gml in Constant 2015


ng the story to data

In the summer of 2010, Constant commissioned artist and
researcher Evan Roth to develop a work of his choice, and
to make the development process available in some way.
He decided to use a part of his fee as prize-money for
The GML-Recorder Challenge, inviting makers to propose an Open Source device ‘that can unobtrusively record
graffiti motion data during a graffiti writer’s normal practice in the city’. In three interviews that took place in
Brussels and Paris within a period of one and a half years,
we spoke about the collaborative powers of the GMLstandard, about contact points between hacker and graffiti
cultures and the granularity of gesture.
Based on conversations between Evan Roth (ER), Femke
Snelting (FS), Peter Westenberg (PW), Michele Walther
(MW), Stéphanie Villayphiou (SV), John Haltiwanger (JH)
and momo3010.
Brussels, July 2010
ER
FS

So what should we talk about?

Can you explain what GML stands for?

GML stands for Graffiti Markup Language 1 . It is a very simple fileformat designed for amateur programmers. It is a way to store graffiti
motion data. I started working with graffiti writers, combining graffiti
and technology back in New York, in 2003.


ue http://csugrue.com
Jamie Wilkinson http://www.jamiedubs.com

214

Tying the story to data

and upload it to an open database. The 000000book-site 8 hosts all this
data and some people are writing software for this.

So there are three parts: the GML-standard, software to record and
play and than there is the data itself – all of it is ‘open’ in some way. Could
you go through each of them and talk about how they produce uploads and
downloads?
FS

Right. It starts with Graffiti Analysis. It is software written in C++
using OpenFrameworks, an Open Source platform designed by artists for
visual applications. Right now you can download the recorder app and
from that you can generate your own .gml files. And from there you can
upload these files into the playback app. In the beginning that was the
only Open Source side of the project. Programmers could also make new
applications based on the software, which also happened.
Last night we met St


mping-off point for making their own
work.
Second was the database. We had this file-format that we loosely defined.
I worked with Jamie to develop the 000000book site. It is pretty nutsand-bolts but you can click ‘upload’ and click on your own .gml files and
it will playback in the browser. People have developed their own playback
mechanisms, which are some of the first Open Source collaborations that
happened around .gml files. There is a user account and you can upload
files; people have made image renderers, there are people that have made
Flash players, SVG players. Golan Levin has developed an application
that converts a .gml file into an auto-CAD format. The 000000book site
is basically where graffiti writers connect to developers.
In the middle between Graffiti Analysis and database is the Graffiti Markup
Language, that I think will have its own place on the web. But so


jects I had a
lot of experience of meeting graffiti writers as a userbase. When you meet
with them, they tell you right away what pieces of the software they think
are missing. So from talking with them we developed a lot of features
that now are in GML like brushes, drips, line-thickness. Some people
had single line tags and some people had multi-line tags so that issue
came up because GML tracks both drawing and non-drawing motion so
we knew that we needed in the file format to talk about pen up and pen
down. I was interested in the connection points between lines also.
We tried to keep it very stripped down. From the beginning we kne


d participate as developers or anonymous contributors were not going to be the same people that would develop a Linux
core. They are students, people just getting into programming or visual
programming. We wanted people to be able to double-click a .gml file
and than everything should verbally make sense so it is Begin stroke.
End stroke. Anyone with basic programming skills should be able to
figure out what’s going on.
ER

Did you have any moment where you had to decide: this does not belong
to g


ng this database.
Now it is not about my personal collection anymore, it is making a handshake between two communities; making them run off with each other
and having fun as opposed to me having to be there all the time to make
introductions.
ER

Is GML about the distribution of signature? I mean: The gestures of
a specific person can now be reproduced by a larger community. How does
that work?

FS

This is an interesting conversation we should have with the graffiti
writers. A tag might be somethin


OK with it and understand
it. Initially when people see something they’ve made being used by other
people, a lot of times it can be a point where a red flag is raised and I am
assuming there are more red flags going to go up.
FS

If you upload a .gml file, can you insert a licence?

Not yet. Right now there is not even a ‘private mode’ on the
000000book site. If you upload, everything is public. There is a lot of
interesting issues with respect to the licence that I have been reluctant to
dea


r to code than to a video file?
FS

is?

ER

Functionally it is code. But would a graffiti writer know what GPL

I am interested in the apprentice-system you were talking about earlier.
Like a young writer learning from someone else they admire. The GML
notation of x-y-time might help someone to learn as well. But would you
ever really copy someone else’s tag?
PW

One of the reasons I think graffiti writing has this history of apprenticeship is because you don’t really have a chance to learn oth


istory with documentarians entering into their community and playing a role but sharing the stories is something writers do
internally, not as much to outsiders. How do you figure out a way to get
graffiti writers to document their stories into the .gml files themselves,
or is it going to take outsiders? How does the format facilitate that?

Do you think the availability of a project like GML can have an impact
on the way graffiti is learned? If data becomes available in a community
that operates traditionally through apprenticeships and person-to-person
sharing, what does it do?
FS

I am interested in Open Source culture being influenced


, a way to have giant paintings circulating in
the city ... There is a lot of room to explore there.
ER

Your experience with the Blender community 14 did not sound like an
easy bridge?

FS

Recently I released a piece of software that translates a .gml file and
translates it into a .stl file, which is a common 3D format. So you can
basically take a graffiti gesture and import it into software like Blender.
I used Blender because I wanted to highlight this tool, because I want
these communities to t


onference
and gave a talk where I explained sort of where I see these things overlap,
I could make a better case than the three minute video they reacted to.
ER

What about Gesture Markup Language instead of Graffiti Markup
Language?
FS

Essentially GML records x-y-time data. If you talk about what it
functionally does, it is probably more related to gesture than it is to graffiti. There is nothing at the core specifically related to graffiti. I am
interested in branding it in relation to graffiti a


to track multiple balls
in the air? I keep calling it Graffiti Markup Language because I think it
is a good story.
ER

15

http://www.blendernation.com/2010/07/09/blender-graffiti-analysis

222

Tying the story to data

PW

What’s the licence on GML?

We haven’t really entered into that. Why would you need a licence
on a file format?
ER
FS

It would prevent that anyone could own the standard.

That sounds good. Actually it would be interesting for the project, if
someone would try to licence it. Legal things matter, but for the things I
do, I am most of all interested in getting the idea across.
ER

I am interested in the way GML stems from a specific practice. How
it is different and similar to large, legal, commercial, global standardization practices. Related, how can GML connect to other standard practices?
Could it be RDF compliant?

FS

PW

Gesture recognition to help out the police?

Or maps of places that are in need of some graffiti? How to link GML
to other types of data?
FS

It is hard for me to imagine something. But one thing is interesting
for example, how GML is used in the EyeWriter project. It has not
so much to do with gesture, but more with how you would draft in a
computer. TEMPT is plotting points, so the time data might not be so
interesting but because it is in the same format, the community might


the time, to use it with his Print Ball project. He took the
tag data from a paralyzed graffiti writer in Los Angeles and painted it on
a wall in Dublin. Eye-movement translated into a paint-ball gun ... that
is the kind of collaboration that I hope GML can be the middle-point
for. If that happens, things can start to extrapolate on either end.
ER

You talked about posting a wish-list and being surprised that your
wishes were fulfilled within weeks. Why do you think that a project like
EyeWriter, even if it interests a lot of people, has a hard time gathering
collaborators, while something much more general like GML seems to be
more compelling for people to contribute to?
FS

16

Benjamin Gaulon, Print Ball
http://www.eyewriter.org/paintball-shooting-robot-writes-tempt1-tag

223

Tying the story to data

I’ll answer that in a second, but you reminded me of something
else: because EyeWriter was GML based, a lot of the collaborations
that happened with people outside of the project were GML related,
not EyeWriter related. So we did have artists like Ben and Golan take
data drawn by TEMPT and do completely different things which made
TEMPT a collaborator with them in a way. The software allowed him to
share his work in a format that allo


the obvious thing that did not work. I have
friends working on OpenFrameworks, it is headed primarily by Zach and
Theo. When you see that forum, it is very involved. It is a deep system,
with many different libraries and lots of code flying around. GML is really
not large enough.
I think what makes sense for this project is when I post news about the
project, I see it ripple in Google Alerts. For people working on it, having
a place where these things show up is already a lot. The biggest success
is the project space, to see all the projects happening.
ER

FS

What happened on the site since we talked?

A project I like, is kml2gml 24 for example. It is done by a friend from
Tokyo. He was gathering GPS data riding his bike around various cities,
and building up a font based on his path. I like projects like this, where
ER

24

Yamaguchi Takahiro http://www.graffitimarkuplanguage.com/kml2GML

228

Tying the story to data

someone takes a work that is already done and just writes an application
to convert the data into another format. To see him riding his bike played
back in GML was really nice. It is super low barrier to entry, he already
did all the hard work. I like that there is now a system for piping very
different kinds of data through GML.
FS

But it could also work the other way around?

Yeah. This is maybe a tangent but depending on how someone solves
the GML challenge ... I was discussing this with Mike (the person that is
developing the sensor based version). He was thinking that if you would
turn on his system, and leave it on for a whole night of graffiti writing,
you would have the gestural data plus the GPS data. You could make
a .gml file that is tracking you down the street, and zoom in when you
start making the tag. Also you would get much more information on
3D movement, like tilt and when the pen is picking up and going down.
Right now all I am getting is a 2D view through vi


aw on a huge scale for one
letter. You could not record anything small.
ER

But it could be nice if you could record with a device that is less conspicuous.
FS

I know. I have just been experimenting with mounting cameras on
spray-cans. A tangent to GML, but related. It is not data, but video.
ER

What do you think is the difference between recording video, and
recording data? You mentioned that you wanted to move away from documentation the image to capture movement. Video is somehow indirect
data?
FS

Video is annoying in that it is computationally expensive. In Brazil 25
I have been using the laptop but the data is not very precise.
Kyle thinks he might be able to back out GML data from videos. This
might solve the challenge, depending on how many cameras you need and
how expensive they are. But so far I have not heard back from him. He
said it needs three different cameras all looking at the wall. I mean: talk
about compu


f this
is going right now, I am just trying to get the footage; I put these pieces
together showing all this movement but I don’t really know what the final
project is. It is more about collecting data so I am interested in having
video, audio and GML that can be synced up, and the sound from these
microphones is something to do something with later. This is research
for me. I like the idea of having all this data related to a 10 second gesture.
I am thinking that in the future we can do interesti


the head-mounted cameras were smart. Unless
you walk in front of a mirror (laughs).
ER

FS

Is it related to the dream of ‘self documenting code’?

I like that. Even doing the challenge is in a way a reflection on this,
how I am fighting to get GML back to the streets somehow, it has a
natural tendency to get closer to the browser, to the screen, and my job
is to get it back to the street. It is so sexy and fun and flashy and that is
important too. My job is to keep the graffiti influence on it


code-based set of rules? If the files were raw video,
it would encourage people to go outside more often? By picking XML
I am defining where the thing heads in a way. I think I am OK in the
role of fighting that tendency. It is not just a problem in GML but with a
lot of work I have been doing with graffiti and technology and even way
back with Graffiti Analysis, before GRL (Graffiti Research Lab), the idea
was always to keep the research very close to the people doing graffiti. I
was intentionally


that the Gothic lettering
is inspired by administrative papers!
SV

I am wondering whether you’re thinking about the standard itself as
a space for hacking?
FS

Graffiti is somehow coded in-itself. Do you mean it would be interesting
to think how GML could be coded in a way for graffiti writers, not for
coders?
There would be more space for that when more people start to program at
a younger age? When it is more common knowledge. If I would start to do
ER

28
29

Interview with François Chastane


ion of the data. In the end Zach said, you are just fighting this
because you have a motion project and you want to project motion and
then I said alright, I’ll do a test. And the tests were so exciting that I felt
OK with it.
ER

In what way does GML bridge the gap between digital drawing and
hand writing? Could you see a sort of computer-aided graffiti? Could you
see computation enter graffiti?
FS

Yeah. When you are in a controlled environment, in a studio, it is
easy but the outdoors part alwa


xt challenge I’d like to run is about recreating the data
outside. I’ve been thinking about these helicopters with embedded wireless
ER

234

Tying the story to data

camera’s, have you seen them? The obvious thing to me would be uploading
a .gml file to one of these helicopters that is dripping paint on a rooftop.
Scale is so important, so going bigger is always going to be better.
Gigantic rooftop tags could be a way to tie it back to the city, give it a
reason? I am thinking of ways to get


a

The Graffiti Markup Field Recorder
challenge

An easily reproducible DIY device that can unobtrusively record graffiti motion data during a graffiti writer’s normal practice in the city. 30
Project Description and Design Requirements:



The GML Field Recorder Challenge is a DIY hardware and software solution for unobtrusively recording graffiti motion data during a graffiti writer’s
normal practice in the city. The winning project will be an easy to follow
instruction set that can be repr


hlight to a graffiti implement. I didn’t want anyone solving the
problem and then, Step one is: ‘Attach a police siren to a spraypaint can’

239

Tying the story to data



The resulting solution should be able to record at least
10 unique GML tags of approximately 10 seconds each in
length in one session without the need for connecting
to or using additional equipment.

I wasn’t thinking this was going to be an issue in terms of memorystorage, but maybe in terms of memory management. I


everywhere, completely crushed DC and his trial was a big deal.
This issue came up and they argued that BORF was a collective, not an
individual. Who knows if that’s true, there were a lot of people around
him, but how do you really know?
ER

FS

GML could help balance the load?

You mean it would not be just the image of a tag but more like signing
at the bank?
ER

240

Tying the story to data

I mean that if you copy and distribute your data, the chance is small
that you can link it to an indi


on
and off the spraypaint can. When you watch graffiti, it is too small of a
movement to even notice but when you are seeing that, it adds another
level of understanding of how they are writing.


All motion data should be saved using the current GML
standard 31 .
FS



Obvious.

All aspects of the winning design should be able to be
reproduced by graffiti writers and amateur technologists.

It wouldn’t be exciting if only ten people can make this thing. This
tool should not be just for peop


34 biography, and I liked his idea that ‘fun’
is part of innovation, right? In a programming sense, it is scratching a
personal itch. The attachment of a prize is more to underline the fun
aspect than anything else.
ER

I am still puzzled about GML and how it is at the one hand stimulating
collaboration and sharing, and than it comes back to the proud individual
that wants to show off. It is kind of funny actually that now two people are
winning the prize.
FS

ER

I understand what you mean.

A


ion I am on at that moment.
ER

Earlier you brought up the contrast between people that were more
visually invested and others that are more interested in the performance
aspect. I wanted to hear a bit more about the continuum in the culture and
how GML fits into that?
JH

35

Anton Marini: Some personal projects of mine, for example specific effects and ‘looks’ that I have a
personal attachment to, I don’t release
https://github.com/kylemcdonald/SharingInterviews/blob/master/antonmarini.markd


other
million of them. While others do not want people painting over them,
the city or other graffiti writers. Also if a tag has been up there for a few
years, it acquires more reverence and it is even worse when it is painted
over.
But I think that GML is different, it is really more similar to a photo of
the tag. It is not trying to be the actual thing.
FS

Once a tag is saved in GML, what can be done with the data?

I am myself reluctant to take any of these tags that I’ve collected and
do anything with it at all without talking closely to whoever’s tag it is,
because it is such an intimate thing. In that sense it is strange


its to referentiality. Like beat jacking for
DJs or biting rhymes for MCs, there must be a moment where you are not
just homaging, but stealing a style.
JH

I’ve seen cases where both parties have been happy, like when Yamaguchi
Takahiro used some GML data from KATSU and piped it into Google
Maps, so he was showing these big KATSU tags all over the earth which
was a nice web-based implementation. I think he was doing what a graffiti writer does naturally: Get out there and make the tag bigger but in
different ways. He is not taking KATSU-data from the database without
shining light back on him.
ER

GML seems very inspired by the practice of Free Software, but at the
same time it reiterates the conventional hierarchies of who are supposed to
FS

249

Tying the story to data

use what ... in which way ... from who. For me the excitement with open
li


not push some of those
boundaries myself.
ER

Even when I am sometimes disappointed by the actual closedness of
F/LOSS, at least in theory through its licensing and refusal to limit who is
entitled and who’s not, it is a liberating force. It seems GML is only half
liberating?
FS

I agree. I think the lack of that is related to the data. The looseness of
its licence makes it less of an invitation in a sense. If the people that put
data up there would sit down and really talk about what this means,


graffiti. I still hold the tag as primal!
ER

It is interesting to be worried about copyright on something that is
illegal, things you can not publicly claim ownership of.
JH

Would you agree that standards are a normalizing practice, that in a
way GML is part of a legalizing process?
FS

250

Tying the story to data

For that to happen, a larger community would have to get involved. It
would need to be Gesture Markup Language, and a community other than
graffiti writers would need to get involved


, Hip Hop and Free Software
are both global insurgent subcultures that have emerged from being kind
of thrown away as fads and then become objects of pondering in multinational boardrooms. So I was hoping to open you up to riff on that:
zooming out, GML is a handshake point between these two cultures, but
GML is a specific thing within this larger world of F/LOSS and graffiti
JH

36

KRS-One Master Teacher. AN INTRODUCTION TO HIP HOP .
http://www.krs-one.com/#!temple-of-hip-hop/c177q

251

Tying the story to data

in the larger world of hiphop. What othe


coming from two completely different strata, from a
class-based analysis which is absent from a lot of discussion. Even on
that level, how to integrate them to me is a political question to some
degree.
JH

ER
FS
ER
FS

Right.

Will any features of GML ever be deprecated?

Breaking currently existing software? I hope not.
Basically I’m asking for your long-term vision?

When the spec was being made of course it wasn’t just me, it was a
group of people debating these things and of course nobody


we tried to get in as many things as we could
think of and have the base stay kind of what it was with the idea that you
could add more stuff into it. It’s easy enough to do, of course its not a
super-rigid standard. If you look at what the base .gml file is, the minimum
requirements for GML to compile, its so so stripped down. As long as it
just remains time/x-y-z, I don’t think that’s going to change, no.
But I’m also hoping that I’m not gonna be the main GML developer. I’m
already not, there’s already people doing way more stuff with it than I am.
ER

FS

How does it work when someone proposes a feature?

ER They just e-mail me (laughs). But right now there hasn’t been a ton
of that because it’s

 

Display 200 300 400 500 600 700 800 900 1000 ALL characters around the word.