What is a gap?
The number beside a song on a show page is how many shows the band played between that performance and the one before it. A gap of 0 means they played it again the very next night; 485 means four hundred and eighty-five shows went by. It is phish.net’s own figure and nothing here recomputes it.
A gap is not a length of time. A song with a gap of 30 in 1995 had been gone about five weeks; the same gap today is closer to a year.
Why does a song page say “shows since” rather than giving a current gap?
Because they are not the same number, and only one of them can be checked here. phish.net’s gap is not reproducible from a show calendar — two songs spanning the same pair of shows can carry different gaps, so there is a per-song term in it that is not published. Printing a number that disagrees with theirs under their name would be worse than printing our own under ours.
So the live figure counts shows the band has played since this song was last played, and calls it that. It is exact, because this site defines it.
What do > and –> mean, and how do they differ?
Both are phish.net’s marks, both say the band did not stop between two songs, and they are not interchangeable. Their definitions:
- –>
- An actual segue: one song jams fluidly and without interruption into another.
- >
- Everything else that runs together — one song stops and the next starts immediately with no jamming between them, or a member begins the next while the last is still ending.
The second one also carries a convention, which is the part worth knowing. phish.net uses > between songs that are simply always played together — Mike’s Song, I Am Hydrogen and Weekapaug Groove; The Horse and Silent in the Morning — and around lead-in and exit songs such as Hold Your Head Up, even where there was an audible gap in the music. A > is not always a claim about sound.
On this site a plain ← or → means neither of those: the song before or after, played as its own song, with a stop between them. The arrow is the absence of a segue rather than the presence of anything.
Definitions are phish.net’s, from their FAQ on segues.
Why does this row have no range bar?
A song needs eight performances inside the ten-year window before there is a usual range to draw. Below that there is no norm for a performance to be early or late against, so the cell carries a dim — rather than an empty track implying a comparison that was never made.
Two quite different songs land there: one played six times in ten years, and one not played in ten years at all. The statistics say which.
What does “due” mean, and why is a song that has been gone for years not on the list?
Due means past its own recent usual gap — measured against that song’s habit, not against a single number for the whole catalog. A staple is late at eight shows and a rarity is not late at eighty.
Recent means the last ten years of shows, counted back from the newest show in the archive rather than from the song’s own last night on stage. That distinction is the whole of the previous paragraph: measured from its own last performance, a song that stopped being played in 2011 still has a tidy ten-year habit ending in 2011, and would be ranked as running late against a band that has since played a thousand shows without it.
Being late is not the same as being expected, though, and more late is not more expected. A song at six times its usual gap is not one anybody is waiting on; it is drifting out of rotation. So the due page sorts songs into four:
- Due
- The band plays it at least every twenty shows or so, and it is now past its usual gap but less than three and a half times past. A song you would not be surprised to hear tonight.
- Slipping
- Well past its usual gap, or too rare to expect on any one night. Could come back, could be on the way out of rotation.
- On the shelf
- Gone more than a hundred shows — long enough that the habit it is being measured against has probably stopped being true.
- Out of rotation
- No recent record at all, and gone a hundred shows or more. Nobody is expecting it, and ranking these would bury the songs somebody might actually shout for tonight — so they have a page of their own, grouped by the year they were last heard rather than by a lateness they cannot have.
That fourth group is three groups, and the difference matters more than the other three lists put together. Dormant means a song used to be otherwise, and that is only true of the ones the band actually played: a song performed once at a Halloween show never had a rotation to fall out of. So the page splits on how many times the song was ever played — 8 or more is dormant, 3 to 7 is a rarity, and a song played once or twice is filed under exactly that — and the archive says the split is worth making. Of every silence of a hundred shows or more it holds, the share ever ended by another performance runs 84%, 65% and 30% down those three. A song played once or twice is the only kind here that is likelier to stay gone than to come back, which is the one thing the old single list could not say.
Slipping, not overdue, because a show page already uses overdue for something narrower: a single performance that came back later than that song usually does. Every song on the due page would be stamped overdue if it turned up tonight, so the word cannot also name one of the lists.
None of it knows what the band has planned. A themed night overrides every figure — the 2021 Halloween runs built around numbers and animals, the elements nights of the first Sphere run, a run played out of a single decade — and the theme is usually not public beforehand.
The hundred-show line is where this site already draws a bustout. It is counted in shows rather than in months on purpose: a gap of thirty-six shows is thirty-six chances to hear it, whether the band took eight months over them or two years. The band averaged about ninety-five shows a year through the 1990s and about forty-four across 2021–2025, so counting in time would quietly mean something different in each era.
Within each list the order is how far past each song is as a multiple of its own usual gap — the figure on the right of every row — not how many shows it has been gone, since a hundred shows is nothing for one song and a decade for another.
What are the eras — 1.0, 2.0, 3.0 and 4.0?
Era is the word this site uses for them, and the one on the chips that group a song’s performances and filter the show list. There are four of them, and they are separated by the three long breaks between them: the hiatus that began in 2000, the split that followed Coventry, and the shutdown of 2020.
Bounded by date rather than by year, because every one of those breaks falls mid-year.
- 1.0
- The beginning through 2000-10-07, the last show before the hiatus.
- 2.0
- 2002-12-31 to 2004-08-15, ending at Coventry.
- 3.0
- 2009-03-06, the Hampton reunion, to 2020-02-23.
- 4.0
- 2021-07-28 onwards.
They are used here for grouping and for counting a show’s place inside its own era — the 312th show of 3.0 — because there is no honest count of where a show sits overall. The method page says why.
Why do some shows say “Not Part of a Tour”?
Because phish.net does. It is where they file everything that was not a leg of a tour: the festivals, the television and radio sessions, the New Year’s runs held somewhere unusual, and the Mexico runs.
Several of those have famous names, and this site does not print them, which is deliberate. The names exist only inside freeform prose notes, spelled inconsistently and often not at all — pulling them out gives a handful of shows a label, some of them wrong, and leaves the rest blank. A blank is honest. A wrong festival name is not.
Where does all of this come from?
Three projects. Phish.net — a project of the volunteer-run Mockingbird Foundation — is the setlists, the gaps, the venues and tours, and the show ratings. phish.in is every Listen link, and holds the audio. fouldomain is the version scores.
Each of them is worth more than the credit line a footer can hold, so they have a page: who they are, what they give away, and where to send money if any of this has been useful.
Why does a show page say “setlist still coming in”?
Because the show is still being played. A report is published mid-show and fills in as phish.net records each song, so the figures on it are real but partial — a median gap over nine songs is the median of those nine, not of the night.
Nothing in the data says when a show has ended, so stability stands in for it: once the song count has stopped moving, the report stops calling itself provisional. The method page says how long.