`
`SCMNTIFIC
`AMERICAI{September 1984
`Computer Software
`
`Volume 251
`
`Number 3
`
`Presenting a single-topic issue on the concepts and techniques needed
`to make the computer do one's bidding. It is software that gives form
`and purpose to a programmable machine, much as a sculptor shapes clay
`
`by Alan Kay
`
`omputers are to computing as in-
`struments are to music. Software
`is the score, whose interpretation
`amplifies our reach and lifts our spir-
`it. Leonardo da Vinci called music "the
`shaping of the invisible," and his phrase
`is even more apt as a description of soft-
`ware. As in the case of music, the in-
`visibility of software is no more mys-
`terious than where your lap goes when
`you stand up. The true mystery to be ex-
`plored in this issue of Scientific American
`is how so much can be accomplished
`with the simplest of materials, given the
`right architecture.
`The materials of computing are the
`tersest of markings, stored by the bil-
`lions in computer hardware. In a mu-
`sical score the tune is represented in
`the hardware of paper and ink; in biol-
`ogy the message transmitted from gen-
`eration to generation by DNA is held
`in the arrangement of the chemical
`groups called nucleotides. Just as there
`have been many materials (from clay to
`papyrus to vellum to paper and ink) for
`storing the marks of writing, so comput-
`er hardware has relied on various physi
`cal systems for storing its marks: rotat-
`ing shafts, holes in cards, magnetic flux,
`vacuum tubes, transistors and integrat-
`
`ed circuits inscribed on silicon chips.
`Marks on clay or paper, in DNA and in
`computer memories are equally power-
`ful in their ability to represent, but the
`only intrinsic meaning of a mark is that
`it is there. "Information," Gregory Bate-
`son noted, "is any difference that makes
`a difference." The first difference is the
`mark; the second one alludes to the need
`for interpretation.
`The same notation that specifies ele-
`vator music specifies the organ fugues
`of Bach. In a computer the same no-
`tation can specify actuarial tables or
`bring a new world to life. The fact that
`the notation for graffiti and for sonnets
`can be the same is not new. That this
`holds also for computers removes much
`of the new technology's mystery and
`puts thinking about it on firmer ground.
`As with most media from which
`things are built, whether the thing is a
`cathedral, a bacterium, a sonnet, a fugue
`or a word processor, architecture domi-
`nates material. To understand clay is not
`to understand the pot. What a pot is all
`about can be appreciated better by un-
`derstanding the creators and users of the
`pot and their need both to inform the
`material with meaning and to extract
`meaning from the form.
`
`MESSAGE embedded in a material medium is the essence of computer soft-
`INTANGIBLE
`image: a scanning-electron micro-
`ware. Ilere the message is made visible in a voltage-contrast
`graph of a small part of an Intel 80186 microprocessor, The features of the image are formeil
`not by the coniluctors and transistors on the chip but by the signals passing through them. The
`trajectory of the secondary electrons emitteil in response to the microscope beam is afiected by
`fiekls at the surface of the chip: regions of higher voltage attract electrons,
`electromagnetic
`weakening the image-forming signal. The microscope beam is pulsed on only when the micro-
`processor is in a particular electronic state: when certain logic elements are "on." The colors of
`lines leading to logic elements. Where
`the lines indicate the voltages in metal communications
`a signal is traveling along a line there is a region of high voltage. The false-color image has been
`processed so that such regions, and thus "messagesr" are seen in light blue. Low-voltage regions
`regions yellow The red lines are conductors at ground poten-
`are green; intermediate-voltage
`tial, or zero volts. The micrograph was made by Timothy C. May of the Intel Corporation.
`
`There is a qualitative difference be-
`tween the computer as a medium of ex-
`pression and clay or paper. Like the ge-
`netic apparatus of a living cell, the com-
`puter can read, write and follow its own
`markings to levels of self-interpretation
`whose intellectual limits are still not un-
`derstood. Hence the task for someone
`who wants to understand software is not
`simply to see the pot instead of the clay.
`It is to see in pots thrown by beginners
`(for all are beginners in the fledgling
`profession of computer science) the pos-
`sibility of the Chinese porcelain and Li-
`moges to come.
`I need spend no more time on
`IJere
`I -f computing's melhods for storing
`and reading marks than molecular biol-
`ogy does on the general properties of
`atoms. A large enough storage capacity
`for marks and the simplest set of in-
`structions are enough to build any fur-
`ther r'epresentational mechanisms that
`are needed, including even the simu-
`lation of an entire new computer. Au-
`gusta Ada, Countess of Lovelace, the
`flrst computer-software genius, who
`programmed the analytical engine that
`Charles Babbage had designed, under-
`stood well the powers of simulation
`of the general-purpose machine. In the
`1930's Alan M. Turing stated the case
`more crisply by showing how a remark-
`ably simple mechanism can simulate all
`mechanisms.
`The idea that any computer can simu-
`late any existing or future computer is
`important philosophically, but it is not
`the answer to all compufational prob-
`lems. Too often a simple computer pre-
`tending to be a fancy one gets stuck in
`the "Turing tar pit" and is of no use if
`results are needed in less than a million
`years. In.other words, quantitative im-
`provements may also be helpful. An in-
`53
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 1
`
`
`
`s
`
`I
`
`I
`l
`
`BseJt
`
`,,
`ll
`
`e lueserde; uezre feur peeds ur es!eJt
`.ld.oq raprsuoC'lueure,rordtur elr1e11pnb
`seru!q o.4Al ruory urlg B dn Surpeeds
`-8eur;o repJo ereu e) 0Z ol puoces Jed
`:ecueJeJlp elqE{reuer B solBr,u (epnlru
`uorldecred a.rrlcefqns eril ol sp!el ll
`er{l Jo qcnw 'luerue^our snonurluoc Jo
`uorlJ!Jelu ,,{.ro1rpnu pu! IEnsrA Jo ,,aJIL,
`'eced s1r uo spuedep
`
`^!lr leql pera^ooslp e,n uerppqc s[
`,{ldtuts urroy {ue otrur pedeqs eQ uec 1 7
`'Jnts eql olut spu!q qloq Eunoqs '(q
`8uq1 qcns ou peuJ!el al!q sn Jo lsotrAl
`stueas l!rJol!ru s11 '.telnduroc eql lnoq!
`e su oruerradxe ueunq tuoJJ peqJ!lep se
`-e; palelndrueru Sureq lo8ur e,rrlceorper
`-rlalel ! pue s8uo1 'suo11ng qlral fleloru
`I!uorloue Jo pul{ l!q1t|rolruoru uols
`Jnls ^\eu sql rllr1lt e{"u ouo u!c lJsluoc
`eolotuer os sruees ssecce 1ecrs.{qd eqt JI
`Sqlndwoc 3o ,{e1c eql sleoJ auo
`-lJos oql :.(ec?JreJur resn,, eq1 qSnorql
`uosred e uae,rleq salurpeu leql eJelt\
`relnduroc eql 8urdeqs sue:8o.rd eql pue
`raqleq^r 'po8 cgrceds e roJ lool E olur
`Eurlr.rm to eEpr:g e EuruErsep sr 1eo8 aq1
`ecuo se,r a3!JJetrw Jesn eqJ .elcrlJ! u!
`'peuErsep eq o1 ruelsfs e ;o ped 1s!l er1
`'1srg eqtr sI ll rAoN
`se pezruSocer sl lJ
`pu! secrlou ol 'esn"ceq freurrrd Eureq
`peluese:d sr l!q/( 'e{rl! sleuorsse;ord
`
`eqa 'relndruoc sieuo ,r1 sesues s.euo otr
`pue senSeelloc ftu s! ,..uorsnllr resn,,
`-eu otlv oIEd xorox eql tE lf pellec l
`rpfur pegfldurs eql sr ,JelueC qcJ!es
`eleur pue) uqeldxe ol splnq euo,{tele
`pu! suollcB s.urels{s oql (lnoq! sessen8
`'lxeu euop eq pForls l!q \
`secrlep pue seldrcwrd eq1 ;o ,{ue1q
`e !q uorsnllr eq? ecuequo o1 pedole,rep
`eJend,lJos ur ecelduoturuoc eruoceq aou
`-urrdluelrodur lsoru eq1 sdeqreg .u8rsop
`1!r{ \ sr ees nof 1eqryy,,) c,r,rrrs,r,tr sr a1d1c
`eqtr uo oEeurr eql :(..1eE no,
`sr
`'ueoJcs
`erp Jo uorl"trueso;de.r InJqll!J e s,{e,tr1e
`'uorsn[r s(resn
`eSeurr eql Sqtepdpeyq
`,{eat urelrec e ur
`-euos seop ,(1e1e1peurun
`-!ru eql Jo atr!ts aql 01 elqelclpe:d 8u1q1
`luql seurSeurr resn oql se) eurqc
`'(olels
`-urzr\,, s!q entol ur lr\ou uorsnlp auo
`Euglulod ! pu! .(suocr,, ..'snueur,, ..,slrrop
`-ur^A. peIIBc seru!4 ,(eldsrp eqJ .ecr^ep
`e lqeserd o1 elqrssod 1l e{Btu s^top
`l! ueeJJs eql uo sarlr^rlJe Jo Jeqr.unu
`sdels lxau eyqrssod Jo snuery .arurl euo
`slcefqo lueserde; suocr lpe,{eldsrp ere
`ecrlap 8urlurod y 'seEeru eleJcuoc s!
`peqsnd sr (esnour ! pall!o seurrleruos)
`ueeJcs eql uo telutod ! elour ol lnoq!
`'snaopuua telncrl.ted 1ce1es {qereql pue
`'suocl Jo swell nuetu
`'
`-ue8.treu ! ol esrr ua,rr8 seq slql Jo IIV
`-rdm ?!ril aJ!./rUJos elrlceJelur Jo uolleJa
`
`I
`096t
`
`9961
`8/61
`oql q8no.rql elaq uaoqs s! sB .sp^rolut rlpurods 1u laqloue auo poorrns SEUNE9 IU1;1AL1OS
`-,reJ,tq,{1;-ru4;qrE roqlBr pezFo8eluc a.rt se8un8uul .sa8untueJ Bqunuu.r8o.rd auros;o alduruxa
`.du1re,ro (spuoq paropr) qa^al eql q8noqle 1j
`-q3;q '(rrr) so8un8uq leaol-rrol eru eraql
`'(mHo) sa8en81e1 q,re1-q3;qu-41n puu (rrn,r) sa8un8uul le,rapq8rq-,(rea .(rtr) se8en8uq 1e^al
`ueql t(sauq atlqil lotttozr.tot!) poqEtqslse q a.rua8 e saien8uel Sugunuulsord Jo uo;;n;o^a eq1 u1
`a8un8uul poao.rdurl eq| ouql rrl'(saur arqtr pa^rnt) epBur E luorueaoldurl uu s.rea,( ilal s ralJc
`lx!u erg ol spueJ lJ puu,.(3urq1 rleu tsorutu' uu lnq ,,3u;q1 plo rageq, u,{;araur lou eq ol uees E
`'a.rue8 ,rau u Bqurocaq atup qree .,{1pelseder pa8uuqr suq ds;1 e8un8uul eq; o-rua8 elqup
`
`elllcefqo eqa 'uorsnlp Jesn egl uo sezrl!l
`-nturs ol fr1gqe s(Jasn er{l {Srldure o1 sr
`-el ls31eeJ3 eql slJexe uosJed v
`.elel
`-dpeur eq u!c uorsnlp srq uer.I,{\ eEBJaA
`-relw loer1sqe o1 pedde lnoqtur pel!ln
`surerSo.rd uepplq eql sB r{cns serrerpeu
`-urls ! uele uorlc! o1u1 1nd ol pepeeu
`lcenp II"o I lerll6. 'rossecord prou e1d
`uorsnlF ern uoq/K pepl,rotd sr eSererrel
`01 r{cll1f\ rllr.a 'loo1 ro ..,1I){,, ! s! slc!
`F.,vr e6era,re1 lcerpq
`! e^los
`.urelqord
`sB slcB uolsnll eql uer{,/(\ peur!llB eq
`s6euo Jo uorsualxe alllc! u! :..1ueEe,, ue
`'spo8 pue esodtnd
`eql ses"c qloq q
`sr teq,tl. Jo tro4uoc s.reutrsap er!alJos
`Ae{ eql sr lxeluoc Fcr4?eql e {perluesse
`s11 Eulcueque pu! uolsnm ue Surleerc o1
`..'sSeulLpuel{,, perrrec;ed
`
`erarrr sruertord lelndtuoc lserlree eq_1.
`-Ics pue suercrlerueqleu {q peuSrsep a
`eq pFor{s {s!} eql lqSnoqt oqnr\ slsrtue
`ereitruos'1ecr8o1 pfre pre,rroglqtl!rls
`ueqt edeqs ol rapreg eq o] tno peurnl
`eterrr srelndruo3 'pesoddns peq l{eq1
`1eq.tr Surop uo pelsrsul ,{eq; .uroqqnls
`-urerSord oql l!q \ uer{t reql!J pr?s s"a
`JO sS"lC naeu B trInseJ ! sv .lu!eu Jelu
`lsel eseql 'Isel er{l JeAo {oo1 su!srlJ?
`uelJo erea. eueldrq ,{reurq eq1;o slopd
`-lcs,{le uole Jou I!crleruoqletu Jer{llou
`pete8ua f1daep eram,'{eqt 1nq 'cgrluj
`-oJ E-IArJeleu eql qlr^\ ecusruoJ B ut
`;o rosrncerd eql uo4o sl leql ecueru
`-lcs IeJntrBN
`'etrle secuercs prrE sUE ,,tA,eu
`{ees pu! esJelrun e ue.Lr8 eJE slsrlue
`-uercs relndruoC 'slr\"I slr Jelocslp ol
`sruerSord Jo ruroJ erll ur sael e{eru slsll
`esJe^run .rrr.eu e sSuuq telnduroc oql pue
`.EJII 01
`ool poqleetq s;erurue;Eord eruos
`-erc Jo ereqdsourle ^p!oq eqt 1o ,,{1deep
`eluBceg ,{eqa 'es:e,rrun elerrrrd e Eurle
`lreqo6 reuSrsep truewrue eql leq,r
`'S
`./$,oI ! Jo slserrd qEH eql,, pell!c uog!g
`trsol l .r.1lnc
`ll lBql
`'re^a.&oq 'pero^ocsrp
`esJelrun e;o poE eql oq ol 8urq1 auo sr
`lo4uoc 01 elq! eq ol raqlou! pu!
`pu! ']l
`u8rsep roJ pleg Jreql eprstno pe4oo1 {eq1
`'uorlerrdsur pu! seepr
`. sSur-at s! elres uec e.rueE p;.re,rod y
`-!tre1u snoJeqc!eI lsolu er.IJ .sur!qc Jo
`JOJ IJO./* 01 ruees 1"q1 seuo eqtr e:e s;oqd
`a;our dee4 uec ,{aql esneJeq ,eunl d
`sy 'dn Euqqgnq uro.r; slqErsur In;relr.od
`sr eJoql lng-,llols sr sser8ord llnseJ B
`'per{srlg!lse si erueS ndeu v
`v
`.sserEord
`-e,rordu1 luecgru8ls " Jelsl sreer{ ,rey
`srebi( e.lour a,eJ ! JoIJV .ep!u sl lueru
`Sureq se pa,rrecrad sr lueue,ro;du4 eql
`u! lnq ..8un11 p1o relteq,, e lsnf 1ou
`fltce.np speel l!rl1 ..Eurql ,treu lsorul!,,
`'erue8 e1qe1s lxeu eql ol
`'{18ug1sere1u1
`slueruelordw rreql pu! sEurql pyo eq1
`lou op
`se^rl!lueserdeJ EuoJls 'J"eddBsrp
`qcns '^epol oAIJIII ?Je lsed qc!e ruo{
`-uEI plo-r!a^-Og eql ur Eurruru!JgoJd s!
`ur uo^e puB NYUrdOd s! un[ou{ egBnE
`-!ru lceJrp sB u1Kou)l lducs luetcu! eql
`{ool lqtrlu eldoed eluos .epoc eurqc
`sJeqlo :slssoJ EurArI s! scrleJ rlcns uo
`plo /fua^ e ue^e lEr{l lno lurod plnorfi
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 2
`
`
`
`species might still be filling a particular
`ecological niche.
`f he computer field has not yet had its
`r Galileo or Newton. Bach or Beetho-
`ven, Shakespeare or Molidre. What it
`needs first is a William of Occam, who
`
`said "Entities should not be multiplied
`unnecessarily." The idea that it is worth-
`while to put considerable effort into
`eliminating complexity and establishing
`the simple had a lot to do with the rise of
`modern science and mathematics, par-
`ticularly from the standpoint of creating
`
`new aesthetics, a vital ingredient of any
`growing field. It is an aesthetic along
`the lines of Occam's razor that is need-
`ed both to judge current computer soft-
`ware and to inspire future designs. Just
`how many concepts are there really?
`And how can metaphor, the magical
`
`IS JUST LIKE
`
`EXCEPT: NEWMESSAGES
`DISPLAY
`GENEALOGY
`YOUR MESSAGES?
`
`EXCEPT: NEW MESSAGES
`DISPLAY ->
`ORIGIN
`CORNER
`CENTER
`BORDER
`FILL
`
`EXCEPT: NEW MESSAGES
`DISPLAY --__->
`TITLE
`SCROLL BARS
`MOVE
`EXPAND
`OPEN
`CLOSE
`MOUSE BUTTON?
`
`EXCEPT: NEW MESSAGES
`DISPLAY -_->
`PANES
`RELATIONS
`SOURCE
`
`U U
`
`<-t Ll
`
`U U
`
`\y
`
`<-t Ll
`EXCEPT: NEWMESSAGESNEEDED
`TO SPECIALIZE EACH ONE
`
`t->
`
`t l t
`
`l
`
`I
`r
`
`V
`
`DATA-BASE
`PLANNING
`BROWSER
`BROWSER
`INHERITANCE
`shows the power of differen-
`PROGRAMMING
`tial description. A generic "objectrt (/op) is displayed as a cloud, One
`can make a rectangle from the undifferentiated object by saying, in
`efiect "I wantsomething
`just like that, except. .. ,'and then specifying
`such properties as the location of the origin (the upper left corner),
`the width, the height and so on. A further elaboration of the idea is a
`"windowrtt a rectangular area of the ilisplay screen that gives a view of
`the output of a program, In creating a window one can allow it to ..in-
`
`THINGLAB
`BOOK-READER
`BBOWSER
`BROWSER
`herit" applicable properties of the rectangle and add new features
`such as scroll bars (to move the winilow about over the material being
`viewed), a title anil facilities for changing the winilow's size anil posi-
`tion. A more complex window with panes is made by adding new
`display methoils to shape the panes and establish communications
`among them (colored arrows'). Paned windows can be manipulated
`to
`make "browserstt: systems enabling one to retrieve resources without
`remembering names. Four examples of browsers are shown (bottoml.
`55
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 3
`
`
`
`process of finding similarity and even
`identity in diverse structures, be put to
`work to reduce complexity?
`The French mathematician Jacques S.
`Hadamard found, in a studv of 100 lead-
`ing mathematicians, that the maioritv of
`them claimed to make no use of ivmbols
`in ,their thinking but were instead pri-
`marily visual in their approach. Some,
`including Einstein, reached further back
`into their childhood to depend on .,sen-
`sations of a kinesthetic or muscular
`type." The older parts of the brain know
`what to say; the newer parts know how
`to say it. The world of the symbolic can
`be dealt with effectively only when the
`repetitious aggregation of concrete in-
`stances becomes boring enough to moti-
`vate exchanging them for a single ab-
`stract insight.
`In algebra the concept of the varia-
`ble, which allows an infinity of instan-
`ces to be. represented and dealt with as
`one idea, was a staggering advance. Met-
`aphor in language usually accentuates
`the similarities of quite different things
`as though they were alike. It was a tri-
`umph of mathematical thinking to real-
`ize that various kinds of self-compari-
`
`son could be even more powerful. The
`differential calculus of Newton and
`I.eibniz represents oomplex ideas by
`finding ways to say "This part of the idea
`is like that part, except for... ." The
`designers of computing systems have
`learned to do the same thing with dif-
`ferential models, for example with pro-
`gramming methods that have the prop-
`erty called inheritance. In recent vears
`models based on the idea of recuision
`have been formulated in which some of
`the parts actually are the whole: a de-
`scription of the entire model is needed to
`generate the representation of a part. An
`example is the fractal geometry of Be-
`noit B. Mandelbrot, where each subpart
`of a structure is similar to every other
`part. Chaos is captured in law.
`Designing the parts to have the same
`power as the whole is a fundamental
`technique in contemporary software.
`One of the most effective applications of
`the technique is object-oriented design.
`The computer is divided (conceptually,
`by capitalizing on its powers of simula-
`tion) into a number of smaller comput-
`ers, or objects, each of which can be siv-
`en a role like thal of an actor in a pl-ay.
`
`The move to object-oriented design rep-
`resents a real change in point of view-a
`change of paradigm-that brings with
`it an enormous increase in expressive
`power. There was a similar change when
`molecular chains floating randomly in
`a prebiological ocean had their efficien-
`cy, robustness and energetic possibilities
`boosted a billionfold when thev were
`first enclosed within a cell membrane.
`The early applications of software
`objects were attempted in the context
`of the old metaphor of sequential pro-
`gramming languales, and the objects
`functioned like colonies of cooperat-
`ing unicellular organisms. If cells are a
`good idea, however, they really start to
`make things happen when the coopera-
`tion is close enorigh for the cells to ag-
`gregate into supercells: tissues and or-
`gans. Can the endlessly malleable fab-
`ric of computer sfuff be designed to
`form a "superobject"?
`f he dynamic spr-eadsheet
`is a good ex-
`r ample of such a tissuelike superob-
`ject. It is a simulation kit, and it provides
`a remarkable degree of direct leverage.
`Spreadsheets at their best combine the
`
`SHEET
`
`VALUE RULE
`VALUE
`f RULE
`IMAGE
`
`WINDOW
`
`;,
`It
`
`,
`
`,
`
`.i rll:
`
`DYNAMIC
`SPREADSIIEET
`is a simutation kit: an aggregate of
`software objects called cells that can get yalues from one another,
`The window selects a rectangular part of the sheet for display. Each
`cell can be imagined as having several layers behind the sheet that
`compule the cell's value and determine the format of the presenta-
`56
`
`tion. The cellts name can be typed into an adjoining cell. Each cell has
`a value rule, which can be the value itself or a way to compute it; the
`value can also be conditional on the state of cells in other parts of the
`sheet The format rule converts the value into.a
`form suitable for
`display, The image is the formatted value as rlisplayed in the sheet
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 4
`
`
`
`geffes established.in the 1970's (objects,
`windows, what-you-see-is-what-you-get
`editing and goal-seeking retrieval) into a
`"better old thing" that is likely to be one
`of the "almost new things" for the main-
`stream designs of the next few years.
`A spreadsheet is an aggregate of con-
`currently active objects, usually orga-
`nized into a rectangular array of cells
`similar to the paper spreadsheet used
`by an accountant. Each cell has a "val-
`ue rule" specifying how its value is to
`be determined. Every time a value is
`changed anywhere in the spreadsheet,
`all values dependent on it are recom-
`puted instantly and the new values are
`displayed. A spreadsheet is a simulated
`pocket universe that continuously main-
`tains its fabric; it is a kit for a surprising
`range of applications. Here the user il-
`lusion is simple, direct and powerful.
`There.are few mystifying surprises be-
`cause the'only way a cell can get a val-
`ue is by having'the cell's own value rule
`put it there.
`Dynamic spreadsheets were invented
`by Daniel Bricklin and Robert Frank-
`ston as a reaction to the frustration
`Bricklin felt when he had to work with
`the old ruled-paper versions in busi-
`ness school. They were surprised by
`the success of the idea and by the fact
`that most people who bought the first
`spreadsheet program (VisiCalc) exploit-
`ed it to forecast the future rather than
`to account for the past. Seeking to de-
`velop a "smart editor," they had creat-
`ed a simulation tool.
`Getting a spreadsheet to do one's bid-
`ding is simplicity itself. The visual meta-
`phor amplifies one's recognition of situ-
`ations and strategies. The easy transition
`from the visual metaphor to the sym-
`bolic value rule brings the full power
`of abstract models to bear almost with-
`out notice. One powerful property is
`the ability to makea solution generic by
`"painting" a rule in many dozens of cells
`at once without requiring users to gener-
`alize from their original concrete level
`of thinking.
`The simplest kind of value rule makes
`a cell a static object such as a number
`or a piece of text. A more complex rule
`might be an arithmetic combination of
`other cells' values. derived from their
`relative oi absolute positions or (much
`better) from names assigned to them. A
`value rule can test a condition and set its
`own value according to the result. Ad-
`vanced versions allow a cell's value to
`be retrieved by heuristic goal seeking,
`so that problems for which there is no
`straightforward method of solution can
`still be solved by a search process.
`f he strongest test of any system is not
`I how well its features conform to an-
`ticipated needs but how well it performs
`when one wants to do something the de-
`signer did not foresee. It is a question
`less of possibility than of perspicuity:
`
`Can the user see what is to be done and
`simply go do it?
`Suppose one wants to display data as
`a set of vertical bars whose height is nor-
`malized to that of the largest value, and
`suppose such a bar-chart feature was
`not programmed into the system. It calls
`for a messy program even in a high-lev-
`el programming language; in a spread-
`sheet it is easy. Cells serve as the "pix-
`els" (picture elements) of the display;
`a stack of cells constitutes a bar. In a
`bar displaying one-third of the maxi-
`mum value. cells in the lowest third of
`the stack are black and cells in the up-
`per two-thirds are white. Each cell has
`to decide whether it should be black or
`white according to its position in the
`bar: "I'll show black if where I am in
`the bar is less than the data I am trying
`to display; otherwise I'll show white"
`lsee illustration on next pagel.
`Another spreadsheet example is a so-
`phisticated interactive "browser," a sys-
`tem originally designed by Lawrence G.
`Tesler, then at the Xerox Palo Alto Re-
`search Center. Browsing is a pleasant
`way to access a hierarchically organized
`data base by pointing to successive lists.
`The name of the data base is typed into
`the first pane of the display, causing the
`subject areas constituting its immediato
`branches to be retrieved and displayed
`in the cells below the name. One of the
`subject areas can be chosen by pointing
`to it with a mouse: the chosen area is
`thereby entered at the head of the next
`column, causing its branches in turn
`to be retrieved. So it goes until the de-
`sired information is reached lsee illus-
`tration on page 591. Remarkably, the en-
`tire browser can be programmed in the
`spreadsheet with just three rules.
`The intent of these examples is nbt to
`get everyone to drop all programming in
`favor of spreadsheets. Current spread-
`sheets are not up to it; nor, perhaps, is
`the spreadsheet metaphor itself. If pro-
`gramming means writing step-by-step
`recipes as has been done for the past 40
`years, however, then for most people it
`never was relevant and is surely obso-
`lete. Spreadsheets, and particularly ex-
`tensions to them of the kind I have sug-
`gested, give strong hints that much more
`powerful styles are in the offing for nov-
`ices and experts alike. Does this mean
`that what might be called a driver-edu-
`cation approach to computer literacy is
`all most people will ever need-that one
`need only learn how to "drive" applica-
`tions programs and need never learn to
`program? Certainly not. lJsers must be
`able to tailor a system to their wants.
`Anything less would be as absurd as re-
`quiring essays to be formed out of para-
`graphs that have already been written.
`In discussing this most protean of me-
`dia I have tried to show how effective-
`ly design confers leverage, particularly
`when the medium is to be shaped as
`a tool for direct leverage. It is clear
`
`that in shaping software kits the limita-
`tions on design are those of the creator
`and the usei, not those of the medium.
`The question of software's limitations
`is brought front and center, however,
`by my contention that in the future a
`stronger kind of indirect leverage will
`be provided by personal agents: exten-
`sions of the user's will and purposes,
`shaped from and embedded in the stuff
`of the computer. Can material give rise
`to mentality? Certainly there seems to
`be nothing mindlike in a mark. How
`can any combination of marks, even dy-
`namic and reflexive marks, possibly
`show any properties of. mentality?
`fitoms also seem quite innocent. Yet
`.{ r biology demonstrates that simple
`materials can be formed into exceeding-
`ly complex organizations that can inter-
`pret themselves and change themselves
`dynamically. Some of them even appear
`to think! It is therefore hard to deny cer-
`tain mental possibilities to computer
`material, since software's strong suit is
`similarly the kinetic structuring of sim-
`ple components. Computers "can only
`do what they are programmed to don"
`but the same is true of a Tertilized egg
`trying to become a baby. Still, the diffi-
`culty of discovering an architecture that
`generates mentality cannot be overstat-
`ed. The study of biology had been under
`way some hundreds of years before the
`properties of DNA and the mechanisms
`of its expression were elucidated, reveal-
`ing the living cell to be an architecture in
`process. Moreover, molecular biology
`has the advantage of studying a system
`already put together and working; for
`the composer of software the computer
`is like a bottle of atoms waiting to be
`shaped by an architecture he must in-
`vent and then impress from the outside.
`To pursue the biological analogy,
`evolutiori can tell the genes very little
`about the world and the genes can tell
`the developing brain still less. All levels
`of mental competence are found in the
`more than one and ahalf million surviv-
`ing species. The range is from behavior
`so totally hard-wired that learning is nei-
`ther needed nor possible; to templates
`that are elaborated by experience, to a
`spectrum of capabilities so fluid that
`they require a stable social organiza-
`full adult potential
`tion-a culture-if
`is to be realized. (In other words, the
`gene's way to get a cat to catch mice is
`to program the cat to play-and let the
`mice teach the rest.) Workers in artifi-
`cial intelligence have generally content-
`ed themselves with attempting to mimic
`only the first, hard-wired kind of behav-
`ior. The results are often called expert
`systems, but in a sense they are the de-
`signer jeans of computer science. It is
`not that their inventors are being dis-
`honest; few of them claim for a system
`more than it can do. Yet the label "ex-
`pert" calls up'a vision that leads to dis-
`57
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 5
`
`
`
`illusionment when it turns out the svs-
`tems miss much of what expert (or even
`competent) behavior is and how it gets
`that way.
`Three developments have very low
`probabilities for the near future. The
`
`first is that a human adult mentality can
`be constructed. The second is that the
`mentality of a human infant can be
`constructed and then "brought up" in
`an environment capable of turning it
`into an adult mentalitv. The third is
`
`rule for each cell rs
`ffiValre
`"Show black if (11 - vertical lo-
`cation) x pixel height is less
`than data [horizontal location]
`else show white."
`
`ffi{til1lvalue rule for each cell is
`either the number itself or a number
`fetched from some other part of
`the sheet.
`
`mWffimmMaximum datum + 10.
`
`.BAR CHART can be constructed out of the standard materials of a spreadsheet A bar is a col-
`umn of cells, where each cell serves as a pixel, or picture element One cell associated with each
`column holds the datum, or value, to be represented by the height of the corresponding bar,
`Within a bar all the cells are governed by the same rule. The quanfity representeil by the heigh{
`of a single pixel is the maximum
`ilatum diviiled by the number of pixels in the longet bar; in
`chart c there are 10 pixels per bar and each pixel represents 25 units. Each cell shows black if
`itb vertical position in the bar muttiplied by the number of units per pixel is less than the dafum
`for that bar; otherwise it shows white. lYhen a new datum is entered in a column (D), a new bar
`appears in that column (c). If a new datum is larget than the previous maximum
`(d), the set of
`bars is replotted (e) on the basis of the new number of units per pixel, which in this case is 36.7.
`58
`
`that current artificial-intelligence tech-
`niques contain the seeds of an architec-
`ture from which one might construct
`some kind of mentality that is genuine-
`ly able to learn competence. The fact
`that the probabilities are low emphati-
`cally does not mean the task is impossi-
`ble. The third development is likely to
`be achieved first. Even before it is there
`will be systems that look and act some-
`- what intelligent, and some of them will
`actually be useful.
`W/hat will agents be like in the next
`W
`few years? The idea of an agent
`originated with John McCarthy in the
`mid-1950's, and the term was coined
`by Oliver G. Selfrldge a few years later,
`when they were both at the Massachu-
`setts Institute of Technology. They had
`in view a system that, when given a goal,
`could carry out the details of the appro-
`priate computer operations and could
`ask for and receivp advice, offered in
`human terms, when it was stuck. An
`agent would be a "soft robot" living
`and doing its business within the com-
`puter's world.
`What might such an agent do? Hun-
`dreds of data-retrieval svstems are now
`made available through computer net-
`works. Knowing every system's arcane
`access procedures is almost impossible.
`Once access has been gained, browsing
`can handle no more than perhaps 5,000
`entries. An agent acting as a librarian is
`needed to deal with the sheer magnitude
`of choices. It might serve as a kind of
`pilot, threading its way from data base
`to data base. Even better would be an
`agent that could present all systems to
`the user as a single large system, but that
`is a remarkably hard problem. A persis-
`tent "go-fer" that for 24 hours a day
`looks for things it knows a user is inter-
`ested in and presents them as a person-
`al magazine would be most welcome.
`Agents are almost inescapably an-
`thropomorphic, but they will not be hu-
`man, nor will they be very competent
`for.some tirne. They violate many of the
`principles defining a good user interface,
`most notably the idea of maintaining the
`user illusion, Surely users will be dis-
`appointed if the projected illusion is
`that of intelligence but the reality falls
`far short. This is the main reason for
`the failure so far of dialogues conduct-
`ed in ordinary English, except when
`the.context of the dialogue is severe-
`ly constrained to lessen the possibility
`of ambiguity.
`Context is the key, of course. The user
`illusion is theater, the ultimate mirror.
`It is the audience (the user) that is in-
`telligent and can be directed into a par-
`ticular context. Giving the audience
`the appropriate cues is the essence of
`user-interface design. V/indows, menus,
`spreadsheets and so on provide a con-
`text that allows the user's intelligence to
`keep choosing the appropriate next step;
`
`Petitioner Nissan North America, Inc. - Exhibit 1013 - Page 6
`
`
`
`An agent-based system will have to do
`the same thing, but the creation of an
`interface with some semblance of hu-
`man rnentality will call for a considera-
`bly subtler approach.
`[ny medium powerful enough to ex-
`I r tend man's reach is powerful
`enough to topple his world. To get the
`medium's magic to work for one's aims
`rather than against them is to attain lit-
`eracy. At its simplest, literacy means flu-
`ency. Familiarity (knowing the "gram-
`mar2') is not enough. People who can
`recognize a book and its words, a type-
`writer and its keyboard or a computer
`and its input-output devices are not lit-
`erate unless they can spend most of their
`time dealing with content rather than
`with the mechanics of form.
`Is the computer a car to be driven or
`an essay,to be written? Most of the con-
`fusion comes from trying to resolve the
`question at this level. The prgtean na-
`ture of the computer is such th'at it can
`act like a machine or like a language to
`be shaped and exploited. It is a medi-
`um that can dynamically simulate the
`details of any other medium, including
`media that cannot exist physically. It is
`not a tool, although it can act like many
`tools. It is the first metamedium, and as
`such it has degrees of freedom for repre-
`sentation and expres