Since the jam chart team has just published a revised chart for Ghost, we thought it would be informative, and perhaps even interesting (for some) to learn more about the process of updating a major jam chart. Ghost is a cherished, fan-favorite jamming song. Going in to this process, we knew that the results would be closely scrutinized, debated, and that we would draw the ire of those who disagree with some of our decisions, like which versions to highlight.

Why did we even feel the need to tinker with the Ghost jam chart? First, the former Ghost chart was assembled in a hurried and somewhat haphazard manner, part of a much larger effort to introduce a new and improved jam chart format that occurred in December 2013. Second, a quick glance at the (now) former chart gave us reason to believe that the chart was overlooking important versions, and underrepresenting particular years. Consider these statistics: the former chart had thirteen versions from 1997 and two from 1999. Likewise, 2003 and 2004 were represented by five total versions, and we inherently knew that the 2.0 era is particularly strong from a jamming perspective. So we set out to do a comprehensive review of all 133 live performances of Ghost, seeking to ensure that a revised chart did not overlook any strong improvisational versions, and that the final chart would reflect the entire performance history of Ghost, covering the high water marks across all years and eras. The team consisted of Marty Acaster (@Doctor_Smarty), Pete Skewes-Coxe (@ucpete), Andrew Stavely (@Westbrook) and me. Below is a description of the processes we employed:

Preparation - Before starting, we wanted to establish whether there is a typical and formulaic pattern to the jam found in most versions of Ghost. After some healthy and heated debate, we concluded that yes, many versions of Ghost share a similar structure. In the typical version, the jam begins with some funky, rocking or exploratory grooving. Then there is a shift, a "release" or "opening" of the jam to a more upbeat, blissful sentiment. I use these self-coined descriptive terms with caution; I am not a musician, and I am certain that musicians, students of music theory and others could provide a more appropriate and accurate term for the shift in mood that seems to happen in most Ghost jams. In addition, while the typical Ghost does follow this basic pattern, the band uses a variety of tecniques to achieve it. Sometimes, the play remains in minor mode throughout, despite this shift to a more upbeat feel. Other times, for example in the version from 12/31/10, there is a clear shift to major mode, which accentuates the blissful nature of the latter part of the jam.

Once we established (at least amongst ourselves) that there is a fundamental pattern to most Ghost jams, we set out to review the 133 versions of Ghost with two goals in mind: 1) to identify the strong, objectively different versions of Ghost which deviate from this normal structure; and 2) to identify the subjectively better versions of Ghost which conform to the typical jamming pattern. The review process was divided into four sequences of review and decision making. These included:

Round 1 - we divided the 133 versions of Ghost evenly among the four of us, and assigned each person a mix of versions from different periods and years. The logic behind assigning each person a mix of versions was two-fold: 1) it would be unfair to everyone else if one person was assigned all of 1997; and 2) when considering a chart encompassing the performance history of a song, it's helpful to listen to versions from different periods, taking note of changes in the style and focus of the jamming. The principal instruction for Round 1 was simply to identify the absolute, must-be-on-chart, no-brainer type versions, like 11/17/97, 5/22/00, and 11/28/09. In addition, each member was also asked to rate his other versions a "no" - not for chart, or a "maybe." For maybe versions, people were asked to assign a weighting of high, medium or low.

Round 2 - of the versions that were voted "yes" or must-be-on-chart in Round 1 (33), we reviewed and cut this list to the 18 versions we felt were indisputable chart material - those mentioned above and others like 7/2/98 and 9/12/99. The remaining 15 yes versions from Round 1 were combined with the 14 highly rated maybe versions to make up the pool for Round 2. To further ensure that we did not miss any quality versions with chart potential, we elevated four versions to Round 2 that we felt had possibly been overlooked or underrated in Round 1. Finally, each participant got to pick one version, a personal favorite or one he thought should get another listen, and add it to the Round 2 pool. It turns out that this last measure was a good one. Several versions that made the final chart arrived there after receiving a rating of less than a high maybe in Round 1, including 12/11/99 and 12/31/09. In all, the pool of versions to review in Round 2 totaled 37.

In Round 2, every version was assigned to two new listeners who had not heard this version in the first round. Versions which received two affirmative yes votes in Round 2 were elevated to the final jam chart. Versions which received one or more maybe votes in Round 2 were pushed to Round 3. In other words, in order to make the jam chart in Round 2, a version needed to be rated a yes or high maybe in Round 1, and then receive unanimous yes votes from the new listeners in Round 2. At the end of Round 2, the jam chart had increased from 18 after Round 1 to 40 total versions. 22 of the 37 Round 2 versions were approved for the jam chart, while the remaining 15 were pushed out to the next cycle of review.

Round 3 - every one of the 15 versions that made up the Round 3 pool had received at least one maybe vote in a previous round. In many cases, these Round 3 versions had received two or even three maybe votes. Looking through this list, as a group, we decided that none of these versions was as strong or as clear-cut in definitively belonging on the chart as the ones that had moved up in Round 2. So we came up with a process for picking the final jam chart versions from this group of 15 “fence straddlers.” Of the 15, we picked 5 versions, bringing the total revised chart to 46 versions, or 45 jamming versions plus the debut (we always include the debut version in major charts). By coincidence and not by intention, 46 versions is precisely the same number of versions as appeared on the former chart.

Highlighted Versions - finally, we went through several mechanisms to determine which versions should be highlighted. Initially, everyone was instructed to vote for 12 versions. After compiling the votes, there were nine versions which had received unanimous support (4 votes). After those nine however, the results became more unclear. There were a couple of versions where the voting split down the middle - two people voted for version X and did not vote for version Y. The other two voted for version Y but did not vote for version X. One person suggested we do a vote in which each rank our choices for highlighting, from first to last. By assigning a successively lower numerical value to every placement vote below first place, we hoped to determine if this method would better indicate a consensus on which versions to highlight. In truth, this second process to pick highlighted versions helped to clarify some things, only to confuse others. We went through several additional iterations of voting for highlighted versions. When we finally concluded the highlighting process, everyone was very comfortable not only with the versions we selected, but also felt we did not leave any deserving versions unhighlighted.

The Revised Jam Chart for Ghost - as a group, we feel strongly that this revised chart is better and more fully representative of Ghost's prominent improvisational history than its predecessor was. Nearly 40% of the versions on this chart are new. In general, there are more versions from 1999 and the 2.0 era than before, and slightly fewer versions from 1997 and the 3.0 era. In addition, there are 11 completely new versions, ones which have never appeared on any previous version of a Ghost chart. A great example of these completely new versions is the Ghost from 7/30/99. A 22 minute improvisational behemoth, this particular version may have escaped general attention because back in the days of "tape trading," few copies of the recording from this Niigata, Japan show circulated. Also making a showing for the first time are two excellent versions from July, 2003. And there are several strong versions from the 3.0 era that have not previously appeared on any Ghost jam chart. Everyone is entitled to his or her own opinion about the content of the new chart. But for the four most directly involved in assembling this chart, we feel satisfied that the final product is more comprehensive than its predecessor was.

A final note: We fully expect to draw criticism from some circles about versions that were included or not included on the chart, and chart versions that were highlighted or not highlighted. Criticism is welcome and an expected result of an effort like this one, especially for a song as near and dear to many as Ghost. Having said that, we strongly encourage you to take some time, and listen to every version of Ghost on the revised chart, as we have. If you do, you may discover some fantastic versions you haven't heard before. And you may come to better understand the rationale behind some of our decisions, even if you disagree with these selections. Regardless, Ghost is rightfully celebrated as one of the premier Phish jamming songs, and we sincerly hope the new chart helps you to more easily navigate and enjoy the many great versions that exist.

If you liked this blog post, one way you could "like" it is to make a donation to The Mockingbird Foundation, the sponsor of Phish.net. Support music education for children, and you just might change the world.