1988 · Elite Systems
Every byte of the C64's 64 KB, one pixel each, 512 to a row. Lit pixels are what this game is: its code, graphics, level data, sound, text, tables and variables, exactly where they sit in memory. Dim areas are screen and working memory; dark is unused.
This game is in 3 parts. The map is of The program that stays, which stays in memory; the band marked as varying holds whichever of the 2 parts loaded over it is there. The Source tab has the listing of each part that has one.
| Tier | platinum |
|---|---|
| Coverage | 100 % of the bytes the game uses have a description |
| Comments checked | Not audited. No audit has checked every comment, and the last sample of 60 comments had 8 wrong, so between 7 and 24 % of the listing’s comments are likely wrong. How listings are checked |
| Media | Counting each copy once, the tape holds 23 files. Of those, 9 are in the listings, 13 are data the game loads and 1 loads the game. |
| Build analysed | TAP of the original Elite cassette (1,530,784 bytes): a 512-byte Novaload boot program in the KERNAL's format, then 26 Novaload files: the loader's second stage, the loading picture, the program in four files, and five question blocks of four files each, which the game loads during play |
| Copy | agent-draft |
| Tools | host: Linux x86_64 (WSL2) on Windows 11 Pro 10.0.26200, Ubuntu, Intel x86_64, emulator: own build of github.com/barryw/vice-mcp, release v3.13.2, commit 1e83aab134; check-emulator 57 of 57, disassembler: regenerator2000 0.9.20, publication: Python 3 and node 24 |
| Model | claude-opus-5-5 |
| Kit version | 0.0.339 |
Every step that measured the Source tab’s comments or made them more accurate, in the order it happened, with the models that did it. Accuracy of the Source explains each step and what the row above makes of them.
| Step | By | What it found | Of the listing |
|---|---|---|---|
| Research | claude-opus-5-5 | Looked the game up online (its manual, reviews or wiki pages) before reading its code. | |
| Annotation | claude-opus-5-5 | Wrote the listing’s comments. 100 % of the program is described, in 2,030 comments. the lead and five helpers, one at a time, each on its own address range or part | |
| Sample 1 | claude-opus-5-5 | 8 of 60 comments wrong, so between 7 and 24 % of the listing is likely wrong. All 8 were then corrected. checked by a fresh agent that wrote none of the comments: 3 variable comments missed a reader or writer in another part or loop, 2 named the wrong colour target (screen colours for colour RAM and back), 1 widened a range of zero-page bytes, 1 said 255 records for 256, 1 misread which byte a store writes; all 8 corrected, with 10 more comments that said the same (the other three operands of the colour-clear loop, the other seven sprite pointers) | |
| Play tab | claude-opus-5-5 | Ported the whole game to JavaScript for the Play tab and corrected 5 comments it showed wrong. the Play tab's port, every routine tested against the game's code and the whole game run in lockstep: 5 comments were wrong (where two steps go after picking their item, which member ends a team pick, how often the subject choice runs, how the question builder spaces a bracket) and 4 more were completed; all 9 corrected |
likely wrong the sample’s range corrected right, or likely right not yet measured
None listed yet. Know a write-up, port or forum thread about this game? Add it to game.json.
What the game is documented to do, with verification status against the binary. Statuses: open (documented, not found yet), traced (in the code, could not be exercised; say what was tried), confirmed (in the code, consistent with the emulator), live (observed directly), differs (the code does something else).
Sources:
$3E04, $3E83, $AEB7-$B456, the string sweep in work/strings-title.txt).| Feature | Status | Where |
|---|---|---|
| A one-player game against the computer, or two players | live | menu "Select game options"; player one on joystick port 1 or the keyboard, player two on port 2 ($1C9C) |
| Each team chooses a specialist subject: Pop, Rock, Rock 'n' Roll, Country & Folk, Classical, Mystery | live | the six icons; subject names at $B32F |
| Each team picks a captain and two members from twelve stars | live | the face grid; team introduction "the members of ...'s team are ... and ..." ($FEC0) |
| Five question blocks on the tape, one loaded for a game | live | "Question block 1" to "5" ($1C3B); four Novaload files a block (orientation.md); a block lasts four games (games_left $2B90) |
| The Jukebox round: choose a number, answer a question on a subject, four answers against the clock | live | "The first round of the game is the Jukebox round" ($AEF6); the number grid 1-12 shuffled over the subjects at $287F; answers in four slots ($3C3A) |
| A wrong answer is offered to the other team for a bonus point | traced | "Incorrect. A bonus point to ..." and "for a bonus point" ($1E2B, $1E6A); the pass to the other team at $3029-$3099 |
| The Your Scene round: an easy question for one point or a hard one for two | traced | announcement $3E04; "Your Scene round - ... will you please choose Easy or Hard" ($B440); Easy asks the team's own subject, Hard another, $3A84-$3AAB |
| Spot the Star: up to three clues, three points on the first, two on the second, one on the last | traced | $B204-$B277; "Correct 1/2/3 points" ($1DF2-$1E18); the next clue at each time-out, $2FC6-$2FDD; every item and clue of the five blocks built by $262E on the kit's 6502. The screen shows the team's captain, not the star ($2DA4-$2DC4) |
| Name the Year: events of one year, the team names it | traced | $B287, $B2AD; built the same way as Spot the Star |
| What happened next: a question with answers A to D | differs | the text ($25F8) and a routine to fill its answer buffers ($2640) are in the game, but no step of the show ($338E) and no call or pointer reaches them |
| The Quickfire Round: as many questions as a team can answer | traced | $B2CF, "End of Quickfire round" ($1EA1); 9 questions on one clock at half speed, a wrong answer a point to the other team ($0D88, $0DDD, $3125-$3162) |
| The final round is the Jukebox again | traced | $AEB7; show steps 25-30 run the Jukebox on the six numbers left ($3362, $338E) |
| 1,275 subject questions, 255 a block | confirmed | built by the game's own $3416 and $3C3A on the kit's 6502 from each block's snapshot (work/allq.js), block 1 again from the listing (test_questions.js) |
$3E83): designed by Ian Upton and Nick Acton, graphics by Dave Percival, music and sound effects by Mark Cooksey, programming by Richard Underhill. The loading picture adds "in conjunction with MPC Games Ltd.". Unused panel credits in the menu ($CC8E-$CD0A) name Steve Beverley for the visuals.$80 up in a question or an answer stands for one of them ($259C).$0936).$4FCB's nine callers).Each part of the game has its own symbol map and listing: The program that stays, symbols.json and listing.json; The menu, the subjects and the teams, symbols.json and listing.json; The quiz, with a question block loaded, symbols.json and listing.json. Write-ups, facts and symbol maps are CC BY-SA 4.0. The game itself is not hosted here.