lamberto coccioli

on music and beauty

The problem with MaxMSP

I had the first glimpse of what was then called simply Max in 1994, when my good friend Diego Dall’Osto introduced me to the software, but I only started using it in 1996, when working at Centro Tempo Reale in Florence. At first, like so many other composers, I was completely taken by the power and beauty of a programming language that allowed me to work in a graphical environment and test the results on the fly, without having to compile the code. Moreover, I had two wonderful mentors, Nicola Bernardini and Alvise Vidolin. They gave me generous advice and help, so that I was soon able to develop my own patches without having prior programming skills.

Soon, though, a number of issues with Max started to emerge, and in various ways, they are still unresolved ten years later. To be fair, many of the issues depend on the way Max, now MaxMSP, is used, but I still find it surprising that David Zicarelli and his company have not acted more energetically to adapt the software to the needs of the growing Max community. I will look at MaxMSP from the two angles that interest me most, usability and sustainability, but first I will try to answer the question of whom this software is written for.

I think that the main problem with MaxMSP is the fact that it sits in a sort of no man’s land between programming languages and software applications. It is too cumbersome and prescriptive as a programming language, but it lacks the user interface and a consistent set of tools that we usually associate with commercial software packages. It may be retorted that this perceived weakness is in fact the main strength of MaxMSP: to give total freedom to artists and musicians so that they can develop their own interactive set-ups without the rigid constraints of commercial software but also without the need to become programmers. My opinion is that in the long term, and looking at the way MaxMSP is now the de facto standard in performing music with live electronics, the problem has become more acute.

Composers that go past MaxMSP’s rather steep learning curve greedily embrace the programme, and start developing their patches, either from scratch, or using existing objects or libraries by members of the community. In the first case they often end up with very inefficient and buggy patches, in the second they create many dependencies, limiting portability and sustainability of their work. Max is great at two things – experimenting with your ideas and prototyping virtual set-ups – but as soon as you enter production mode, it becomes quite unfriendly. There is a historical reason for this; Max was first developed at IRCAM, an institution characterised by a rather rigid separation between composers and music technology assistants. The idea was that composers dealt with the creative part, while the assistants provided a human interface to the technology tools. This meant that the code was looked after by the technologists, and composers didn’t need to engage directly with it. Also, a big institution like IRCAM ensured the long-term preservation of the works, by employing assistants to maintain and upgrade the patches as needed.

This initial dichotomy is part of MaxMSP’s genetic code: the software is used mainly by composers and artists, but is written for programmers. This is why I find difficult to identify the target audience of the software: it is too complex and difficult to learn to be mastered fully by artists, but its true potential is wasted in the hands of programmers, who will also complain that as a development platform MaxMSP lacks many important features. In fact, I haven’t found yet a good application built with MaxMSP. So it looks like the MaxMSP target user is either a highly talented composer-technologist, equally versed in computer programming and music composition, or a creative team supplying the required skills. Not surprisingly, MaxMSP is frequently found in higher education.

Let’s look now at MaxMSP from the usability perspective. MaxMSP provides out-of-the-box quite a number of graphic objects, and has recently added powerful new functions, like Java support, the ability to encapsulate/de-encapsulate patches and create/save prototypes [template patches that can be used everywhere]. Nevertheless, the actual user interface is entirely the responsibility of the user – there are no standard Graphic User Interface models or templates. The result is that a given patch – say a sound spatializer – can be realised in many different ways, each one providing a very different user experience. Testing and comparing of patches is thus made very difficult, as the same spatializer engine can be visualised omitting certain parameters altogether or hiding them in remote subpatches. Sharing of patches, or having your live electronics performed by someone else, is also compromised, since every user builds their patches according to their personal needs and taste. If you add the fact that MaxMSP has no easy way for commenting or documenting patches, you see how hard it can be sometimes to reconstruct signal and control flow in a complex patch, even for the person that wrote it!

Probably it is from the sustainability point of view that MaxMSP fares worse. The software gives artists and musicians the impression to be in control, but in fact locks them into a closed system, difficult to scale, adapt or maintain over time. I’m talking here mainly from the perspective of live electronics concert performance, the kind of mission-critical application where everything has to work as planned. My experience over the years is that in order to work properly a MaxMSP patch has to be tweaked or rewritten every year or so, especially if external dependencies are included in the patch. In some cases, objects and libraries are not upgraded when the software is, and an alternative must be found or developed from scratch. Conflicts between objects with the same name can also prevent patches from functioning properly.

As I said, MaxMSP is an invaluable tool for trying out ideas, experimenting and prototyping, but falls short of usability and sustainability requirements, the two areas that matter most for a creative, musical use of the software and for the long-term preservation and maintenance of patches and the artistic works that depend on them. MaxMSP remains the first choice for musicians working with live electronics, but I think I have identified a gap that needs to be filled if we really want to empower musicians and offer them more accessible tools for interacting with technology.

4 Comments

  1. Hello, I wrote a piece on my blog about this article. I like your article, it got me thinking about Max πŸ™‚

    Please have a read and tell me what you think.

  2. http://www.nici.kun.nl/mmm/papers/dh-93-b.txt

    Above should be a link to a fascinating series of letters regarding Max published in Computer Music Journal back in 1993. It addresses many of the same conceptual issues you speak about and features the opinions of many key academics involved in computer music composition.

  3. good reading, but in the end it is you, who is still differentiating between composers and programmers … are you coming from IRCAM, too? πŸ™‚

    • Hi 110, Don’t you think it’s a useful distinction, or at least, two extremes of the same continuum? I’m certainly not coming from IRCAM, but the whole point of the post was to demonstrate that composing music is fundamentally different from writing code. I’m all in favour of empowering musicians to compose and perform live electronics as part of their musical process, but I believe this can only be achieved by providing musicians with tools designed for them, allowing them to think “in music”, rather than forcing them to become programmers.

Leave a Reply

Your email address will not be published.

*

© 2017 lamberto coccioli

Theme by Anders NorenUp ↑