Ž .Automation in Construction 9 2000 387–392www.elsevier.comrlocaterautcon

ž /Phase x — memetic engineering for architecture

Urs Hirschberg ), Florian Wenz 1

Chair for Architecture and CAAD, Faculty of Architecture, Swiss Federal Institute of Technology Zurich, CH-8093 Zurich, Switzerland¨

Abstract

Ž .Phase x was a successful teaching experiment we made in our entry level CAAD course in the Wintersemester1996r1997. The course was entirely organized by means of a central database that managed all the students’ works throughdifferent learning phases. This set-up allowed that the results of one phase and one author be taken as the starting point forthe work in the next phase by a different author. As students could choose which model they wanted to work with, the whole

Ž .of Phase x could be viewed as an organism where, as in a genetic system, only the ‘‘fittest’’ works survived. While somediscussion of the technical set-up is necessary as a background, the main topics addressed in this paper will be thestructuring in phases of the course, the experiences we had with collective authorship, and the observations we made about

2 Ž .the memes that developed and spread in the students’ works. Finally we’ll draw some conclusions in how far Phase x isrelevant also in a larger context, which is not limited to teaching CAAD. Since this paper was first published in 1997, wehave continued to explore the issues described here in various projects 3 together with a growing number of other interestedinstitutions worldwide. While leaving the paper essentially in its original form, we added a section at the end, in which weoutline some of these recent developments. q 2000 Published by Elsevier Science B.V. All rights reserved.

Keywords: Memetic process; Collaborative creative work; Collective authorship; CAAD education

1. Introduction

At the chair for Architecture and CAAD, over thepast years we have gathered much experience in

Ž .using the World Wide Web WWW as a teaching

) Corresponding author. E-mail: hirsch@arch.ethz.ch1 E-mail: wenz@arch.ethz.ch.2 The term ‘‘meme’’ was first introduced by Richard Dawkins

w xin 1976 1 . In analogy to genes allowing the replication of life,the meme is, in Dawkins’ theory, the basis of cultural replicationprocesses.

3 Most of the websites of these projects are available athttp:rrspace.arch.ethz.ch.

environment for our courses. The web’s possibilitiesto communicate and to share multimedia-data havebecome an integral part of the way we teach CAAD.While this integration has many administrative ad-vantages, the main benefit in educational terms is therapid and intense exchange of ideas that it promotesbetween the large number of students that take ourcourses every semester. In the past we have achievedthis by establishing the students’ homepages as theplace where they would present their coursework.We also developed a means to get more of anoverview of the work being done: WWW-pages thatwere generated by scripts that collected documentsfrom all the students homepages and displayed themin a ‘‘panoptic view’’.

0926-5805r00r$ - see front matter q 2000 Published by Elsevier Science B.V. All rights reserved.Ž .PII: S0926-5805 99 00022-9

( )U. Hirschberg, F. WenzrAutomation in Construction 9 2000 387–392388

2. Goals

The positive experiences we gathered with theŽ .mentioned web-based means led us to the Phase x

experiment. We wanted to further intensify the men-tioned advantages, namely the exchange of ideas andexpertise between the students. Forcing the studentsto take one of their colleagues’ work as the basis oftheir design in every phase actually results in a shiftof how individual authorship can be observed. In thetraditional cultural model, individual authorship isdominant. The influences between different authors

Ž .are often unclear and always debatable Fig. 1 . InŽ .Phase x the works themselves form the primary

thread and the contribution of every author involvedin the process can clearly be shown. In this way, in

Ž .Phase x we did not only intensify this exchange ofŽideas, we also made it absolutely transparent Fig.

.2 .There is another problem we addressed with this

set-up. As our course is an elective class that stu-dents usually take parallel to design studio work, thetime students can spend doing work for our class islimited. With the many different software-tools andconcepts that are covered in our course, students tendto spend too much time overcoming operative prob-lems, rather than concentrating on the intelligent useof the computer as a design medium. In reaction to

Ž .this, in Phase x each phase consists of simple well-

Fig. 1. Gutenberg Galaxy, the traditional cultural model: differentauthors influence one another through their works, but the influ-ences are hard to trace.

Ž .Fig. 2. Phase x process: relationships between works and authorsare explicit and transparent.

defined tasks. Doing the exercises gives instant grati-fication rather than prolonged frustration. Part of thisis also the fact that the students never have to startfrom scratch: they are faced with an object that oneof their colleagues has prepared in the previousphase. Therefore their creative energy is spent onfinding an intelligent answer or reaction to this ob-ject. This is a more well-defined design task thanstarting from scratch, but certainly not a less chal-lenging one.

3. System

Students who take our course get an account inour workstation cluster. In these accounts they dotheir assigned tasks and store the results. They alsohave homepages where they can put these results on

Ž .display. In Phase x we made this use of the home-page optional, because the central database held allthe documents and related information of every workof every student and displayed them in various ways.This was realized by creating a custom WWW-inter-face to a Mini SQL database, which would manageall the data related to the course essentially in twotables: one for the authors and one for the works.

To start working on a new phase, students wouldchoose a result of the previous phase that interestedthem and request it from the system. To request awork, the users had to identify themselves with apassword. Users registered this way could not re-

( )U. Hirschberg, F. WenzrAutomation in Construction 9 2000 387–392 389

Ž . Ž .Fig. 3. Phase x interface. An Object from phase 7 with parentand children. The Interface was used to navigate through thedatabase, as well as to request and submit works.

quest another work, until they had submitted the onethey were currently working on. Upon requesting awork, a whole folder was placed in the students’accounts, containing a CAAD drawing file andsometimes other additional data. When the work wassubmitted, it became available for other students to

Ž .work with in the next phase Fig. 3 .Using the database and the authentication, it was

possible to smoothly exchange the works betweenstudents and prevent any student from using theirown results. On top of that, it also made it easy tostore a lot of additional information with every work,like the amount of time that it was worked on, whichauthors had worked on it and when, etc. This infor-mation could later be used to analyse the wholecourse in various ways.

4. Phases

In each phase, a new CAAD concept was intro-duced and applied. Starting from 2D compositions

Ž .on the plane in phase 1 , the phases got more andmore complex, covering solid modelling as well asdifferent object-oriented principles, like parametric

Žor hierarchical modelling. AutoCAD and to some.extent Inventor, resp. VRML was the tool used forŽ .most phases, except for phase 9 , where sculptor, a

Virtual Reality modelling program developed at our

Ž .chair was used and phase 10 , where light-simula-tions were created with the program Radiance.

The phases were prepared in such a way that thedesign-tasks could be fulfilled with only a fewmacro-like commands that were assembled on aspecial menu. Complexity was introduced merely bythe fact that many different authors inscribed theirpersonal ideas into the same objects. As in previoussemesters, the time students spent for a task variedgreatly. Those with previous CAAD experience, ordetermined to get a more in depth knowledge of thetools, could easily use all the standard AutoCADcommands as well. But for the ones more interestedin the concepts than in the nitty gritty of the soft-ware, the macros made it possible to concentrate onthe design quickly.

All phases dealt with geometric modelling in avery abstract, almost hermetic way. Only in the lastphases questions of architectural space becamesomewhat relevant. There is a tradition for this ap-

w xproach at our chair 6,7 . The software tools andmethods we introduced allow to concentrate purelyon the principles of composition. A result is nevertaken to be the representation of something but isalways the — though intangible — thing itself. Theideas or memes each object contains are entirelyexpressed in terms of color, proportion and geome-try, the language of computer graphics. That there isindeed such a language and that it is established

Ž .enough to make the Phase x experiment meaning-ful, was the hypothesis at the outset of our experi-ment.

Analyzing the whole body of works at the end ofthe semester, we generated analytical charts of the

Ž .development Figs. 4 and 5 . One amazing fact we

Fig. 4. y-Axis: AUTHOR color scheme: GENEALOGY nodeweight: CHILDREN connection: ACTIVE x-axis: PHASE.

( )U. Hirschberg, F. WenzrAutomation in Construction 9 2000 387–392390

Fig. 5. y-Axis: GENEALOGY color scheme: GENEALOGYnode weight: CHILDREN connection: ACTIVE x-axis: PHASE.

came to realize then, was that only four out of theover 120 different 2D-compositions that were cre-ated in the first phase had any descendants in the

Ž .tenth generation Figs. 5 and 6 . We had not ex-pected the natural selection taking place to be sodrastic.

5. Memetic engineering

Beyond the scope of an academic educationalŽ .experiment, Phase x addresses issues that touch the

core of the networked society, currently under con-struction. As we can tell from the development of the

Ž .Fig. 6. Phase x survivors.

Internet, an environment where individual ideas arebasically everybody’s freeware generates a very rapidevolution of the system itself. The British scientistRichard Dawkins first suggested in his book ‘‘The

w xSelfish Gene’’ 1 that cultural evolution is based onsimilar mechanisms as biological evolution.

Ideas or memes, as the smallest units of memeticevolution, tend to replicate by separating themselvesfrom their authors and getting picked up by thepublic. This is especially true in the digital realm,where anything can be sampled and reused by suchsimple operations as ‘‘Copy’’ and ‘‘Paste’’.

Ž .Phase x assumes this is also true for architec-tural content and is designed around the principles ofmemetic evolution. By splitting a rather complexdesign process into clearly defined units, compatiblememes are generated. These memes are first strippedfrom their authors by being placed into the publicrealm of the database and can then be copied withoutloss of substance as digital files to the next author. Infact, the total number of memes by a single authorwas the only way that the system measured the

Ž . Ž .performance of students Fig. 7 . Phase x was thusdeveloped as a showcase study for architecturalmemetic engineering and we expect that similar

Fig. 7. Performance analysis for one author.

( )U. Hirschberg, F. WenzrAutomation in Construction 9 2000 387–392 391

mechanisms will be used in the professional realm inw xthe future 2 .

6. Collective authorship

One major unsolved problem with memetic pro-cesses is that the traditional concept of authorship,which relates an object to a single author and a

Ž .single point in time Fig. 1 , does not make senseany more. In fact, the notion of single authorshipdoes in no way record the mutual synergies ofcollaborative creative work that is predominant onthe Internet and leads to very complex legal implica-tions in the economic system.

Ž .Phase x replaces single authorship with collec-Ž .tive authorship Fig. 8 because all relationships

between works, authors and timeline are recorded inthe database and can be rendered and evaluated. In

Ž .Phase x authorship is a variable entity that is rela-tive to its current significance in the system. A singlesubmission may have very little impact on the sys-tem at first, whereas at some later point, its signifi-cance suddenly increases, because of actions of otherauthors that relate to it. As authorship equals identity

Ž .in Phase x , a single author finds himself beingconstantly redefined by other coauthors and subjectto a dynamic model of personal identity.

7. Implications

Ž .Without its architectural content, Phase x is asocial experiment that can be applied to a widevariety of collaborative creative processes and if weare to define cultural evolution as such a process,

Ž .systems like Phase x can be very effective culturalengines.

Applied to for example architectural competitionswith authors from different disciplines, the resultscould be viewed from different angles and precondi-

Fig. 8. Thread of one model through all phases; collective author-ship.

tions and could be further developed at any laterpoint, a concept which is far superior to the currentpractice of selecting a single contributor as the win-

Ž .ner. Phase x is only an early testbed for suchsolution oriented systems and must be followed upby similar experiments that provide different anglesand goals.

8. Recent related work

In the last couple of years, we have refined andextended the teaching methods presented in this pa-per in several other projects. While we cannot de-scribe them in much detail here, we would like torefer to some of these projects briefly.

8.1. Multiplying time

The Virtual Design Studio Multiplying Time wasan international collaboration between several aca-demic Institutions around the globe. Students fromthree timezones, distributed evenly around the globe,worked on a common design problem. They formedan international think-tank, operating 24 h a day.While abstract, geometric models were exchanged,

Ž .in Phase x the VDS projects dealt with architecturalŽdesign tasks. In this sense the VDS of 1997 a

.second VDS took place in 1998 was the first real-Ž . w xworld application of the Phase x principle 3,5 .

8.2. Fake.space

Ž .Fake.space is the sibling course to Phase x ,Ž .taught with Phase x in a two-semester cycle. It does

not take up the idea of memetic engineering, in theŽ .way that it happens in Phase x , but experiments

w xwith collective story-telling in a similar context 4 .

( )8.3. Phase x

Ž .The Phase x course has been repeated once andthe third version is being prepared at the time of thiswriting. While many technical details have been

Žchanged different phases, different modeling soft-.ware used and the handling of the site has become

more elegant, the essentials are still as described inthis paper.

( )U. Hirschberg, F. WenzrAutomation in Construction 9 2000 387–392392

8.4. Information architecture

Ž .To quote the paper, the first Phase x was indeedan ‘‘early testbed’’ for a wide variety of approacheswe are now engaged in, for networked teaching aswell as networked collaboration. It has also been animportant reference for other research projects car-

w xried out at our chair 8 . While we were quite ontarget with respect to the significance of the first

Ž .Phase x course for our own work, we came tounderstand some of its implications only later. Wenow see the design and lay-out of the database andthe website, as well as the various possibilities ofinteracting and visualizing these interactions that themore recent projects provide as case studies in ‘‘in-

Ž .formation architecture’’. Phase x proposes a newform of collective authorship which promises toenable new ways to collaboratively deal with com-plex problems. At the same time, it shows the needfor environments in which such collective authorship

Ž .can grow. Thus Phase x makes a strong point forinformation architecture as one of the real challengesfor architects in the networked world.

Acknowledgements

Ž .Phase x was a team effort not only in the waystudents collaborated. Gerhard Schmitt, Cristina Be-somi, Bige Tuncer, Fabio Gramazio, Urs Hirschberg,

Patrick Sibenaler and Florian Wenz were the mem-bers of the teaching team that conceived, preparedand taught the course. Many other members of thechair for Architecture and CAAD contributed to thecourse. Among these, Daniel Burckhard should bementioned, who was instrumental in developing thescripts for the graphic database representation andthe CD-ROM version of the site.

References

w x1 R. Dawkins, The Selfish Gene, Oxford Univ. Press, Oxford,1976.

w x2 F. Heylighen, The Evolution of Memes on the Network,Memesis, The future of evolution, Ars Electronica ’96,Springer Verlag, WienrNew York, 1996.

w x3 U. Hirschberg, B. Kolarevic, G. Schmitt, D. Kurmann, B.Johnson, D. Donath, The 24 Hour Design Cycle, Proceedingsof the CAADRIA ’99 Conference, Shanghai, forthcoming.

w x4 U. Hirschberg, Fake.space: an online community and a jointenquiry into the nature of space, Proceedings of the CAADRIA’98 conference, Osaka, pp. 281–290.

w x5 B. Kolarevic, G. Schmitt, U. Hirschberg, D. Kurmann, B.Johnson, Virtual Design Studio: Multiplying Time, Proceed-ings of the ECAADE ’98 conference, Paris, pp. 123–130.

w x6 L. Madrazo, Design Education with Computers, Proceedingsof the CAAD Futures ’91 conference, Zurich, 1991, pp.¨77–96.

w x7 G. Schmitt, Principia: Entwurf als Sprache, Architectura EtMachina, Vieweg Verlag, BraunschweigrWiesbaden, 1993.

w x8 G. Schmitt, Information Architecture, Basi e futuro del CAAD,Testo and Immagine s.r.l., Torino, 1998.