The William J Mitchell Archive: Digital Historiography and the fight against banality

Writing in the Introduction to The Electronic Design Studio a book he co-edited William J Mitchell was to proclaim that architects should:

“should think of design systems as open, flexible, constantly evolving knowledge-capture devices rather than static collections of familiar tools and dispensers of established wisdom. When we can do this, I think we will see the emergence of design systems that do not just mechanically assemble banalities, but that have real style and flair”


In 1986, Mitchell was appointed Professor of Architecture and director of the Master in Design Studies Program at the Harvard Graduate School of Design. He later joined MIT in 1992 until his death in 2010. He is perhaps best for his book City of Bits.

William J Mitchell archive

The recently acquired William J Mitchell archive at Melbourne University enables an insight into Mitchell’s career as a thinker in relation to the development of Computer-Aided Design in architecture as well as the early history of parametric design.

Mitchell’s archive points to the problems with the contemporary parametric gospels as well as raising questions about the relationships between architectural theory, historiography and digital archives. In 2011 the Dean of the Faculty of Architecture, Building and Planning Prof. Tom Kvan organised for William J Mitchell’s archive to come to Melbourne University. Professor Kvan, had been a former student and collaborator of Mitchell’s.

3.5 inch floppy discs

Included in the collection now at the University of Melbourne are a variety of digital storage media including 104 iOmega Zip disks created between 1995-2001, 51 Maxell 8mm HS-8/112 Data Cartridges first released in 1989, as well as 241 3.5 inch floppy discs of various kinds created between 1985 and 1993.


In the mid to late 1980s, Mitchell was exploring the capabilities of the Apple Macintosh computers in the network he was building at Harvard’s Graduate School of Design (GSD). In 1986, there were three Apple models available on the market these were the Macintosh Plus, the Macintosh 512Ke, the Macintosh XL. All of these models were aiming at the consumer and education markets.


The disks in the Mitchell archive has been examined using Forensic Toolkit (FTK). FTK is mostly used in digital forensics and law enforcement but can be directly applied to archival contexts.  FTK can interrogate digital media to identify, preserve, recover and analyze digital information.


As result of our work with FTK The software title that appears to predominate on the 3.5-inch disks in the archive relates to Topdown. These disks includes disks that were student submissions as part of teaching at Harvard. Topdown was Milton Tan’s PhD thesis was undertaken at Harvard between 1987 and 1991 which Mitchell supervised; Topdown was developed and coded by Tan, but the work probably started at UCLA in collaboration with Mitchell and also Robin S Liggett who knew how to code.

 The birth of parametrics: Topdown

Topdown was a “knowledge-based design system” that also employed some parametric functions. While AutoCAD 10 had been released in 1988 Topdown was not merely a linear drafting tool drafting tool like AutoCAD. Topdown was developed in Lightspeed Pascal for the Macintosh and in Microsoft Pascal under Microsoft Windows for the IBM P5/2. But in 1989 the system was not yet able to utilize object-orientated programming.


As Mitchell was to liken, it at that time it, most computer drafting systems were seen to “work in a bottom-up fashion.”  As Tan and Mitchell contended Topdown was software that was an attempt to make a tool inherently parametric. Mitchell contended that a “CAD system should automatically maintain structure as a designer manipulates a geometric model.”

In using Topdown in his teaching, Topdown appeared to allow Mitchell’s students to replicate the generative exploration and selective decisions of classical architecture. However, the classical orders were no longer drawn but selected and elaborated via the graphical user interface using a Macintosh personal computer.

Emulation for digital historians 

Further research into the archive will allow for the emulation of the Topdown program itself.  Emulation is the concept that a computer program in one digital device can imitate or mimic another application or device without needing the original hardware. In other words, digital entities or fragments produced on earlier machines can be emulated on later platforms allowing them to be viewed as if they were running on the original system. The development of emulation services suggests that architectural historians must also know computer languages, hardware, networks, storage and memory capacities, graphics, user interfaces, display technology, as well as regimes of application and software development and individual applications.

Shape grammarists 


Mitchell’s and Tan’s work and effort to develop Topdown followed on from the shape grammar work of Stiny and Gips. In the early 1970s, it was Stiny and Gips (above) who advanced theoretical notions of shape grammars and parametric design in a series of papers.


Whereas Stiny saw the link between shape grammars and abstract art, Mitchell saw the connection between shape grammars and the games and systems inherent to classicism in architecture. This is obvious in the publication in 1990 of The Logic of Architecture: Design, Computation and Cognition.In many respects, there was an attempt to replicate the systematic and layered nature of classicism in the Topdown software.


Positioning the digital fragment 

The Mitchell archive suggests the need for architectural historians to develop new methods of archival research. Given the history of incremental and disruptive innovation in computing, the architectural historian and digital archivist must recognize this state of flux. The history of the digital model in architecture is only a partial history if it does not position the digital fragment into its appropriate technological context.

The shift to digital practices, as exemplified in the Mitchell archive, indicates a need to revise principles governing museology, collections, and conservation. In this context, is it merely a matter of developing new technology based methods and practices to extract historical data? Or does the extant digitization of an archive shift how fragments emerge and are selected as quotations in historical narratives?

Forget chronology, style and type.

Historiographical theories and methods can no longer rely on the old taxonomies of physical media based on the verities of chronology, style, and type (isn’t Parametrics now all about overturning the old orders?) In relation to digital archives, instruments of architectural theory and criticism will need to be cognisant of technology. Digital archives suggest the need for socio-material or technical methodologies that account for the materiality of technology and its interconnection with architecture.

In some ways architecture, its canon, norms, and histories is now a creature that has become a kind of cyborg. Architectural theory as an abstract game, as it is for some historians and theorists, sourced in critical theory, continental philosophy and logic will no longer suffice. I would suggest that in this new digital context, historiographic perspectives that recognise socio-technical work and digital ethnographies of the social sciences is the work that needs to be done.

Assembling the banalities.

Mitchell’s Topdown indicates that he saw parametric design as a way to attempt to link the canon and norms of architectural design to computing. It was an effort to both translate and assert architecture as a canon of knowledge into and against the crude digitization of architectural knowledge through machine computing and an effort countering the mindlessness of engineering optimization and efficiency. The paradox is that parametric design methods have themselves now come to exemplify an architectural approach that appears to mechanically assemble the banalities that Bill Mitchell himself loathed.

This blog is based on a paper was written in conjunction with Peter Neish the research data curator at Melbourne University. It was presented at SAHANZ 2017 you can access the full paper here. We would like to acknowledge Geoff Laurenson, and Millicent Weber, who imaged and cataloged the disks as well as Naomi Mullumby, Sarah Charing and Professor Tom Kvan for their role is securing and documenting the archive.

The Failure to Fail Fast: The parametric and BIM fail in architecture.

I recently came across two quite disparate fragments of knowledge in my travels across the so called interwebs. The first was an article at e-flux, one of those curiously named architecture websites, by the eminent American architectural academic Joan Ockman. Ockman’s article, which can be found here, details the trajectories of history and theory in architectural discourse since the 90s.

Elon is really Iron Man 

The second fragment was more fleeting. This was a glimpse, as one tends to get these days when scanning and cramming your brain with your social media feeds. I saw a post in my Facebook feed about Space X.  Space X is Elon Musk’s, the Iron Man like entrepreneur, attempt to develop cheap low earth orbit rockets. In the process a few of Elon’s rockets have crashed.

As Elon says, “Failure is an option here. If things are not failing, you are not innovating enough.” This led me to a few other interweb mantras and business school type aphorisms such as: “Innovators today are told to run loose and think lean in order to fail fast and succeed sooner” and of course there is all the Lean Start-up and Lean Design thinking encapsulated in the “Fail Fast, Learn Faster, Deliver Fastest”

It was then easy to worry if I had missed something in the past few years. To worry if I missed the whole lean design movement. Had I missed another potentially career propelling and thought leading bandwagon?

The managerial ethos. 

Thankfully, Ockman’s article had a few choice quotes that helped me to think a bit more deeply. The article helped me to join the dots, in my mind, between the proponents of the lean design, fail faster, movement and the unfolding catastrophe that is the digital “revolution” in our profession of architecture. A “revolution” disguised in futuristic rhetoric that is diminishing the domain and agency of architectural practice and knowledge. Ockman writes:

“Now that capitalism is the most revolutionary force in the world, a triumphant managerial ethos has given rise to a host of new specializations laser-focused on issues of optimization, performance, and delivery.”

Citing the last issue of Assemblage, the influential (and oh-so-pedigreed) architectural journal, as a point at which political and critical theory departed from architecture, she argues that:

“Instead of history/theory today, what we now have is research. Research is the holy grail of contemporary architecture education, and the “laboratories” in which it is carried out–by white-coated architectural technicians, figuratively speaking–are its shrines. As for criticism: arguably, we now have something like “curation.” History/theory has turned into research/curation.”

In the current climate of neoliberal universities we, myself included, all prey to the idea that curation is research (but that’s probably the topic of another blog).

Productive creatives

But then, just before I got diverted into a curation-is-research reverie, there was this little gem:

“Yet in an increasingly commodified system in which architecture students are in training to become future members of a productive (and debt-ridden) class of “creatives” and, at the same time, are not shy about exercising their rights as educational consumers, the tradition of scepticism and negativity associated with critical thinking holds less and less allure.”

The need to fail

My thought linking all these interweb fragments is that the education, research and digital practices now inscribed in the global system of architecture does not allow architects to fail. It doesn’t allow us to fail quickly enough.

I don’t think Parametric design has failure built in to its processes. In the studio, once the designer is committed to a particular digital model it becomes a kind of juggernaut. Once the model’s relational geometries are loose, design is then just addition and refinement; addition and refinement in the service of optimisation.


As architects we are not teaching our architecture students to fail. In the neoliberal university it is easier to teach the architecture students that everything is ok.  Consequently, design teaching has become focused on serial techniques and technical problem solving disguised as “productive” and waste minimising techniques. A lather of doing good for the world.

The rhetoric of techno-future 

What also bound my own disparate thoughts together is the thought that the rise of the digital in architecture and its associated rhetoric of the future has, by and large, escaped critical scrutiny. The abandonment of theory in the 90s, in the name of a post-critical position, in architecture has led to the erasure of politics in our discourse. It is worth reading through Ockman’s article to see the outline of this history.


Both BIM and Parametric design seek to optimise and configure patterns rather than socio-material systems. In both of these architectural methods, the everyday experience of the user has been replaced by the gaze of the operator, design iteration has become the spinning of the model in the shimmering screen, experimentation has become additive rather than truly generative, collaboration is reduced to the efficient exchange of data and there is no sense that architects should learn how to fail and fast in the design process.


Let’s face it, Parametric architects don’t care much for history and theory. Who needs it when you need a job in an office after you graduate. Who needs the politics of the everyday when you can play in the spectral sphere of the digital. As a result parametrics as a movement in architecture has done little to free architectural discourse from a global system that perpetuates: entrenched privileges of professional strata, a culture of design optimisation and design research that is techno-utilitarian rather than thought provoking.

As noted here Architects need to think more about digital disobedience.

The Rise of the Box Building: Bananas in Pajamas and BIM software.

I fear that the latest digital software dictates our design decisions without architects really thinking that much about it. Instead of jumping on the new great new BIM technology bandwagon we need a different debate.  I worry that some of the readers of this blog may be a bit tired of my seemingly old school rants about the hazards of digital technologies and design. But they need to be voiced, or written about, before it is too late. Architects need to resist architectural design and design knowledge becoming a sub-system of a commodified production process. Debating the merits of the prevalent software brands is critical to developing a resistance to anything that diminishes our field of knowledge.

B1 and B2: The predominant global software brands 

Let’s call the two predominant software tools beloved by our profession B1 (Trade name of large white almost extinct animal) and B2 (Weird trade name that conjures up Cousin IT). How did our discourse become beholden to these global brands? In order to protect the guilty, I prefer not to name them by their trademarked and branded names. That would give their developers too much dignity. You can work out who I mean.

Like the Bananas in Pohjamas, also named B1 and B2, both are entities that are the result of the new digital media arena that architects work in. A landscape, dare I mention it, intertwined with the emerging digital-military complex. I wont dwell on this broader point, as today I would like to focus a bit on the hazards of B2.

Firstly, for those readers who need further prompting, B1 is a software brand with animal logo of an almost extinct mammal. In total, there are only about 20,000 of the white species of animal left. B1 allows you to design and create plastic and fluid curves and shapes. Arguably, and supposedly, B1 allows you to generate a design. The emphasis here being on the word generation.

The B2 database

In contrast B2 is different to B1 it is not a modelling tool. It is essentially a database. Yes, an actual database that allows you to do some 3D drawing. You can even do 4D in B2. Wow. Googly Moogly Batman: you can slowly watch the Banana being peeled in order to meet supply chain logistics and OH&S logics. All the information created by the B2 can then be used as the B2 created banana withers and dies. All very sustainable. Or so it is claimed.

B2 does have some add-ons which augment it. But in the rush for technical skills, and post graduation jobs, many students and indeed studios are being hampered by the lack of generative capability. Disturbingly, I am starting to see more and more design studios employing the B2 software tool as a generative and primary tool. No conceptual drawing, no generative diagrams, no annotated sketches, no exploration of options, no physical models. Just jump in and start the model.

CAD and BIM Monkey Magic

Who needs the fluff of design when you need the BIMMY B2 skills to get a job, to become a CAD BIM monkey eating B2 bananas. Who needs that when you can quickly whip up an orthogonal framework and put stuff into it. Yes, using B2 in a design studio you can quickly develop a convincing orthogonal structural frame; and a so-called system; and  lo and behold fill your overall frames with some little boxes; or even slightly bigger boxes; Holey Moley Batman these could be rooms: you can then easily pretend you have designed and actual building. A building that is little more than an overall orthogonal frame filled in with boxes and frames and segments.

Pleasuring the reward centres 

But B2, unlike B1, does not create a NURBS wonderland and it has a limited ability to manipulate individual polygons. The pleasure and experience of using B2 is quite limited. You can easily pull stuff out of the B2 database, as that is what it is made for. Coffee tables, dining tables, office tables, chairs, sofas and trees. Not to mention all sorts of windows and doors. It’s not about generative design: It’s about scrolling, clicking and selecting and then placing. Not so different to Ebay. Each time an architect undertakes this process in building a digital model, a reward pulse goes from your eyes once the database object is placed, to the reward centres of your brain. You then feel good using a database even though you have populated your building model with slop. You feel like you have achieved something. You feel as if the model you are working on is real.

Architectural Design requires thought and effort to conceive, generate, manipulate and then recast. It is an iterative process. Sometimes, two steps forward and one step back. The upshot is that with software tools like B2, limit this process, and encourage the least course of resistance to be followed in the design process. Architectural studios and graduate schools are quickly becoming populated with the results of an over use of B2. Our discipline is getting getting swamped with B2 boxes.

A guide to recognising the B2 designed Box

These projects are B2-like boxes, they are easily recognised, and the following guide should help you to spot them as well.

1. They are boxes: Usually with a few additions and subtractions. Addition, subtraction, orthogonal segmentation and division are about the limits of compositional nuance. Of course, you might find a few abberrant curves, But these will be outliers.

2. They are boxes: The box finishes at the lines of its border. Everything is contained within and there is no effort to either extend or consider how the design might extend into or be a part of a surrounding context. No need to think of architecture’s broader urban responsibilities. The bunny is definitely in the box.

3. They are boxes: and utilise a segmentation that is commensurate with the most advanced, but simplistic, prefabricated building techniques. It’s always a melange of concrete and aluminium panels. No need to think about constructional craft or detailing. Its flat packed world of timber and chipboard.

4. They are boxes: and whilst a section may have been cut through the model for display, it is at worst a section that shows an undifferentiated layer cake of walls and ceilings, at best a few gaps have been dropped out or erased to make some interior spaces. There is no crafting, shaping and contouring of sections.

5. They are boxes: The only light that illuminates these creations is the final oh-so-awful V-Ray renders. B2 software does not allow the architect to think about how light might enter or be manipulated in these toxic creations. I mean who cares. You can’t dial up or select actual light from the database.

6. They are boxes: They are dumb and inchoate boxes that have abandoned architectural theory and history except in the most superficial way. There are no cultural tones or thoughts in these creations. No authentic design research and experiment. They are lacking in irony and there is never any subversive hint or self-awareness in their own making. I hate it when these types of projects win prizes.

Architects decried the modernist box of the 1950s international style. But these new boxes are more insidious. Who needs a critical theory of architecture when you can appease your pleasure centres by using a cool database. Who needs theory when you can be part of the B2 Banana cult future.

A future that is a retrograde technological utopia devoid of architecture.