Cryptograss:Connectome/Live Music Archive

From PickiPedia: A knowledge base of bluegrass, old time psychedelic jams, and other public domain music
Jump to navigationJump to search

What the Live Music Archive (archive.org's etree collection) says about who played with Sam Grisman, and how it says it. This is a survey for a parser; nothing has been built or written to the connectome yet (Justin, #connectome, 9 Nov, block 26,380,430). Checked by blu on 9 Nov against archive.org's search and metadata APIs. All 206 Sam Grisman Project items were read.[unverified]

Counts

  • 206 items with creator "Sam Grisman Project", covering 139 show dates. 40 of those dates have more than one item, because several tapers recorded the same night.[unverified]
  • 87 of the 139 dates have at least one source whose description names someone besides Sam: a lineup, a "with" line, or a marked guest.[unverified]
  • David Grier is the creator of no etree item, but 18 items name him in their descriptions. He appears as a guest with Yonder Mountain String Band, Donna the Buffalo, the Travelin' McCourys, Tony Trischka, Rapidgrass Quintet and the Jason Carter Band, and on Arabesque recordings.[unverified]

The license

The license is set on the band's collection, not on each show. SamGrismanProject's rights field holds the band's own letter to the LMA, dated 27 Jan 2023, giving permission for "public download and noncommercial, royalty-free circulation". That's the taper agreement in writing. Only 2 of the 206 items carry their own licenseurl (CC BY-NC 4.0 and CC BY-NC-ND 4.0). So a parser reads the license from the collection and the item's own field together, and cites both.[unverified]

Structured fields (read as-is)

creator, date, venue, coverage (city), taper, transferer, source (the gear), lineage. Each one is a separate field. The taper is a connectome person too: someone who was there.[unverified]

Who played: free text in description

The lineup is never a field. It sits in the description's HTML, in one of these shapes:[unverified]

  1. A header and a list. "Lineup:", "Band:" or "The Band -", followed by one name per line, sometimes with "Name-Instrument & Vocals". A guest list can follow it under "special guests :" or "The Guests -".[unverified]
  2. A "with" line under the band name: "with Max Flansberg and Richie Stearns", "with Dawg (David Grisman)", "with Guests incl Maria Muldauer".[unverified]
  3. Name - instrument lines: "Peter Rowan - guitar", "Jayna Manning - guitar/vocal".[unverified]
  4. {{{1}}}[unverified]
  5. Nothing: many sources give only band, date, venue and setlist. Some notes are about the microphones only.[unverified]

The .txt info files (173 of 206 items have one) mostly repeat the description. The one checked (2023-01-23) has only a setlist.[unverified]

What the parser has to survive

  • Name spellings vary from taper to taper: Max Flansberg, Flansburg and Flasburg; Tod and Todd Livingston; Dominic and Dominick Leslie; "Chris J. English" and "Chris English aka Hollywood". The parser copies names as written and merges nothing. A person maps names to pages.[unverified]
  • Names broken across lines by the description's HTML, for example "Bryan" / "McDowell".[unverified]
  • Several sources for one show: fold them by date and venue, and keep every source's lines, because one taper lists the band and another doesn't.[unverified]
  • Placeholders: "with Guests", "Cousins Vinnie and Joe".[unverified]

The parser this calls for

Dumb, like the rest. For each item, it reads the structured fields and the license, then finds lines matching the four shapes above. Each line is kept verbatim, with the item URL as its source. It splits only where the shape is mechanical: "Name - instrument", a header list, and "and" or "&" inside a "with" line. A guest mark is tied to the setlist songs that carry the same symbol. Everything else goes through verbatim for a person to read. The output is the same per-source reading Sideman --album gives, with the show as the record.[unverified]

Second path: Kyle Tuttle

Justin asked for another person we've worked with (#connectome, 10 Nov, block 26,380,889). I searched the LMA for everyone in 4masks's lineups first: Kyle Tuttle 147 items, John Mailander 39, Cory Walker 20, Maddie Denton 19, Harry Clark 16, Gaven Largent 4, and Skyler Golden, Kaitlyn Raitz and Justin Holmes 0. Most of Denton's, Clark's and Walker's hits are East Nash Grass. I followed Kyle, because his 147 items span 33 bands and many tapers. All 147 were read.[unverified]

New shapes, beyond Sam's

  • {{{1}}}[unverified]
  • Band names that carry the lineup in the file artist and creator tags: "The Bender ft. Mike Robinson, Andy Hall, Max Johnson, Kyle Tuttle", "Velvet Elmo ft. Kyle Tuttle, Mike Robinson and Max Johnson".[unverified]
  • The lineup in notes instead of description (13 items), sometimes under "The Band:".[unverified]
  • Song-inline credits in the setlist line itself: "Love Please Come Home (w/ Kyle Tuttle on Banjo)", "Wild Bill Jones - With Kyle Tuttle", "Get No Better (John Hartford) (with Kyle Tuttle & Dan Andree)".[unverified]
  • Numbered or bracketed footnotes instead of symbols: "(5) Played with Kyle Tuttle on banjo", "[1] with Kyle Tuttle", "#14 & 15 w/ Jeff Austin & Kyle Tuttle".[unverified]
  • A guest line that names a whole band: "w/ Leftover Salmon, Vince Herman on guitar, Drew Emmitt on mandolin, … and Kyle Tuttle on banjo" (Billy Strings, 28 Aug 2026).[unverified]
  • Prose: "With Kyle Tuttle on Foggy Mountain Breakdown." (Del McCoury Band, 6 Apr 2024); "Featured special guests Kyle Tuttle and Patti Steel."[unverified]
  • Banter tracks: "Kyle Tuttle Intro", "Anders Banter - Kyle Tuttle Intro". These are evidence he was on that stage, not of which songs he played.[unverified]
  • Cross-links between items: a Yonder Mountain show links the Kyle Tuttle Band's opening set on the same night.[unverified]

Traps (look like a connection, aren't one)

  • "Artists at Large :" followed by a list, copied onto every set at one festival on 20–21 Sep 2025 (Greensky, Travelin' McCourys, LaMP, Kyle Hollingsworth Band, The Last Revel, Pressing Strings, Sam Grisman Project). It's the festival's roster, not who played in that set. Only the set's own marks ("+ w Kyle Tuttle") count.[unverified]
  • "Kyle Tuttle Band opened": the same bill, not the same stage.[unverified]
  • Tapers' own doubts: "Alex? - Banjo"; "Kyle Tuttle on banjo (he wasn't in the multitrack so you can't hear him … Ya had to be there)"; "Kyle Tuttle - Mandolin" on one Larry Keel set. None of the other 146 items that name Kyle puts him on mandolin (many name no instrument at all). Copied as written, flagged, never corrected.[unverified]
  • Run-together text, where the HTML lost its line breaks: "Black Mountain BoysJerry Garcia - A Bluegrass JourneyWoodward Theatre…".[unverified]
  • Compilations: "Jam Cruise Jam Room", 2019 jam tunes from several bands and dates. Not one show.[unverified]

What the parser gains

Read notes as well as description, and read the file list: a file's comment is the best song-level source the archive has. Resolve footnote marks of every kind (* ^ # $ % + @ & ~ §, "(n)", "[n]") back to the setlist lines that carry them. Treat "Artists at Large", "opened" and banter tracks as their own scopes ("same festival", "same bill", "on stage"), never as a player on a song. Fold one festival's sets by date and venue, so one roster block doesn't become seven connections.[unverified]

Third path: Justin Holmes to Vickie Vaughn

Justin asked whether the Artificial Dumb alone could find a path from him to Vickie Vaughn (#connectome, 10 Nov, block 26,387,308). Homeplace (arthel and PickiPedia) gives Justin 60 people he has played with. Sideman's helpers find 20 record titles credited to a Vickie Vaughn on MusicBrainz and Discogs. Their credits name 205 people she shares a record with. The LMA was searched for every one of those names that has two words or more: 5,274 items.[unverified]

The best path found has a song at every hop:[unverified]

  • Justin Holmes and Kyle Tuttle: the same studio cut of "Silver 44" on 4masks (arthel songs_and_tunes/silver-44.yaml; also "Victory" on 4masks and "August" on Vowel Sounds). Corrected: this said Barlow's Jig, where the banjo is Cory Walker (Justin, talk page, 11 Nov).[unverified]
  • Kyle Tuttle and Molly Tuttle: "My Mountain so High $@" at Grassfire Festival, 5 Jul 2025, Peter Rowan with the Sam Grisman Project. $ is "with Kyle Tuttle" and @ is "With Molly Tuttle". SGP2025-07-05, taped by Jamie Burks, covered by Sam Grisman Project's LMA permission.[unverified]
  • Molly Tuttle and Vickie Vaughn: "Girl, Have Money When You're Old", track 9 of Safe, Sensible And Sane (Alison Brown and Steve Martin, 2025), on Discogs.[unverified]

The parser, measured

tools/lma_lineup.py on candidate branch pickipedia:blu/lma-lineup puts each line that names a person into one shape with a scope (song, show, stage, bill, festival, none, or "song?" for "Song (Name)", which can be a guest or the songwriter). Traps are shapes of their own. Lines naming one of 115 people from both circles:[unverified]

  • 98.1% of 1,630 lines read, on the 1,030 items the shapes were written against;[unverified]
  • 94.7% of 436 lines read on 400 items held out, which is the honest number.[unverified]

"Read" means a shape was assigned, not that it was the right one; that takes a person checking a sample. The 5% left over is long-tail prose: "Pre Show Music: Molly Tuttle on the Meadow Stage", "between The Sons of Otis Malone and Sam Bush", "foh - Adam McIntosh". Those lines are shown verbatim for a person to read.[unverified]

Retrieved from "https://pickipedia.xyz/index.php?title=Cryptograss:Connectome/Live_Music_Archive&oldid=16013"