Mario Carpo
This essay originated as a
‘paper presented at the confer-
ence “Architecture between
Spectacle and Use” at the Clark
Art Institute in Williamstown,
Massachusetts, in April 2005.
The complete transactions will
be published by the Sterling
and Francine Clark Art
Institute in association with
Yale University Press (2006).
1. “Architecuares non standard” Centre
Pompidou, Paris, December 10, 200}-
March 1, 2008, with an accompanying
publication ofthe same tile, ed. Frederic
Migayrou and Zeynep Meanan (Pais
Edvions du Centre Pompidou, 2005),
wwas followed by "Metamorph" the Sth
International Architecture Biennale in
Yenice, directed by Kure W. Forster,
September 12-November 7, 2004, and
toto conferences dealing with the subject:
*Non-Standard Praxis, Emergent
Principles of Architectural Praxis
swith/in Digital Technologies” convened
ae MIT September 24-26, 2094 (ard
tecture.mit.edu/project/asp), and
“Devices of Design,” organized by the
Canadian Centre for Architeceure and
the Fondation Daniel Langlois for Art,
Science and Technology in Montréal
londation-linglois.org/devicesofde-
sign), cochaved by Mirko Zardi
(ECA) and Jean Gagnon (Fondation
Daniel Langlois), November 18-19, 2004
2. See Mario Carpo's review ofthe exhi
bition, “Archicectures non standard”
Journal ofthe Sciey of rchtectaral
Historians 64/2 (2008): 234-1.
3. See Michael Speaks, “After Theory”
Arhitectural Record (June 1005): 72-75
4. Gilles Deleuze, Difference et répéiton
(Paris: Presses Universitaires de France,
1968); se exp. chap. 4 “Syathive wale
dela difference”: 218-5; Gilles Deleuze,
Le pis Leibniz tle baroque (Pari
Eadsions de Minuit, 188); se esp. chap.
2, Les plis dans ime,” 20-17; his was
‘anilated by Tom Conley as The Fold
Leibnis andthe Baraque Minneapolis
University of Minnesota Press, 1993),
chap. 2, “The Folds inthe Soul,” 14-26.
Tempest ina Teapot
The theory of nonstandard architecture is the latest avatar to
date of the digital revolution in architecture, now already in
its second decade. Several major events that followed
Frédéric Migayrou’s exhibition at the Centre Pompidou in
Paris last year! have resulted in a certain degree of consensus
about what nonstandard architecture might be. However,
there is less consensus on why we might need it. European
critics often characterize nonstandard architectural forms as
a revival or a vindication of 20th-century organicism, but
they fail to explain why organicism is a better idea now than
it was a century ago. More ominously, in the United States
at least, nonstandard technologies are sometimes advocated
as the harbinger of new architectural values that are inher-
ently antitheoretical, anticritical, proudly consumerist,
hedonistic, and market-friendly.’ These number among the
reasons some architects and theoreticians of an older genera-
tion have taken a stance against all that relates to digital
technologies in architecture and design. But the digital need
not be seen as a Trojan horse for the commodification of
design; neither is it true that all of the theoreticians and
designers who have embraced computer-based design and
manufacturing are neocapitalist zealots. In fact, I know some
who are socialists.
In its simplest technical definition, nonstandard seriality
means the mass production of nonidentical parts. This
implies a complete reversal of the mechanical paradigm, in
which mass production generates economies of scale on the
condition that all items in the same series are identical, as in
the traditional assembly line. On the contrary, digital tech-
nologies applied simultaneously to design and manufacturing
can generate the same economies of scale while mass-pro-
ducing a series in which all items are different, albeit differ
ent within limits. On the eve of the digital revolution, Gilles
Deleuze famously anticipated this technological shift in his
studies of difference and repetition, and derived from his
somewhat peculiar interpretation of Leibniz’s differential
calculus the dual notion of objet and objectile,+ which still
aptly defines the basic principle of a nonstandard series: one
objectile in many objets — in a language that Aristotle would
995. towever, when one mabematical lan
frag choven ter pci parame
Se vironment far digil dvgn ou
production it canbe argued tha ome
Sachets enprencions are more ee
rom than others and more clegae
qutons ny beget more clans al
farm. Ths objection wes pitted out 0
ime recenily by Greg Lyanyand would
require lengthier dacusion than
te hee
prom diferent vantage points, tech
blogs and abervrs ofthe prs
visual environmen re coming fo similar
Conlovons. Se Wendy Hui Ryong
Shun, “On Sofware or the Perience
{Wal Knowledge in ry Rav 8
(Weiner 2008) 27
More precy en and coffe service
(Ale offer and Ten Tower) See Greg
Lynn, "arom alee in
‘Arcbitctare nonstandard el Miggyrou
dnd Mennan, 31h orginal projet
inched 50,00 variation of which 99
wrere fabricated sion to three
thors copes according commercial
informacion farised by Ales Sp).
8. See Obj Cearck Beaued and
Bernard Cache), “Yer une architecture
toc in dciere on tended
be
have found more congenial, one form in many events; in the
language of contemporary computer engineering, one algo-
rithm in many digical files.
A nonstandard series is not defined by its relation to the
shape of any individual item in it, whether round or square
or formless. What defines it are the variances or differentials
between all sequential items in the series, regardless of any
specific form, Te also follows that all items in the same non-
standard series constitute a set, where each item has some-
thing in common with all of the others. In technological
terms, all items in a nonstandard series share some algo-
rithms, as well as some of the machines used to process those
algorithms and to produce the objects themselves. In visual
terms, a nonstandard series comprises a theoretically unlim-
ited number of items that can all be different, but must also
be to some extent similar insofar as certain design parame-
ters and technical processes that were used to make them
leave a trace that is detectable in all end products. Similarity
and resemblance, however, are not scientific notions, and
they are notoriously difficult to assess and measure. The
classical tradition, which is based on imitation, hence on
similarity, has been trying to nail down a workable notion of
similarity almost from the beginning of time, a precedent
seemingly ignored by computer and cognitive scientists
today.
Algorithms, software, hardware, and digital manufac
turing tools are the new standards that determine the general
aspect of all items in the same nonstandard series, as well as
which aspects of each item may change, and within which
limits, randomly or by design.* However, unlike a mechani-
cal imprint, which physically stamps things out, an algorith-
mic imprint lets forms change from one item to the next in
the series. Itis thus safe to say that all items in a nonstandard
series share the same style, in the original sense of the term,
which referred to the tool used for writing, not to the inten-
tions of the writer. Today's stilus is not the style of the
designer but the inevitable trace left by the software the de-
signer uses — which, like Cicero's stilus, is man-made and
produced by specialized technicians.
Comparing two nonstandard series recently created by
two of the most alert interpreters of the digital environment
may help to illustrate this point. Greg Lynn's series of 99
teapots for the luxe Italian manufacturer Alessi? and Bernard
Cache’s Projective Tables,* an open-ended series, both fea-
ture items that are visibly different from, yer strikingly simi-
lar to all other items in the same series. The peculiar “stilus”
1009. The opposition between “identical”
and “differential” reproduction was
posed by Greg Lynn, who relates the for-
‘mer to the modes of operating “simple
machines” and the latter to “intricate
‘machines” in the introduction to his
Folding in Arcbiteeture (London: Wiley,
2004), 12, frst published as 4D Magezine
16, no.}~4 Profile 102 (1993).
10. Mario Carpo, “Pattern Recognition,”
in Metamorph, Focus: 9th Imernational
Architecture Exhibition, La Biennale di
Venezia, e. Kurt W. Forster (Venice:
Marsilio, and New York: Rizzoli
International, 2004), 44-58.
of each series derives in part from different technological
platforms employed, but mostly from different adaptations
of existing design software that were implemented by the
two authors, each creating his own range of variabilites and
his own set of self-imposed limits. Lynn favors software
based on differential calculus, whereas Cache developed an
interface based on projective geometry. Each choice can be
justified, but neither was inevitable. In both cases, the figural
result is distinctive: Lynn's differential calculus begets
smoothness and continuity of surfaces; Cache’s projective
geometry begets angular intersections of planes in three
dimensions.
The shift from mechanical to algorithmic production
prefigures a parallel and equally crucial shift in our visual
environment at large. We are leaving behind a world of
forms determined by exactly repeatable visible imprints and
moving toward a new visual environment dominated by
exactly transmissible but invisible algorithms. In the age of
mechanical reproduction, the identification — hence the
meaning ~ of forms depended on their identicality. But in
the new world of algorithmic or differential reproducibility,?
visual identicality is replaced by similarity, which is far less
easy to identify. It follows that our capacity to infer meaning
and value (including market value) from objects made in a
nonstandard environment does not depend on the identifica~
tion of visually identical forms, but rather on the recogni-
tion of hidden patterns ~ a risky and unpredictable cognitive
process.!0
To some extent, the ongoing discussion of the so-called
crisis of indexicality in contemporary architecture may be
seen as.a sign of the impending and much more momentous
demise of identicality, upon which much of our architectural
and visual environment depend. Most mechanically repro-
duced objects and forms are indexical of the imprint that
made them, but most handmade, premechanical, and algo-
rithmically produced items are not. The erratic drifis of
manual copies may distort or confuse the sign of the original
archetype, and as a result conceal the identity of its maker;
the unlimited variances of high-tech, digitally controlled
differential reproduction may have the same effect, up toa
point. Trademarks, coats of arms, imprints, logos, flags, uni-
forms, passports, banknotes, photographs, and the like were
indispensable signs of identity in an age when signs (and
forms) were mechanically printed, and the identicality of
the sign stood for the identity and the authority of its maker.
But identical reproduction is often meaningless and worth-
10111 See the ecent exhibition “Trade
Show" curated by Rebecea Uchil,
Massachusetts Museum of Contemporary
Ary, January 29~May 3, 2005 Cmassmo-
ca.ory/visual_arts/images/tradeshow/T
radeShow Catalog pdf), 16-17.
12. There is also a certain logic tothe fer
thar much of what defines our present
digitally driven visual envionment may
be characterized 35 Be
reverse, or equal and opposite 1 some of
the categories that Benjamin famously
brought to ce forefront with his 936
essay "The Work of Artin che Age of
Mechanical Reproduction,” tans. Harry
Zoho, in Muninatons: Eeays and
Reflection, ed. Hannah Arends (New
York: Schocken Books, 1968), 217-5.
1, Robin Evans, “Translations from
Drawing to Building,” 44 Fle 12 (1986)
3-18, subsequently published in Robin
Evans, Tanslaion from Drawing to
Buslding and Other Entaye (London: The
Architectural Associaton Publications;
Cambridge, Massachusetes: MIT Pees
1999), 194-95
14, With a diferent bia, similar argue
teas are made inthe nedoctin to
Digital Tecnico Neil Leach, Dad
Taenbul and Chris Willams (Landon:
Wiley Academy, 2008), 4-1 seep. #5
'5Or ales it came oe inthe course
athe ith censarySigiied Giedion
Famouly acd he beginnings ofthe
‘modern sembly line tothe andar
zation of mana bicut-makingin
England durig the Napoleonic Wars in
theVicualing Office af the Navy a
Depfor, an in Portsmouth, 104-10),
anda shown how wguencing and
rdiaaion ofthe atnanal gesture
preceded the introduction ofthe st,
Pitcesof machinery. See Mechencation
Taber Command: 4 conrbution any.
snus Bray (1998 rope, New York
Norcon an Co 1969), The for
real equencing of the five rues or
ting 0 pint one
phe inthe evolution modes of bere
frag production, rom arian drinks
lke wine to industrial ones suchas
Cota Coa
16, Se, for example the ange of prod-
ucts and sevice offered by Gehry
Technologies, based in Lov Angeles
gehryechacogiscom), Scores of Is
fSinous companies around the world
have already ily adopted negated
CAD. CAM ethnlogjes fora vary af
bie makes, from prelabricaed bul
ing components, o sone cing, to
ined devices
less in a digitized environment, where recognition depends
on more complex patterns of cognitive perception, and iden-
tification is increasingly taken over by new technologies of
algorithmic encryption. Authors and artists may not be
aware of this incumbent and possibly epochal wave of
change," but corporate lawyers grappling with the right of
copy in the new digital economy have not failed to take
notice. Identical copies were essential and mostly inevitable
in the mechanical world, but they are accidental and mostly
evitable in the digital world — just as identical reproduction
was accidental and mostly avoided in the premechanical
world. There is an ironic symmetry (even a vindication of
sorts) in the fact that some of our contemporaries, baffled by
the apparent randomness and illegibility of differential
reproduction, now attribute to identical reproduction some
of the nostalgic and “auratic” value that Walter Benjamin
famously attributed to premechanical, handmade originals
early in the last century, when art theory first embraced the
logic of identical reproducibility.!?
The rise of CAD-CAM technologies also heralds a new
division of labor in the production of the built environment,
one which reverses the trend the West has been witnessing
since at least the 15th century. Alberti was the first to stipu-
late that architects should stop making things and design
things instead. In the centuries that followed, with a few
notable exceptions, drawings were the inevitable bottleneck
through which all architecture had to pass before it could be
built: drawings had to be translated into buildings ~ once
again, an instance of identical reproduction - and only that
which could be drawn and measured in drawings could be
built. And built by others: as Alberti mandated, that servile
and mechanical operation is not the architect's business.
CAD-CAM technologies have already changed this early-
modern, then modern paradigm. Anything drawn with a
CAD program is already measured in three dimensions, so it
is geometrically buildable; in some cases, CNC milling
machines or similar tools can automatically build these pla-
nar drawings, printing them out, as it were, in the three
dimensions of an X-Y-Z space. The whole technological sys-
tem, from screen to manufacturing, or from file to factory,
emulates the process of traditional hand-making, from con-
ception to formation, including the various phases of feed-
back and interaction between ideas and the qualities of the
matter to be formed. Digital technologies act as an almost
organic, rather than prosthetic, extension of the hand of the
maker. Again, this may point to the forthcoming rise of new
102GREG LYNN, PRODUCTION PROTO:
‘TvPES, ALESSI COFFEE AND TEA
‘Towers, 2001. PHOTO: Gree LYNN
Form, BrLow: TECTON (BERTHOLD
‘LupeTKIN), PENGUIN Poot,
LONDON, 1934
17, For an account of Berthold Lubetkin’s
dealings with penguins, and bis photo-
{graphs of some of the original penguins
Contemplating the model of the pool, sce
Job Allan, Berthold Lubeekin
Architecture and the Tradition of Progress
London: RIBA Publications, 1992), esp.
the image on 198,
forms of digital artisanship, and for most of today’s digital
avant-garde, the separation between design and production
has already been breached."* Both Lynn and Cache are con-
ceivers and makers, designers and artisans, industrialists and
engineering consultants, and yet they are none of these,
because in a digital environment, the traditional division of
mechanical labor no longer applies, and some roles and tasks
that were separated by the mechanical revolution are already
being reunited by the digital one.
However, in spite of our newly acquired technological
abilities, variations for variation’s sake are not likely to stand
the test of time. Much has already been said about the appar-
ent frivolity of a technological revolution that culminates in
a series of 99 slightly different and horrifically expensive
teapots. One might argue that making good tea is a fairly
standard procedure," and that if the function of making tea
does not change, there is no reason why the form of the
teapot should. One response would be that design experi-
ments are typically performed at less than the scale of a
building, and that more crucial objects of manufacture than
a teapot may follow (and some already have, although in a
less glamorous fashion). For the sake of analogy, we may
remember that one of the most celebrated demonstrations of
modernism’s use of reinforced concrete was built in London
in 1934 for the exclusive benefit of a small community of
penguins. Penguins are noble and elegant beings to be sure,
but generally incapable of articulate feedback, and hence as a
group, they are of modest sociological interest.”
Modernism was not about penguins, and nonstandard is
not only about teapots. These nascent nonstandard technolo-
10318, See Stanley M. Davis, Future Perfect
Reading, Massachusers: Addison-
Wesley, 1987), where the expression
“mass customization” fist seems to have
‘curred; and Joseph B. Pine, Mate
Gustomizaton: The New Fromier in
Business Competiin, foreword Stanley M
Davis (Boston: Harvard Business School
ress, 1999). The vem was brought to
theatention af architects and designers
by William J. Mitchell in che late 190s,
See also Daggnar Steffen, *Produire en
masse pour chaciny" drcbiteeure
4 Aujaurd 383 Jay -Avsgust 2004):
102-07
gies may deliver mass production and economies of scale
without the ironclad rules of product standardization that
accompanied the industrial mode of production. Custom-
made products at the same cost as mass-produced, standard-
ized ones would mean better, cheaper, and more suitable
products for many. When the form of objects must follow
human functions, standardized mass production generally
aims at the average user or customer, and neglects the statis-
tical margins. When one size must fit all, users who don't fit
the mold — whether physically or ideologically, by choice or
by necessity — must pay more to have what they need made
to measure. Nonstandard technologies promise to alleviate
this tax on diversity.
Nonstandard production may also appear to multiply
the variety of some commodities beyond necessity, fostering
artificial cycles of demand and consumption. Indeed, the
concept of mass customization began as a marketing strategy
before the advent of CAD-CAM technologies,!* and offering
multiple options for the same product is an old gimmick of
mechanical mass production, but the apparent multiplication
of choice in a mechanical environment (often limited to cos-
metic aspects, such as color) should not be confused with the
scope of seamless nonstandard seriality in a digital environ-
ment. Earlier this year, the latest generation of Objectile’s
Projective Tables was shown in a Paris art gallery, and priced
and sold accordingly. The exhibition included a computer
station, where the customer was invited to set variables for
designing the table of his or her choice; a digital file could
later be emailed to the customer for further verifications at
home (for example, to double-check the measurements)
before the order was sent to the factory, and the table cut
and made to measure, shipped, and delivered, either flat-
packed or already assembled and ready for use. True, uncu-
rious Parisian customers could simply buy one of the tables
on display in the gallery and walk away with it. However,
neither the factory nor the technology is fictional. When the
whole system is put to use as planned, there is no reason for
Objectile’s industrially mass-produced tables (a bookshelf is
also in the works) to be more expensive than Ikea’s. Unlike
Ikea furniture, each of Objectile’s pieces will be made-to-
measure and one of a kind, yet like Ikea’s “Billy” bookshelf,
a standard in millions of homes around the world,
Objectile’s furniture, despite its differences, will be easily
recognized by all who have seen just one piece.
Ac the scale of architectural design, the tectonic implica-
tions of nonstandard technologies are equally vast and easier
104Onjectite, Projective Tastes.
to measure. When form must follow structural constraints,
such as load and stress, standardization inevitably leads to
oversizing and the waste of materials. Regrettably, this is
perfectly rational within the economy of a mechanical envi-
ronment. Take a beam, for example. Often only one section
is subject to the greatest stress, and thus requires the maxi
mum momentum of inertia, or simply the maximum size
and strength, whereas all other sections could be decremen
tally smaller. However, as beams with uniform sections are
cheaper to mass-produce than beams with variable sections,
and the savings in the cost of mass-producing standardized
beams vastly offsets the cost of the material wasted in each
beam, all sections in a beam are often made equal to the one
that is subject to the maximum stress. This kind of wasteful-
ness has characterized civil engineering for most of the sec~
ond half of the 20th century. But the pattern can be reversed
with nonstandard technologies, which make it possible to
envisage structural components that are both mass-produced
and made-to-measure, thereby achieving economies in both
105ted this potential
of nonstandard technologies in his dis-
‘cussion of some structures built in the
1990s by Shoei Yoh, in which eraditional
building materials and craftsmanship
swore merged with advanced digital tech-
iologies. Greg Lynn, “Classicism and
Vitality” im Shoei Yob, ed. Anthony
Tannacci (Milan: L’Arca Edizioni, 1997),
15, See alto Lynn's “Odawara Municipal
Sports Complex” in Shoei Yoh, 67-70; and
“Shoei Yoh, Prefeetura Gymnasium,” in
Folding in Architecture, 79
20. Manuel De Landa, *Material
Complexity," in Digital Tectonics, 20-21.
21, similar isomorphism between
Gothic visualiy and scholastic philoso
phy was famously suggested by Erwin
Panofsky in Gorbic Architerure and
Scholasicim (Latrobe, Pennsylvania
Archabbey Press, 1951).
22. Lewis Mumford, Technic and
Civilization (New York: Harcourt, Brace,
1938); see esp. chap. 8.1, “The Dissolution
of ‘The Machine’ and chap. 8.2,
‘Toward an Organic Ideology,” 368-72,
2}, John Ruskin, The Seven Lamp of
Architecture (London: Smith, Elder and
Co,, 1849); se esp. chap. 5, “The Larmp
of Life” XXI: 14; see also John Ruskin,
The Stones of Venice CLondan: Smith,
Elder and Co,, 1851-58), §: (1851), ILIV.
5,198,
Manio Canpo 1s associaTE PROFES-
SOR IN ARCHITECTURAL HISTORY AND
‘THEORY AT THE ScHooL OF
ARCHITECTURE, PaRIS-La VILLETTE,
AND CONSULTING HEAD OF THE
Srupy CenTRe AT THE CANADIAN
(CENTRE FOR ARCHITECTURE,
MonrREAt.
the processes of production and the use of materials.” Digital
tools will also facilitate the use of materials characterized by
nonlinear elasticity, such as plastics or ceramics, and even
timber or stone, which, being natural and not standardized,
include irregularities that make their structural behavior
unpredictable.2 Nonstandard technologies could interact
with such irregularities and adapt form and design to the
variability of nature almost as easily as artisanal manipula
tion always did. In short, structural design could again
become an art form, as it was up until a century ago, when
building material was rare and human intelligence abundant:
for most of the last century, the reverse has been the case.
Moreover, we are now aware, as the preindustrial world
always was but the industrial world often wasn’, that most
building materials are nonrenewable resources, and that
their supply is limited. Ingenuity in structural design is no
longer just a matter of pride on the part of the maker: it is a
matter of social responsibility for all users.
In sum, it appears that the rise of the nonstandard para-
digm, based on digital technologies, implies the complete
reversal of some technical, economic, social, and visual prin-
ciples that have characterized the mechanical age for almost
five centuries. It is a commonplace of media studies that the
new digital environment is in many ways closer to the world
of script that preceded print than the age of printing, which
digital technologies are replacing. Digital technologies
applied to design and manufacturing may prompt similar
interpretations. Although projected into the future, this also
points to a pre-Albertian technical and visual environment, a
new scholasticism of sorts, based on algorithmically defined
fixed genera and endlessly morphing species.” This is a
world that one of the greatest moralists of the 20th century,
Lewis Mumford, prophesied as forthcoming: a neotechnical
age that would stamp out the paleotechnical, mechanical age
that he decried as the source of all evil. This is also a world
that the greatest moralist of the 19th century, John Ruskin,
situated in the distant past, in the golden age of Gothic arti-
sanship, before the slavish art of the Renaissance turned
artisans into machines (Marx might have called them prole-
tarians) and European architecture lost all light and all life,
as Ruskin thought, forever. But the forthcoming neotechni-
cal, nonstandard world cannot be described in Benjaminian
terms, and many architects and critics trained in the 20th
century would find it meaningless. Mies van der Rohe
probably would not have understood it, and Rem Koolhaas
probably won't.
106