← Building
Music Technology · Research System2026In Progress

DJ Playprint:Find the Record Inside the Set

Source FeasibilityIdentity ResolutionData ProvenanceRecord MatchingInterface PrototypingBrowser QA

The Recording-Finder Problem

A copied tracklist can tell me what somebody typed. It cannot tell me whether the record exists under that spelling, which version played, whether the link still works, or where the artist actually sells it.

DJ Playprint starts with a DJ name and follows the evidence until it reaches a recording. Every useful result needs a chain that can survive inspection.

  1. 01DJ identity
  2. 02Set evidence
  3. 03Timestamp
  4. 04Recording identity
  5. 05Version
  6. 06Listen or buy

Copying the string is the easy part. Proving what it points to is the product.

Current proof / four evidence cards

Four DJs, Four Source Conditions

Checked July 21, 2026

Batu and Four Tet are working profile artifacts. Seth Troxler and Carl Craig are corpus-level stress tests whose dedicated interfaces have not been run. Each card keeps that status visible, then opens into the source receipt and boundary behind it.

Profile / 01Working profileJuly 20, 2026

Batu

Identifier-rich baseline

Batu’s solo-primary NTS episodes provide unusually strong track-level identifiers. That made the first profile a clean test of exact recording matches without mixing in B2B or guest selections.

25solo-primary episodes
359track plays
128ISRC-backed plays
19MusicBrainz links
Open evidence receipt
  • 354 unique normalized artist-title pairs
  • 85 NTS-supplied Deezer links
  • 50-record destination pilot with exact-identifier gates
  • 13 artifact tests and 22 desktop/mobile browser checks passed
Source mix
25 solo-primary NTS episodes
Stable identity
128 ISRC · 19 MusicBrainz
Destination pilot
50 strict records
Artifact gate
13 data · 22 browser checks

Boundary: Batu’s identifier coverage is unusually favorable. It proves the strict path, not average archive quality.

Desktop DJ Playprint prototype showing the Batu discovery view and source-backed track lanes
Discovery begins with what the source can prove.
Mobile DJ Playprint prototype showing searchable Batu track evidence
Batu track evidence at a true 390 × 844 viewport.
Profile / 02Working profileJuly 21, 2026

Four Tet

Coverage-rich portability smoke

Four Tet produced more documented sets and far fewer stable recording identities. MixesDB supplied breadth; NTS supplied authority; the profile kept the gap between a typed track string and a verified recording visible.

25bounded complete sets
549track rows
18strict MusicBrainz matches
0NTS stable IDs
Open evidence receipt
  • 212 MixesDB archive pages discovered
  • 26 NTS episodes and 126 solo-primary NTS track rows
  • 50 strict MusicBrainz candidates: 18 exact, 32 unresolved, zero API errors
  • 29 unit/regression tests, 15 link checks, and 29 browser checks passed
Source mix
25 MixesDB build sets · 26 NTS episodes
NTS identity
0 stable recording IDs
Resolver pilot
18 exact · 32 unresolved
Artifact gate
29 tests · 29 browser checks

Boundary: Community tracklists provide coverage, not recording identity. The 25 sets are a bounded slice, not Four Tet’s complete history.

Desktop Four Tet DJ Playprint profile showing source coverage and strict identity evidence
The second profile makes source inequality visible.
Mobile Four Tet DJ Playprint profile showing source-backed track rows
Four Tet track evidence at a true 390 × 844 viewport.
Profile / 03Corpus receiptJuly 20, 2026

Seth Troxler

Volume-rich, attribution-heavy

Troxler has the largest set trail in the research corpus and the dirtiest text. Merged lines, inconsistent dashes, and a heavy B2B history make source attribution the hard problem before recording resolution even begins.

142strict identified sets
2,121identified track rows
1,886normalized tracks
191recurring tracks
Open evidence receipt

Corpus pattern

Repeat signals

Strict sets only

Exact artist-title pairs repeated across strict solo-live and radio sets. They are listening leads, not taste rankings.

Strict identified sets by three-year band

All 142 strict sets have a usable year.

Start with the repeats

  1. Good For MeDJ Dan & Jerome Robins · 2013–20185 sets
  2. SiliconDANNY DAZE · 20144 sets
  3. GrowlerPig&Dan · 20164 sets

Resolved trailhead

Floorplan / Never Grow OldOriginal album version · M-Plant MPM16The set string resolves to the original Paradise album version. The later Re-Plant remains a separate recording.
  • 205 MixesDB pages contain parsed track lines; 139 pass the strict identified-set analysis
  • Three primary NTS episodes contain 50 track rows and no stable recording identifiers
  • 74 of 248 title-matched MixesDB pages are collaborative or B2B
  • Floorplan’s “Never Grow Old” resolves to the original M-Plant MPM16 album version, not the later Re-Plant
Strict source mix
139 MixesDB · 3 NTS
Title-matched span
2006–2026
Shared-set risk
74 collaborative or B2B pages
NTS identity
50 rows · no stable recording IDs

Boundary: The 2,121 rows are normalized track strings from a research corpus, not 2,121 proven solo Troxler selections. Shared sets and merged lines still need set-level attribution before this becomes a working profile.

Profile / 04Corpus receiptJuly 20, 2026

Carl Craig

Deep-history alias test

Craig supplies the deepest archive and the cleanest released-record coverage in the original stress test. The problem moves from finding track strings to preserving versions and linking aliases without erasing the credited artist.

100strict identified sets
1,462identified track rows
1,113normalized tracks
176recurring tracks
Open evidence receipt

Corpus pattern

Repeat signals

Strict sets only

Exact artist-title pairs repeated across strict solo-live and radio sets. Alias links stay separate from the credited artist.

Strict identified sets by four-year band

Ninety-nine strict sets are dated from 1994–2025; one is undated.

Start with the repeats

  1. ThrowPaperclip People · 1994–20179 sets
  2. Rushed69 · 1995–20248 sets
  3. Strings Of LifeRhythim Is Rhythim · 2005–20188 sets

Resolved trailhead

Paperclip People / ThrowOriginal · Planet E PLE653473The set string reaches Planet E’s official track page. Remake and Basic Reshape remain separate versions.
  • 134 MixesDB pages contain parsed track lines; 99 pass the strict identified-set analysis
  • One primary NTS episode contains 24 track rows and no stable recording identifiers
  • Paperclip People’s “Throw” appears in nine sets; 69’s “Rushed” and Rhythim Is Rhythim’s “Strings of Life” appear in eight
  • The “Throw” resolver receipt reaches the original Planet E version, direct Bandcamp purchase, and release evidence
Strict source mix
99 MixesDB · 1 NTS
Title-matched span
1991–2026
Shared-set risk
43 collaborative or B2B pages
NTS identity
24 rows · no stable recording IDs

Boundary: Paperclip People, 69, Designer Music, and Innerzone Orchestra are connected to Craig, but they remain distinct credits. Forty-three collaborative or B2B pages are shared evidence, not solo selection claims.

The Sources Are Unequal

The research spikes tested Batu, Four Tet, Seth Troxler, and Carl Craig across radio archives, community databases, and music catalogs. The result was not one universal scraper. It was a source hierarchy.

Authoritative structured lane

NTS Radio

Public endpoints preserve source-owned episode and track data. Batu is identifier-rich; Four Tet shows how sparse the same lane can be.

Supported discovery lanes

Mixcloud, BBC, independent radio

Official APIs or parseable public pages can extend coverage, but every adapter needs its own terms and attribution review.

Community leads

MixesDB and 1001Tracklists

Useful for discovery and corroboration. They do not become authoritative simply because the tracklist looks complete.

Excluded from automation

Resident Advisor

Technically extractable, but outside the production ingestion boundary under the current terms and robots policy.

Rinse remains useful as a recording node, but the sampled episode pages did not carry populated tracklists. Audio recognition may eventually fill some gaps. It stays outside the first product because it changes the cost, rights, and confidence model all at once.

A Link Is Not a Match

NTS supplied 84 unique Deezer catalog IDs in the Batu corpus. A current API audit found 59 readable records and 25 catalog records that were not readable from this region. Eighty-two API ISRCs matched the NTS ISRC. The 50-record pilot admits a track only when the destination is readable now and the ISRC matches exactly.

Set evidenceExact recordingRelease evidenceChecked destination

Ploy’s “Ramos” is the first complete cross-archive receipt: set evidence, exact recording, official listening page, Discogs release evidence, and direct Bandcamp purchase. Other retailer buttons remain labeled searches until the release and version are verified. Three literal unknown-ID rows remain open. I did not guess them away.

What Broke During the Proof

Failure / 01

The research table widened the phone page

The table initially forced horizontal overflow across the whole document. It now scrolls inside its own container while the page remains at viewport width.

Failure / 02

Navigation landed below the wrong thing

Selecting a tool on mobile opened its content beneath the discovery hero. The interface now moves the chosen workspace into view before the person has to hunt for it.

The Batu and Four Tet profiles passed dedicated integrity, link, desktop, and mobile checks after their repairs. Seth and Carl remain corpus receipts until the same artifact gate runs for them. The distinction stays visible.

From Batu to Any DJ

Two profile runs prove that the pipeline can operate under unlike source conditions. Two additional corpus receipts expose the next hard cases before the interface claims to solve them. Seth forces collaborative attribution and dirty-line handling. Carl forces alias and version resolution across three decades. The general product should search on demand rather than copy the entire DJ internet into one brittle database. A cache hit can return existing evidence immediately. A cache miss can resolve identity, discover supported set sources, run the appropriate adapters, normalize tracks, and preserve every source URL and checked date.

01

Matched

One identity, source-backed sets, and a music list with visible evidence and checked destinations.

02

Ambiguous

Several artists share the name or alias. The person searching chooses the identity before extraction continues.

03

Sparse

Recordings or leads exist, but no reliable tracklist has been found yet. The gap stays visible.

IdentitySourcesTrack evidenceNormalizationRecording and releaseDestinationsCache

Build boundary / July 21, 2026

What Is Proven and What Is Not

Proven now

  • Evidence and provenance model
  • Batu and Four Tet profile-scoped extraction
  • Seth and Carl corpus-level stress tests
  • Exact-identifier and strict text-match gates
  • Searchable desktop and mobile prototype
  • Automated integrity and browser checks

Still open

  • Arbitrary DJ identity resolution
  • Production source adapters
  • Commercial ingestion permissions
  • Release-level purchase resolution at scale
  • Dedicated Seth and Carl artifact runs
  • Coverage testing beyond four DJs

The repository stays private while the source and data-rights boundary is reviewed. The next build turns the profile runner into a search API and tests more DJs who do not have Batu’s unusually strong NTS record.