The rest of the claims, measured (2026-07-30)

A companion to UPDATE-2026-07-30-engine-claim-and-site-corrections.md. That file recorded the one claim of theirs that was correct and the fix I shipped for it. This one records what happened when I measured the rest.

Ten items. Every figure here was measured against the artifacts the two repositories ship, not against my own working copy. Both bundled release archives hash to the values in their own SHA256SUMS.txt, so the artifacts are genuine and I measured the same bytes they did. Each item carries the command that reproduces it.

  1. GSE was MIT licensed for the whole period in which GRIP-EMS was designed and built
  2. Finding 5 overstates the export, in the build they hashed
  3. The legacy import block is quoted from the second line of the branch
  4. The Priority expansion is not GSE’s design
  5. Ten years, not 12 to 13
  6. Six counting errors
  7. Two of five rows in the scienter table are a third party’s words
  8. The predicate act, from their own three documents
  9. The memo drops context that their own capture supplies
  10. The site owner line

Items 7 and 9 are the Discord ones. Item 7 is about who said the words the scienter table attributes to the developer. Item 9 is about what their capture records around those words that the memo built on top of it leaves out.

Where this was supposed to go, and why it is here instead

I wrote these up as ten separate issues and two pull requests, one set per repository, so each item could be checked on its own and corrected in place. I could not file any of them.

From my account, GitHub returns HTTP 403 for both forking and issue creation on both repositories:

POST /repos/LarryThiessen/GSE-to-GRIP-EMS-Conversion-Copyright-Violations/issues
{"message":"Blocked","status":"403"}

POST /repos/LarryThiessen/GSE-Addon-vs-GRIP-EMS-Addon-Copyright-Violations/issues
{"message":"Blocked","status":"403"}

Both repositories report allow_forking: true, neither is archived or disabled, both are public, and public read and clone both work normally, which is how I measured everything below. A control fork of an unrelated public repository from the same account in the same minute succeeded, and the API rate limit was not near its ceiling. The block is specific to these repositories.

I am recording that rather than working around it. A pull request or an issue from a second account would be circumventing a platform control, and I am not doing that. So the corrections are published here, where the timestamps are checkable, and the delivery routes that remain open are the CurseForge claim thread, where the numeric errors below actually matter, and a direct message.

1. GSE was MIT licensed for the whole period in which GRIP-EMS was designed and built

CURSEFORGE-COMPLAINT-FIELDS.md:37 presents the licence as Copyright (c) 2026 Timothy Minahan. All Rights Reserved. with no date attached, and THE-STORY.md describes the conduct as reverse engineering “the free, All-Rights-Reserved addon”. The LICENSE file in TimothyLuke/GSE-Advanced-Macro-Compiler has exactly two commits in its whole history.

gh api "repos/TimothyLuke/GSE-Advanced-Macro-Compiler/commits?path=LICENSE" \
  --jq '.[] | [.sha[0:8], .commit.author.date] | @tsv'
gh api repos/TimothyLuke/GSE-Advanced-Macro-Compiler/contents/LICENSE?ref=9c15b61c \
  -H "Accept: application/vnd.github.raw" | head -3

9c15b61c, 2021-07-06, is the MIT License, “Copyright (c) 2016 Timothy Minahan”, granting permission “to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies”. 5f293d0a, 2026-04-14, replaces it with the All Rights Reserved text the filing quotes.

Date Event
2021-07-06 GSE LICENSE committed as MIT
2026-02-06 GSE wiki Priority page last edited, see item 4
2026-03-18 GRIP-EMS first commit, “Initial project skeleton”
2026-03-21 GRIP-EMS v1.0.4, the first release in their own version scan
2026-04-14 GSE relicensed to All Rights Reserved
2026-04-25 GSE introduces PlatformID, per their section 3
2026-04-30 the Discord thread their capture covers

The design period, the build period and the first twenty four days of public releases all sat inside the MIT window. I am stating dates, not arguing law. The sworn field text quotes a licence that took effect on 2026-04-14 without noting that the prior licence was MIT, which a reader needs in order to weigh the claim.

Against my own interest: the “ARR Licence” message in their capture is dated 2026-04-30, after the relicense, so it was accurate when it was written. The licence history does not touch that.

2. Finding 5 overstates the export, in the build they hashed

grip-cmi-evidence-exhibit.md:129 says the export “strips only locale fields”.

PrepareExportVersions, the function quoted, is at line 22 of Import/GRIPExport.lua and handles one version’s steps and actions. It never touches sequence metadata. That is assembled further down the same file, in GE.Export, and the share payload carries sixteen provenance and identity fields: originalAuthor, originalAuthorIdentity, originalAuthorRealm, originalCreatedAt, originalSignature, originalSignatureV2, signatureAlgorithm, lastModifier, lastModifierIdentity, lastModifierRealm, lastModifiedAt, modifierChain, forkedFrom, forkedFromChain, provenanceSource and privacyMode.

unzip -o evidence/GRIP-EMS-v2.3.5.zip -d v235
sed -n '22p;464p;544,559p' v235/GRIP-EMS/Import/GRIPExport.lua

In v2.3.5, the archive they hash, GE.Export is at 464 and that block is 544 to 559. In v2.3.16 it is 594 to 609 with GE.Export at 514. Before the payload is built, Identity:BuildExportPayload scrubs the local user’s own identity for privacy and leaves foreign-author entries alone, which is the opposite direction from stripping somebody else’s attribution.

The rest of that sentence is correct and I am not contesting it. There is no license check and nothing gates re-sharing on origin: grep -ric 'licens' returns zero across Import/*.lua and Engine/Transmission.lua. So the heading at line 127, “with no provenance or redistribution guard”, is half right. The redistribution half stands. The provenance half is the one that line 196 carries into the 1202 argument.

3. The legacy import block is quoted from the second line of the branch

grip-vs-gse-forensic-comparison.md:261-269 quotes five lines for v2.3.5 and :309-319 quotes seven for v2.3.16, under the heading “GRIP blanks the origin identity on GSE-legacy import”. Three things sit just outside the quote.

unzip -o evidence/GRIP-EMS-v2.3.16.zip -d v2316
sed -n '876,934p' v2316/GRIP-EMS/Import/LegacyImport.lua

Line 920 is else. The elseif branch above it, 878 to 919, carries every provenance field an incoming payload actually has plus a modifier chain entry; the quoted block is the other branch, for content arriving with no provenance at all, which is what a GSE export is. The first line of their own code block is seqData.originalAuthor = seqData.author or "Unknown (GSE legacy)", which keeps the author. Their quote ends at 927; line 928 is seqData.provenanceSource = "gse-legacy" and line 933 appends a modifier chain entry reading "imported-from-gse".

The empty strings inside the quote go to fields holding an EMS identity hash, an EMS realm and an EMS signature. A GSE sequence never carried any of the three, so there is nothing to preserve into them.

4. The Priority expansion is not GSE’s design

grip-vs-gse-forensic-comparison.md:48-54 calls the triangular expansion “GSE’s signature algorithm” and “GSE’s distinctive design”, “for over a decade”, and line 177 calls it “the single most probative point in this exhibit”.

Their reading of my code is right. SF:ExpandPriority does emit [1, 1, 2, 1, 2, 3] for three steps. The attribution is what does not hold, on three independent grounds.

It is in GSE’s own repository, in another addon’s form. spec/prioritycheck.lua on master carries the walk at lines 35 to 41.

gh api repos/TimothyLuke/GSE-Advanced-Macro-Compiler/contents/spec/prioritycheck.lua?ref=master \
  -H "Accept: application/vnd.github.raw" | sed -n '35,41p'
gh api "repos/TimothyLuke/GSE-Advanced-Macro-Compiler/commits?path=spec/prioritycheck.lua" \
  --jq '.[] | [.sha[0:8], .commit.author.date] | @tsv'

limit = limit or 1
if step == limit then
  limit = limit % #macros + 1
  step = 1
else
  step = step % #macros + 1
end

That file entered the repository on 2017-06-14, commit a2c17c99, “Priority checks and update for #302”. Note what it counts: #macros. GSE’s own engine at GSE/API/Storage.lua:2155-2171, the code their exhibit quotes as GSE’s implementation, counts #actionList. macros is GnomeSequencer’s table name.

It is semlar’s, published before GSE’s repository existed. GnomeSequencer, WoWInterface file 23234, author semlar, listing created 2014-10-29, 280,195 downloads. The listing reproduces ExampleSequences.lua inline and the reproduced file carries the same five statements with the same three variable names, under the comment “This example increments the step in the following order: 1 12 123 1234 etc. until it reaches the end and starts over”.

Stated against my own interest: the version on that page today is 7.3.0.1, updated 2017-08-31, so the inline listing is the 2017 file rather than provably the October 2014 upload. What the page establishes without qualification is authorship by semlar, a listing dating to 2014, and that the body is his rather than GSE’s. Their own sworn field text names the work “GSE - Gnome Sequencer Enhanced”, which is the same lineage read from the other end.

GSE publishes the expansion in implementable detail. The wiki page “GSE3: Advanced Loop Step Functions - Priority and Reverse Priority”, public, last edited 2026-02-06, states the weighting in words, “in the case of a 7 line macro the first line is attempted 7 times more than the last line with line 2 6 times more and line 6 2 times more”, then prints the literal expansion 1 / 12 / 123 / 1234 / 12345 / 123456 / 1234567, then the reverse variant, then a seven pass worked example with spell names.

That page was last edited forty one days before my first commit. The “no convergent evolution, no independent invention” argument at line 24 needs the behaviour to be obtainable only by reading GSE’s source. It is on GSE’s own wiki, in words and in a printed table, and it was not GSE’s to begin with.

5. Ten years, not 12 to 13

TimothyLuke/GSE-Advanced-Macro-Compiler was created 2016-06-24. Ten years to today.

gh repo view TimothyLuke/GSE-Advanced-Macro-Compiler --json createdAt

The figure appears five times in one repository and twice in the other. Six of the seven are prose. The seventh is CURSEFORGE-COMPLAINT-FIELDS.md:37, field text prepared for a CurseForge filing, and line 95 of the same file is the attestation reading “All information I have provided is true.” Their own sentence at line 37 dates the addon to 2015 or 2016 and calls it 12 to 13 years of prior art in the same breath, so the two halves disagree with each other before anyone checks the repository date.

GnomeSequencer is older than the GSE repository, but that cuts the other way. It makes GSE the downstream work.

6. Six counting errors

Taken one at a time none of these matters. Taken together they sit in the passages that establish the package’s measurement rigour, including the paragraph stating that its line numbers “were re-derived from those exact artifacts”.

Where Claim Measured
forensic :170 and :295 “272 entries” then “244 entries” 272 and 288 total entries; 44 directory entries each; 228 and 244 files
forensic :96 “GRIP’s entire ~9,000-file tree” 228 files in v2.3.5, 244 in v2.3.16
forensic :306 StepFunctions.lua “is 505 lines in both” 504 in both
forensic :309 LegacyImport.lua “grew from ~900 to 2,396 lines” 2052 in v2.3.5, 2395 in v2.3.16. No shipped release has it near 900
forensic :96 and :309 the blanking block “previously cited at 857-872” 870 to 876 in v2.3.5. Line 261 of the same file cites 870 to 874 for the same block 
for z in evidence/GRIP-EMS-v2.3.5.zip evidence/GRIP-EMS-v2.3.16.zip; do
  echo "$z"; unzip -Z1 "$z" | wc -l; unzip -Z1 "$z" | grep -c '/$'
done
find v235 -type f | wc -l; find v2316 -type f | wc -l
wc -l v235/GRIP-EMS/Engine/StepFunctions.lua v2316/GRIP-EMS/Engine/StepFunctions.lua
wc -l v235/GRIP-EMS/Import/LegacyImport.lua v2316/GRIP-EMS/Import/LegacyImport.lua

The entries row is the one worth a second look. 272 counts every entry including directories and 244 counts files only, so side by side under one word the package reads as though it shrank by 28 entries between the two releases. It grew by 16. Their hashes, byte sizes and Lua counts are all correct, so this is a counting method that changed between two rows rather than a problem with the artifacts.

7. Two of five rows in the scienter table are a third party’s words

This concerns grip-1202-cmi-analysis.md in the Conversion repository only. The other repository carries an earlier version of that memo with no such table.

The table at line 56 is headed “Evidence (developers’ own words)” and line 64 says the innocent-conversion reading “is contradicted by the developers’ own words”. Two of the five rows quote CzarTheMad: the row at line 59, “Perhaps it would be easier for me to strip the gse tools stuff from gse”, message 1499389371697336340; and the row at line 61, “I’m worried you’ll stop the conversation in addon. Or TL will do so much obfuscating that it makes it time prohibited”, message 1499389764921724940.

Their own capture attributes both correctly and repeatedly, with permalinks, at captures.md:147, :149, :165 and :297. So this is not a dispute about who said what. It is that the memo then presents them as the developer’s words. Their own record also says who CzarTheMad is: captures.md:140 gives “CzarTheMad (MOD badge) and Sataana (Developer role)”, and THE-STORY.md:156 lists the roles as Sataana tagged Developer, MFDOOM and itmeteemo Operations, and “bearded_dad_bod/CzarTheMad/Pershizzle” as BETA Testers.

captures.md is byte identical in both repositories, so both citations resolve wherever you check them. On the second row the memo also drops the leading word: their capture has “Yes I’m worried you’ll stop the conversation in addon”, which is a reply to a question.

I am addressing the attribution and nothing else in that table.

8. The predicate act, from their own three documents

The 1202 and inducement theories both need a user copy that infringes. Three passages in their own repositories bear on whether one exists, quoted rather than characterised.

SLG-Sequences-LICENSE.txt clause 1 grants “a personal, non-exclusive, non-transferable, revocable license to download and use this software and its sequence/macro content … for your own personal, non-commercial use in World of Warcraft, including modifying it for your own personal use.”

grip-cmi-evidence-exhibit.md:201, and curseforge-complaint-final.md:69 in the same words: “A user converting their own installed copy for personal use is not the target.”

README.md:71 in the Conversion repository: “An investigation into whether the community site lazygrip.net hosts / redistributes copies of these sequences returned no matches … That hosting angle is deliberately excluded; do not include it.”

So the act a user performs is inside their grant and their own documents say it is not the target, and the act that would carry the theory was searched for and not found.

Two related figures. curseforge-complaint-final.md:25 states “roughly 80% of my sequences are distributed privately to supporters/subscribers and are not published openly”, and grip-cmi-evidence-exhibit.md:274 maps licence clause 3 onto “the ~80% privately-distributed sequences”. THE-STORY.md:7 and captures.md:13 both state “ScaryLarryGames publishes all his current content free on CurseForge”, and README.md:63 states “The 14 works claimed are all of the rights holder’s CurseForge projects … each published All Rights Reserved.”

9. The memo drops context that their own capture supplies

Their capture file is careful. It is the memo built on top of it that is not, and the gap between the two is measurable without leaving their repository.

captures.md:150, message 1499389964071469108, CzarTheMad at 12:40:24, two messages after the “strip the gse tools stuff” line: “Want to still be able to use GSE w/o the companion app. And to be able to import gse strings into ems for the forseeable future.” Removing the GSE.Tools dependency from his own copy of the GSE addon so it runs without the companion is a different act from removing owner identifiers from sequences. The memo row at line 59 titles that same quote “Intended to remove the identifying data”.

captures.md:295, on a different “strip” reference: “This particular ‘strip’ reference is about UI simplification, not CMI, included for completeness/accuracy; do not conflate with the CMI-stripping conduct.” That is the right instinct, written by them, and it is not applied to the row above.

captures.md:190 prints the licence message in full. The two fragments quoted at line 204 as proof of “knowing engagement with, and conscious legal hedging around, GSE’s license terms” are “In theory (I cant advise it since the GitHub is under ARR Licence)” and “(Again, legally I cant advice it)”. Both sit inside a sentence declining to advise, and the subject of the sentence is reading GSE’s public GitHub commits, not sequence identifiers.

The source file flags its own reliability 46 times: 33 instances of “date TBD”, 9 of “truncated”, 2 of “need full message”, 1 “could not be loaded”, and 1 statement that the capture is “necessarily incomplete”.

for p in 'date TBD' 'need full message' 'truncated' 'could not be loaded' \
         'necessarily incomplete'; do
  printf '%-26s %s\n' "$p" "$(grep -oi "$p" evidence/discord/captures.md | wc -l)"
done

Their transcriptions are close to exact and their attributions in the capture are right. It is that the memo presents five fragments as a chain while the file they came from records the goal, the distinction and the caveats sitting around them.

10. The site owner line

README.md:47 in the Addon-vs repository lists me as “(site owner)” of LazyGrip.net. Line 37 calls the ownership “circumstantially the same operator” and line 50 makes that the “single most important nexus fact”.

I do not own the lazygrip GitHub organisation and I am not a member of it. I am a repository collaborator whose permission the GitHub API returns as write, given by the maintainer so I can push branches and open and triage pull requests on the addon-facing parts of the site. I hold neither admin nor maintain, and viewerCanAdminister is false for me there. I do not own or operate the domain, the hosting, the database or the moderation.

Pull requests 25 and 26 on lazygrip/lazygrip-gg were opened by me and merged by me, and I said so in the companion update before anyone asked.

Their own OPERATOR-IDENTITY-RESOLVED.md section 2 already reaches most of this. It credits the Workshop tools to Beard3d_Gamer and Slowdog, on the site’s own in-panel credit and on their own descriptions of their roles, and it records the domain registrant as redacted. README.md:47 predates that document and contradicts it.

What I am not claiming

This is a record of measurements, not a legal document and not advice. Where their citations are right I have said so, here and in the companion update: PlatformID, HelpURL and gse.tools are zero; the blockedPlayers gate is real and their description of it is accurate; their Lua counts, hashes and byte sizes are correct; StepFunctions.lua:248-262 is byte identical across the two releases and in fact the whole file is; their section 7a migrate field set holds across all four archives; and the engine claim was theirs and correct.

Two defects of mine are visible in their evidence and I am not disputing either: the converter refusal on one of their sequences, and a spell ID that changes across a web conversion round trip. Both are logged and being fixed.

Verification