Thursday, September 26, 2013

Interactive song-suite: debut postmortem.

First off: plenty of exciting news all around. Chris Klimas has come out of hiding and delivered a talk on his neglected creation, Twine. Choice of Games is on a roll, delivering the conclusion of their Choice of Romance trilogy, part 2 of Choice of the Vampire, and a follow-up to Heroes Rise hot off the presses last Friday. Just when I'd finally completed my exhaustive documentation over at Mobygames of all known works released written in their ChoiceScript language, GameFly majorly borked an unwanted, unnecessary site redesign and all my studious scholarship there, including much on hyperfiction and storygames, has been thrown into jeopardy. Fortunately, this obscure site is safe: its only risk factor is the bottleneck of my free time.

Now, as promised, pledged and threatened, I did debut a performance of an allegedly interactive song-suite with my jug band of the damned, the Creaking Planks, at the opening night of the 6th annual Accordion Noir Festival, joining two of my grand passions in a bizarre arranged marriage. We pulled it off, in a sense, though things were hairy.

Here's a link to the performance.

If you think back to the node graph I shared in an earlier post, it may not come as much of a surprise to you to learn that I planned out too many songs, which caused a few problems: primarily, it amounted to just too much for me to write in the allotted time, despite working overtime spitting out red-hot rhyming couplets in enormous quantities like it was 1995 and I was producing lit for elite BBSes in exhange for hot fresh warez. Additionally, I was penny wise and pound foolish in initially devising a plot reason for music re-use -- and then having to adhere to the plot contortions needed to bring the story to that point rather than allowing the threads to evolve organically.

The overall performance was too long -- about 40 minutes, vs. the 25 I was shooting for. (3 minutes per song, I thought, plus one minute per for audience-choosing haranguing, applause, etc.) I felt that six-part story-chains with five choices made by the crowd would be a good, satisfying length, but due to my choice of audience selecting mechanism (in homage to its since-expired suggestor, though as I said, I did admire the way it made the audience not only participants but outright performers) every individual song became too long, because of a need to telegraph to the audience what the options were going to look like when the moment of choosing came upon them. That moment was indicated with the use of a purloined front desk handbell ("please ring for service") and when mic'd (could I have less "ding!" in the monitor, please?) it made for an excellent signal: we were worried about getting input from a confused audience, but the problem was not that they did not know when to give input or how to offer it, but that they stubbornly kept choosing dead-end choices that led to parts of the work I hadn't managed to finish writing by showtime. The Choice of Games folks talk a lot about how in order to be satisfying, choices extended in gamebooks need to be weighty, well-distinguished from each other, and generally equally interesting: a reader's curiosity could lead either way but consistent choices are made for role-playing purposes. Apparently the most interesting choices I had offered up were to songs I hadn't managed to complete the writing of. Going into the debut performance with an incomplete work was a crapshoot, but I thought it was like the practice of at least going through the motions of offering a choice, when there's a 50% chance they'll go along with the choice you hope for them to choose anyhow.

The telegraphing was needed because much of the lyrical content was abstract and bizarre (because, two days before the show, I had to take whatever I could come up with), and I therefore couldn't assume that anyone was successfully following along until they heard the ding! Even without setting up the choices, these segments were making for long songs: get enough instrumental introductions, verses, choruses, and bridges, and you have a complete song -- but a sprawling one. A Choose-Your-Own-Adventure children's book can offer you a choice half a page after it starts, after only 30 seconds of reading, but a complete song takes 4-5 minutes to present, perhaps analogous to those only-nominally CYOA works which only offer choices after pages and pages of text.

The opening choice, "would you like a happy song, or a sad one?", was a weak initial choice, but I wanted to warm up the crowd with a low-stakes easy-pitch choice, like in the opening of the outstanding interactive comic Meanwhile (which flavour of ice cream? Chocolate or Vanilla?) My impish and perverse take on things was probably a complication (entailing appearing to deliver the crowd the reverse of what they asked for -- though having a happy/sad song and a sad/happy song was handy when I had to take half of the choices off the menu at the last minute, with the remaining song being arguably happy and/or sad and suiting a choice of either option.) Also there were some admittedly stupid and silly choices (will your secret dietary supplement be eating rainbows or eating cuddles?), in the psychedelic children's literature spirit of CYOAs (You have the day off. Will you: build a robot and campaign for it to become President or the USA or teach your dog how to fly and buzz the crater of the seemingly-extinct volcano?)

I wasn't thrilled by my re-use of backing music -- eg. currently, all possibilities for "Song #3" are musically identical to each other, simply re-skinned with different lyrics, and similarly for the "Song #4"s and "Song #5"s -- but given the scope of the project, it was a necessary evil: even supposing one is able to come up with 30 distinct songs it's not fair to expect a band to pick up 30 songs in two cramming rehearsals with the knowledge that 24 of them will not be used. (Also logistically it's handy, as this way the band only needs to remember one sequence of six songs whose rhythms and chord patterns always run the same way rather than needing to thumb through a heavily tabbed manuscript to get to the next one song out of the 30 possible -- after quickly arriving at some on-stage consensus regarding just which audience selection was chosen, in this model a process which is always subject to a certain degree of tie-breaking conductorial interpretation.) My music situation was admittedly better than in my YouTube predecessor, the Haircut, which is all just one song, admittedly penned and rendered by a sassy and cool individual with neat visuals. Additionally, adhering to standardized song templates for every level in the chart ensured that songs that didn't actually have that much to express needed a great deal of time-dragging lyrical filler in order to meet their cohort's quota of verses and choruses instead of just getting to the point and moving on. In the fullness time, after words are written for all sections, I can see the musical portion of individual song nodes being customized further until they are all more or less unique. In the meantime, thanks for doubtlessly inspiring my instrumental groves go out to Portishead, Yann Tiersen's Amelie OST, Howard Shore's "Gollum's Song" from the Two Towers final credits roll, and Raghu Lokanathan's "The Steam That Turns The Wheel".

An opening night surprise! I had been writing the work in plaintext and rehearsing it in the Twine IDE, not needing to render it down to a final HTML file yet since of course it hadn't been completed yet, crawling toward half-baked-ness one coffee break at a time. But I was sure that when showtime struck, I'd be reading it off the laptop in the least labour-intensive way possible: just scrolling down a single browser window with no-precision-needed strikes of the space bar and following choices with a single easy-to-quickly-execute hypertext click. However, after sound check, I was reminded that compiled files in Twine are of course HTML, and consequently manual indents are removed, instantly obscuring all of my cues distinguishing between which groupings of words are verses and which are the chorus! I went to the onboard help seeking a reminder of how to paste the relevant indenting HTML commands into the passages and found that the required information was unobtainable without being online, a capability that the concert hall did not offer. So the performance ran through my reading from the Twine IDE, which necessitated my stalling a bit between songs (something I'd been hoping to avoid) while closing windows of completed songs, locating and opening new windows for upcoming songs, scrolling to the top of passages prior to getting their songs started, etc. In short, a lot of hassle. It turns out that I needed the inter-song banter anyhow to explain to the audience how and why the new song-section they were being delivered wasn't necessarily a best match for the choice they thought they had made (though as the drummer pointed out, if I'd made less of a fuss about it, due to the somewhat random stream-of-consciousness flow of the work's plots, if I hadn't been so explicitly apologetic and salvage-spinning, they might not have clued in that I was only accepting half of their input -- not a great track record for a piece fundamentally about taking audience input, but admittedly still more than most music pieces offer. Far from a complete success, but definitely a partial one.)

In a nutshell, I'm very glad that I pursued this trailblazing avenue (though I would definitely think twice before penning a sequel -- it takes a lot of time to write the songs filling out an exponentially expanding tree of bifurcating branches, though of course with greater planning a more streamlined graph can be plotted which accepts player choice but still manages to direct a focused narrative down a hallway of sorts); there were elements I struggled with but they virtually all arrived at after a great deal of consideration and, well, soul-searching: by and large most problems arose as a calculated lesser of two evils rather than as unexpected trouble encountered along the way. I would like to continue polishing up this prototype and presenting future versions at eg. the annual Horace Phair party in Portland, the ArtsWells festival 2014, perhaps as a Fringe Festival show somewhere someday. Wherever it crops up, I'm pleased that I've concocted a clever and charming gimmick that allows us to offer something musical we can deliver that no one else really can: I can go out there with my band and play a game with audiences.

But first I will need to finish writing it.

With this out the door in a beta state of sorts, I'm prioritizing working on a couple of non-hyperfiction interactive fiction works, but I look forward to continuing to flesh it out (eg. Horace Phair cited above is in just three weeks). It means that I should also look at coming up with some more traditional gamebook materials to share here on this blog and take a break from my project status updates; it turns out that I've already picked much of the easy-to-convert low-hanging fruit, and curiously as I get older (and, well, after a somewhat frosty exchange with Emily Short) I am putting a higher priority on posting these conversions with the permission and blessing of the original authors. Further, I just don't have a great deal of free time these days, a vital fuel for this blind brute force technique of presenting the works: a programmer can figure out a way to get you the complete lyrics to "Ninety nine bottles of beer on the wall" much quicker than a mere typesetter can. But I have a pile of half-finished conversions, so we'll see if I can't get a couple of them wrapped up before the year's end.