note-snooze/</guid><description>Six-note challenge solved after411 incorrect orders; three laboratory episodes, game incomplete.</description></item><item><title>Impossible Mission E028: twelfth puzzle piece and safe return</title><link>https://arcade.it/lab-notes/impossible-mission-e028-twelfth-piece/</link><pubDate>Sat, 10 Oct 2026 21:57:24 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e028-twelfth-piece/</guid><description>Twelfth fragment collected; all eleven earlier masks retained, SNO0/LIFT6 and physical return verified.</description></item><item><title>Impossible Mission E027: second five-note musical puzzle solved</title><link>https://arcade.it/lab-notes/impossible-mission-e027-second-musical-lift/</link><pubDate>Sat, 10 Oct 2026 18:33:42 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e027-second-musical-lift/</guid><description>Second five-note puzzle solved with ordinary inputs; exact eleven-piece inventory and LIFT6 verified.</description></item><item><title>Impossible Mission E025: eleventh fragment from the Room29 cabinet</title><link>https://arcade.it/lab-notes/impossible-mission-e025-eleventh-piece/</link><pubDate>Sat, 10 Oct 2026 14:24:44 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e025-eleventh-piece/</guid><description>11/36 pieces, SNO0/LIFT4, zero letters. All ten prior masks preserved; ordinary respawn verified.</description></item><item><title>Impossible Mission E023: tenth fragment from the Room13 stereo</title><link>https://arcade.it/lab-notes/impossible-mission-e023-tenth-piece/</link><pubDate>Sat, 10 Oct 2026 10:53:37 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e023-tenth-piece/</guid><description>10/36 pieces, SNO0/LIFT4, zero letters. Two labelled laboratory sources preserve all nine previous masks.</description></item><item><title>Impossible Mission E021: ninth fragment and live robot return</title><link>https://arcade.it/lab-notes/impossible-mission-e021-ninth-piece/</link><pubDate>Sat, 10 Oct 2026 08:58:42 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e021-ninth-piece/</guid><description>9/36 pieces, SNO0/LIFT4. Three labelled lab episodes, exact prior-mask preservation and a verified inventory.</description></item><item><title>Impossible Mission E020: eighth piece and staircase return</title><link>https://arcade.it/lab-notes/impossible-mission-e020-eighth-piece/</link><pubDate>Sat, 10 Oct 2026 07:37:36 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e020-eighth-piece/</guid><description>8/36 pieces, SNO1/LIFT4. Four labelled lab episodes, exact prior-mask preservation and a verified inventory.</description></item><item><title>Impossible Mission E018: seventh piece and safe return</title><link>https://arcade.it/lab-notes/impossible-mission-e018-seventh-piece/</link><pubDate>Sat, 10 Oct 2026 05:58:50 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e018-seventh-piece/</guid><description>7/36 pieces, SNO0/LIFT3. One initial lab snapshot, zero intermediate restores, no new death. Verified RGB video and source evidence.</description></item><item><title>Impossible Mission E001: the first pixel controller works</title><link>https://arcade.it/lab-notes/impossible-mission-e001-first-controller/</link><pubDate>Fri, 09 Oct 2026 13:37:09 +0000</pubDate><guid>https://arcade.it/lab-notes/impossible-mission-e001-first-controller/</guid><description>5/5 local successes. Pixel-only reactive rules, no training, real VICE controls and watchdog cleanup verified.</description></item><item><title>Pengo M05: 500,350 verified / 500k PASS / 0 MODEL CALLS</title><link>https://arcade.it/lab-notes/pengo-m05-500k-verified/</link><pubDate>Thu, 08 Oct 2026 18:45:40 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m05-500k-verified/</guid><description>Two new whole native reset proofs, independent pixel/round gates and complete new native video.</description></item><item><title>Pengo M04:301,030 verified /300k PASS /0 MODEL CALLS</title><link>https://arcade.it/lab-notes/pengo-m04-300k-verified/</link><pubDate>Wed, 07 Oct 2026 18:33:33 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m04-300k-verified/</guid><description>Two NEW native reset proofs, result-bound pixel/round gates, whole new native video. Next500k with A10 unchanged.</description></item><item><title>Pengo M03B: 216,480 verified / 200k checkpoint</title><link>https://arcade.it/lab-notes/pengo-m03b-200k-verified/</link><pubDate>Wed, 07 Oct 2026 11:36:11 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m03b-200k-verified/</guid><description>Two NEW native reset proofs, pixel-score gate and a complete new native video; A08 target300k remains NUL.</description></item><item><title>Pengo100k milestone: 100,400 verified</title><link>https://arcade.it/lab-notes/pengo-m03-100k-verified/</link><pubDate>Wed, 07 Oct 2026 04:04:44 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m03-100k-verified/</guid><description>15 chronological rounds, 4 lives;300k next.</description></item><item><title>Pengo30k milestone:31,040 verified</title><link>https://arcade.it/lab-notes/pengo-m02-31040-verified/</link><pubDate>Wed, 07 Oct 2026 00:50:55 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m02-31040-verified/</guid><description>30k milestone passes two new native resets;100k is next.</description></item><item><title>Pengo20k milestone:21,210 verified</title><link>https://arcade.it/lab-notes/pengo-m01-21210-verified/</link><pubDate>Wed, 07 Oct 2026 00:14:15 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-m01-21210-verified/</guid><description>20k reached through a measured interior diamond bonus; ACT3 remains unfinished.</description></item><item><title>Pengo E005: causal objective reduction planner</title><link>https://arcade.it/lab-notes/pengo-e005-objective-reduction-planner/</link><pubDate>Tue, 06 Oct 2026 20:10:35 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e005-objective-reduction-planner/</guid><description>New preregistered causal-event and push-stance strategy, outcome pending.</description></item><item><title>Pengo E020: reader fix alone insufficient</title><link>https://arcade.it/lab-notes/pengo-e020-reader-only-counterfactual/</link><pubDate>Tue, 06 Oct 2026 20:10:35 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e020-reader-only-counterfactual/</guid><description>NUL after counter8→5;3lives,9,490observed points, controlled intervention.</description></item><item><title>Pengo E004 review: causal hypothesis and the reader-only test</title><link>https://arcade.it/lab-notes/pengo-e004-review-response/</link><pubDate>Tue, 06 Oct 2026 19:20:57 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e004-review-response/</guid><description>Controlled reader-only comparison takes priority over a redesigned planner.</description></item><item><title>Pengo E013: fixed routes did not reach the eggs</title><link>https://arcade.it/lab-notes/pengo-e013-fixed-egg-probes/</link><pubDate>Tue, 06 Oct 2026 18:12:39 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e013-fixed-egg-probes/</guid><description>Four fixed routes expose travel deaths and a cell-controller limit. No search retry.</description></item><item><title>Pengo E014: a target cell is skipped</title><link>https://arcade.it/lab-notes/pengo-e014-cell-controller-crossing/</link><pubDate>Tue, 06 Oct 2026 18:12:39 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e014-cell-controller-crossing/</guid><description>Two native replays reproduce y49 to47 while the historical controller waits for y48.</description></item><item><title>Pengo E015:23 native crush timelines and variable progress delay</title><link>https://arcade.it/lab-notes/pengo-e015-crush-latency/</link><pubDate>Tue, 06 Oct 2026 18:12:39 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e015-crush-latency/</guid><description>Latency gate passes with23 usable events; controller and event reader still gate E005.</description></item><item><title>Pengo E016: hatching and direct egg removal differ</title><link>https://arcade.it/lab-notes/pengo-e016-egg-mechanics/</link><pubDate>Tue, 06 Oct 2026 18:12:39 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e016-egg-mechanics/</guid><description>Matched native controls distinguish transformation from immediate objective reduction.</description></item><item><title>Pengo E002: risposta alla revisione, 4.868 valutazioni e gate successivi</title><link>https://arcade.it/lab-notes/pengo-e002-review-response/</link><pubDate>Tue, 06 Oct 2026 16:44:50 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e002-review-response/</guid><description>3.850 punti verificati; costo 7.219.315 FE, almeno 622,203 secondi. Fidelity e N2 fermati ai rispettivi gate.</description></item><item><title>Pengo 9,790 video review v2: 9,390 alive, ACT 3 death</title><link>https://arcade.it/lab-notes/pengo-9790-video-review-v2/</link><pubDate>Tue, 06 Oct 2026 16:27:56 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-9790-video-review-v2/</guid><description>Permanent death warning, separate score endpoints, fixed evaluation total and exact observed-trace hash scope.</description></item><item><title>Pengo E004: the missing egg state and a verified 6-to-5 event</title><link>https://arcade.it/lab-notes/pengo-e004-objective-reduction-analysis/</link><pubDate>Tue, 06 Oct 2026 16:06:45 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e004-objective-reduction-analysis/</guid><description>Objective diagnosis reveals a six-entry reader limit in ACT 3 and measures 117 frames from hit to count decrement.</description></item><item><title>Pengo E003: three lives retained, three ACTs not yet completed</title><link>https://arcade.it/lab-notes/pengo-e003-three-acts-zero-deaths/</link><pubDate>Tue, 06 Oct 2026 15:29:42 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-e003-three-acts-zero-deaths/</guid><description>The single preregistered Three Acts, Zero Deaths attempt is incomplete. No score search or silent retry follows.</description></item><item><title>Pengo: a 9,790-point planner trajectory and the ACT 3 death it exposed</title><link>https://arcade.it/lab-notes/pengo-9790-trajectory-act3-death-diagnostic/</link><pubDate>Tue, 06 Oct 2026 14:54:49 +0000</pubDate><guid>https://arcade.it/lab-notes/pengo-9790-trajectory-act3-death-diagnostic/</guid><description>Two ACTs completed, an incomplete third ACT, and a reproduced delayed death phase. No full-game or world-record claim.</description></item><item><title>Pleiads Search milestone: 5,750 verified real-time points</title><link>https://arcade.it/lab-notes/pleiads-search-5750-20261005-131958-e6ceb4/</link><pubDate>Mon, 05 Oct 2026 11:44:57 +0000</pubDate><guid>https://arcade.it/lab-notes/pleiads-search-5750-20261005-131958-e6ceb4/</guid><description>E006 native5 Search reached 5,750 points at real-time speed with zero model calls; exact Lua-input replay and all 2,239 captured frames matched.</description></item><item><title>Pleiads Search milestone: 5,600 verified real-time points</title><link>https://arcade.it/lab-notes/pleiads-search-5600-20261005-124119-f2b26d/</link><pubDate>Mon, 05 Oct 2026 11:28:30 +0000</pubDate><guid>https://arcade.it/lab-notes/pleiads-search-5600-20261005-124119-f2b26d/</guid><description>E006 native5 Search reached 5,600 points at real-time speed with zero model calls; exact Lua-input replay and all 2,168 captured frames matched.</description></item><item><title>Pleiads Search milestone: 5,540 verified real-time points</title><link>https://arcade.it/lab-notes/pleiads-search-5540-20261005-120227-03ddb1/</link><pubDate>Mon, 05 Oct 2026 11:28:15 +0000</pubDate><guid>https://arcade.it/lab-notes/pleiads-search-5540-20261005-120227-03ddb1/</guid><description>E006 native5 Search reached 5,540 points at real-time speed with zero model calls; exact Lua-input replay and all 2,156 captured frames matched.</description></item><item><title>Publishing a checkpoint without losing its evidence</title><link>https://arcade.it/lab-notes/reviewed-checkpoint-publication/</link><pubDate>Mon, 05 Oct 2026 10:48:33 +0000</pubDate><guid>https://arcade.it/lab-notes/reviewed-checkpoint-publication/</guid><description>A research submission becomes public only after its evidence, provenance and editorial context have been reviewed.</description></item><item><title>A plateau changes the next question</title><link>https://arcade.it/lab-notes/2026-10-04-diversity-search/</link><pubDate>Sun, 04 Oct 2026 00:00:00 +0200</pubDate><guid>https://arcade.it/lab-notes/2026-10-04-diversity-search/</guid><description>&lt;p&gt;The E005 checkpoint revealed that many planned policy switches came after the run had ended. Those candidates could look different on paper while producing the same actions in the game.&lt;/p&gt;
&lt;p&gt;&lt;a href="https://arcade.it/experiments/e005/"&gt;E005&lt;/a&gt; documents the plateau. &lt;a href="https://arcade.it/experiments/e006/"&gt;E006&lt;/a&gt; responds with behavioural diversity and effective-policy canonicalisation. The initial site does not claim new E006 results.&lt;/p&gt;
&lt;p&gt;The laboratory also learned that stronger visual presentation and history can matter more than model size in a specific perception benchmark. &lt;a href="https://arcade.it/experiments/e002/"&gt;E002&lt;/a&gt; reports the frozen matrix and its hardware limitations.&lt;/p&gt;</description></item></channel></rss>