what even is artistic license
the point of art, subjectivity in science, the decisions of translators, and the visualization of data
i’m a TE1 in nature, ai, and performance, in which we make animal flocking patterns into music. yesterday2 i Zoomed with my boss and four other ppl (one from YC) making an app for middle schoolers to do this, and the central question was: how faithful should we be to the data? how much creative freedom should we allow?
at a certain point, infinite creative freedom just means a blank page—in which case we don’t even need the data to begin with, except maybe as some sort of reference drawing. but if we try to maximize faithfulness to the data, we run into three problems:
the process isn’t really fun anymore (which matters a lot to middle schoolers).
the process isn’t really art anymore (but does it need to be?).
true objectivity in translation (of data into art, but also other things) is impossible.
i’ll address these problems one at a time in what is the point of this, subjectivity and translation in multiple fields, and translation in multiple fields but actually.
what is the point of this
making animal flocking patterns (and other natural phenomena) into music might sound really cool, but what’s the point?
one could say that the point is to make great art, and in that case, we really should have infinite creative freedom. the data should be our starting point, our palette, but our canvas should be our own to paint as we see fit. this seems to be what many artists think the point is.
(the max-msp3 expert on the project said: “is the data the instrument or the composition?”)
maybe the point is to educate people? if we’re starting with data, maybe we should be trying to impart an understanding of this data upon the masses, who aren’t that good at reading scientific papers? this seems to be what many scientists think the point is.
or maybe the point is to raise awareness of climate change and ecological collapse. to get people off their butts and encourage them to do something about these issues. this seems to be what many environmental studies people think the point is.
(the environmental science person on the project said: “We don’t donate to the Amazon rainforest based on a scientific paper. We donate based on art.”)
depending on what you think the point is, it seems like a different amount of creative freedom should be allowed.
if the point is just to make great art (is “just” the right word here? is that demeaning?), then we should be allowed infinite freedom.
if the point is to educate people, it seems like we should be allowed very little freedom, because any artistically-motivated change in the piece will by nature lead to more noise in the data4. it’d be like if you painted over parts of a QR code because you thought it looked better, and now the QR code doesn’t work anymore.
one question i have is whether people actually gain any further understanding of data by listening to it as music. we listened5 to a piece in class where the notes on a cello corresponded to global temperature highs over time, and you could tell that the notes were rising, but that’s a piece of information anyone could understand after being told in a second. i can’t think of any information that would be hard to understand when reading a paper or watching a TED talk, yet easy to understand through music (altho lmk if you have counterexamples!).
it is true that music probably makes more of an emotional impact than reading a scientific paper. so if the point is to make an emotional impact, to get people off their butts, to raise awareness, then maybe more artistic license should be allowed.
but at that point, why not just create a composition which you think emotionally represents climate change/ecological collapse without using the actual data? i’m sure you could provoke a greater emotional reaction that way—music makes people feel emotional partly because of intrinsic qualities like relationship between frequencies and stuff, which can all be much better manipulated from scratch than if you have to pay heed to some random dataset which probably doesn’t correlate well with the musical-emotional patterns you’re looking for.
maybe the emotional impact comes from listeners knowing what they’re listening to is data. maybe the solution is to nominally use the data, but in reality try to manipulate the music a lot in order to make a large emotional impact?
anyway, how much artistic license we should allow varies with what we think the actual point of this whole thing is. it seems like some people think the point is just to grab media attention for a good cause, but other people want to do more artistic things with it. does making cool art detract from the noble effort of helping the world?
to me, it seems like the whole point of this thing is to make cool art AND help the world, and… c’mon, there must be some way to do both of these things at once, right?
subjectivity and translation in multiple fields6
one problem with the idea of sticking with the data 100% is that there are still choices to be made in the process of translation. what pieces of data do you map to what musical parameters? at a certain point, even though some decisions seem natural (the speed of an animal to musical velocity, for example), others are basically just toss-ups (what do you map to musical timbre? what do you do with an animal’s Z-coordinate?). how motivated are your choices?
to complicate this even further, the original dataset probably already has a bunch of noise and inaccuracies. a lot of people in my class were quick to pounce on this fact in our discussion as an excuse for artistic license. science is subjective, and the data is already flawed, so our piece won’t be a perfect representation anyway—why shouldn’t we modify it just a little bit more?
personally, i think this is a little bit of a cop-out. yes, every dataset has noise and probable inaccuracies, but it’s still better than anything else we have. every dataset is a map of its territory, and maps by their very nature do not have as high a fidelity as what they’re actually trying to capture. so of course data is slightly flawed and simplified. of course choices have to be made in the process of gathering it. otherwise it wouldn’t be doing its job. this subjectivity isn’t an excuse for fudging data—that’d be like if you took a map of the world and you messed up the coastlines a little bit because you thought it looked better. it might be a nice piece of art, but it’s not something sailors can use to navigate anymore.
(one possible solution: like a real mapmaker, make the parts of the map that don’t really matter beautiful, like the compass and borders, while preserving the coastlines and everything that is actually supposed to represent the world.)
disregarding the ‘subjectivity of science,’ there’s still the problem of decisions in how to translate the data to musical parameters. fortunately, these kinds of decisions are made all the time by translators in other fields, so maybe we can look to them for inspiration.
translation in other fields but actually
the man the myth the legend hans zimmer says that the number one rule of film composing is “stick to the story”. he says that the only thing you should be worrying about is your director’s vision. you shouldn’t be setting out to write pieces that are memorable, or interesting musically, or groundbreaking enough that they will advance your career. you should write for the story, and the story alone.
but at a certain point, translating a story into music for film necessitates some (ok, a lot of) artistic choice.
similarly, when composing any sort of music, there’s a line that needs to be drawn between the composer and the performer. how much is written on the page, and how much is allowed to vary between performers (and even performances)?
at the one extreme, composers can mandate which notes a violinist should play dragging their bow up across the strings, and which notes they should play moving down. at the other, some experimental composers don’t even use traditional clefs, but make weird clef art they expect performers to interpret as music.
screenwriters have to make similar choices when adapting books to movies. which scenes should stay in? which should be cut? the mediums are different, so often the story needs to change with them7.
in terms of our current undertaking—making an app for middle schoolers to turn animal flocking patterns into music—there actually seems to be three parties at play. the data is the instrument, the middle schoolers are the performers, but we the app-makers are the composers. at the Zoom meeting yesterday, we talked about how there will, at one point, be an ‘advanced’ tab in the app where users can specify which columns in their data should map to which musical parameters. but until then—and in the basic version of the app, always—we the app-makers are making these parameter-mapping choices. choices are everywhere!
i’m sure there are a billion other examples of this kind of decision-making in art, but honestly the one i’m most familiar with is the most basic implication of the word ‘translation’: natural language.
in translation theory8, there’s a dichotomy between strict and loose translation. an absurdly strict translator would look up every single word of the original language in their source text and write its definition in the target language. an absurdly loose translator would read the book in its original language, get an idea of it in their head, and then write a book in the target language that they think “captures the meaning of the original” in some way.
all translators fall somewhere in between these two extremes. to some degree all writing is translation, between what you have in your head and the words you lay out on the page, and like all translations, writing is imperfect, so maybe as translators our duty should be to minimize this imperfection—to try and get as close as we can to the original authorial intent, rather than the words the author actually put down on the page.
(this gets into death-of-the-author stuff, but i’ll write another post about that sometime soon!)
one thing that’s interesting to me is that while we consider some translators to be much better translators than others (e.g. jhumpa lahiri is super famous and amazing), and it’s usually because of the quality of their writing, it still feels a little bit taboo to say that translators should be entitled to any sort of artistic license with their work. in an optimal world, every translation of a given book X into a given language Y should be exactly the same, because the translation is what perfectly captures the original authorial intent. right? so why does jhumpa lahiri get props for being a beautiful writer?
is translation an art or a science?
my boss told us that he was talking about mapping flocking patterns to musical parameters with a friend of his at the med school, and she got a bit angry with him. she said “you shouldn’t be making any decisions. you’re just a messenger.”
like i wrote before, i do think the idea that scientists don’t make any decisions, and are just natural conduits for the truth of the world, is a lie. however, it’s still true that in science you aim to make the fewest amount of ‘decisions’ possible. you’re trying to maximize the similarity of your representation, your map, with the actuality of the world.
whereas in art, the decisions often are the art.
(although some artists might say that they too are just trying to maximize the similarity of their representation with the actuality of the original artistic vision they had in their head.)
how fuzzy is our representation allowed to be?
EM is taking a class on narrative mapmaking right now, directly following a class on data visualization, and she’s told me that the data visualization professor really liked minimalism—he hated legends and unnecessary annotation—whereas her mapmaking professor wants as much contextual information displayed as possible.
but shouldn’t the data speak for itself?
should art speak for itself?
(if so, what should it say?)
yeah, not TA. TE actually stands for “temporary expert” and it’s my (really awesome) job at the CCAM. long story short: i told my boss i was sad i couldn’t take his class next semester, he said you should audit it, i said sure and did so, and after the first class he emailed me and was like hey do you want to be paid for this class to look over ppls code as part of your CCAM job? and ofc i was like sure!!
yesterday when i wrote this; now more like a week ago
a visual programming language which turns basically any kind of data (i.e. animal flocking patterns, represented through timestamps, coordinates, and animal ids) into any other kind of data (i.e. music, represented through pitch, velocity, etc.). max can do a lot of things but is mostly used for music.
and non-random noise too, which is even worse!!
and two people performed a version of it live! it was really cool!
this section title is really linguistically ambiguous and thus exemplifies its topic of ambiguity in translation. i originally intended “subjectivity and [translation in multiple fields]”, hence the graphic, but it could also be “[subjectivity and translation] in multiple fields,” and if so, are both of them in every such field, or are there fields with only one of them?
dune 2 is amazing go rewatch dune 2 guys
if you want an introduction to this without having to read dense academic literature, go read yalie rf kuang’s book Babel. although it is unfortunately also incredibly depressing.