Matthew Kirschenbaum: My name is Matt Kirschenbaum. I am here from MITH who is co-sponsoring this talk with HCIL, the Maryland Institute for Technology in the Humanities. And I also teach in the English depart­ment here. I want­ed to vol­un­teer to intro­duce Thom and Mark because I’ve often said that Thom Haigh here is respon­si­ble for sav­ing a year of my life. And what I mean by that, when I was research­ing my book on the lit­er­ary his­to­ry of word pro­cess­ing, Thom’s pri­or research on word pro­cess­ing was an enor­mous resource and sent me in many direc­tions I nev­er would have come across oth­er­wise. So Thom saved a year of my life.

He’s also one of the very best com­put­er his­to­ri­ans work­ing today, as you will see and hear in just a moment. He runs the Special Interest Group on Computers, Information and Society, which also offers its mem­bers that rarest of won­ders nowa­days, an active, vibrant, and pro­duc­tive list­serv, and that’s large­ly due to Thom’s stew­ard­ship of that space. He is cur­rent­ly an Associate Professor in the Information School at the University of Wisconsin Milwaukee.

Mark Priestley… So, I can­not claim that Mark has saved a year of my life. I’ve only just met him. 

Mark Priestley: Give me time.

Kirschenbaum:can tell you that Mark Priestley, who will be co-presenting with Thom Haigh, is an inde­pen­dent researcher with a a PhD and Science and Technology Studies from University College London. Before that, he worked for many years as a pro­gram­mer and soft­ware devel­op­er, and Thom and Mark, and I believe one oth­er indi­vid­ual are the co-authors of ENIAC in Action just pub­lished by the MIT Press. 

So take it away, guys.

Thomas Haigh: We’re very please and excit­ed to be here. The book is avail­able now. They are beau­ti­ful­ly pro­duced. And they’re mere­ly $38 despite the absolute­ly love­ly qual­i­ty pro­duc­tion. So even if you don’t like the words, you’ll love the pictures.

Now, the book real­ly does a num­ber of dif­fer­ent things. Some of what we’re doing in it is in some ways the most in-depth, tech­no­log­i­cal­ly involved recon­struc­tion of pro­gram­ming prac­tice that’s ever been attempt­ed. We also do some things that if you have an inter­est in in the his­to­ry of com­put­ing kind of on the geeki­er side of it, if you want to know why mod­ern com­put­er archi­tec­ture is the way it is, you’ll find an expla­na­tion here that’s dif­fer­ent from any­thing that’s been pre­sent­ed before. If you want to know you know how and where John von Neumann’s First Draft of a Report on the EDVAC” is com­ing from, etc. 

But one of the priv­i­leges of of writ­ing a whole book on a sin­gle arti­fact is that with that nar­row­ness of scope, in some ways we can be very broad in oth­er ways, so in terms of the ques­tions we engage with, the dis­ci­pli­nary per­spec­tives that we try to incor­po­rate, and the time scale. So the book does­n’t stop, as most peo­ple do, sev­en­ty years ago this week when ENIAC is unveiled to the pub­lic. It does­n’t even stop in 1955 when ENIAC is decom­mis­sioned. It goes right up to the present day. The last chap­ter is basi­cal­ly ENIAC in cul­tur­al memory. 

So it does a lot of dif­fer­ent things. When I was writ­ing it, I think I had a sen­tence in the intro­duc­tion which Mark per­suad­ed me to take out that basi­cal­ly any­one is going to be bored by this book at by some point depend­ing on what per­spec­tive they’re com­ing from, but hope­ful­ly every­one will find some­thing of inter­est in it. 

And we have these cool posters. I did­n’t real­ize how many peo­ple would be here, but I think there are about a dozen of them. Mark is going to kind of briefly explain to you what this inscrutably tech­ni­cal but very attrac­tive flow dia­gram means.

The part I’m focus­ing on here, as the title tells, is basi­cal­ly the labor his­to­ry side of ENIAC, not so much the tech­ni­cal side from the book.

We should thank our spon­sors. The research is spon­sored by Mrs. LD Rope’s Second Charitable Trust and Third Charitable Trust. We’d like to thank some con­tri­bu­tions by [Ann Graf, Peter Sachs Collopy, and Stephanie Dick.]

Conventional History of Computing

Several images of plaques, book covers, and web pages making claim to various sometimes conflicting firsts in compurint

So, the con­ven­tion­al his­to­ry of com­put­ing was dom­i­nat­ed in its days, and still is if you ever read the com­ments on a pop­u­lar arti­cle, it’s always bat­tle for firsts. Was it Atanasoff, or Alan Turing, or Atanasoff, or Konrad Zuse, or Eckert and Mauchly who did ENIAC, that invent­ed the first com­put­er? This is a ques­tion that com­put­er his­to­ri­ans are not, frankly, very inter­est­ed in. But it seems to obsess with the pop­u­lar kind narrative. 

You see this most clear­ly here. Alan Turing in The Imitation Game, played by Sherlock Holmes as essen­tial­ly the same char­ac­ter, right? The autis­tic, lone, insane­ly bril­liant genius with no rela­tion to any­thing actu­al­ly exists in human­i­ty. At some point he turns to the oth­er bril­liant code­break­ers. Now, Bletchley Park seems to be about six peo­ple in the movie, but the oth­er five are most­ly just there to be not as smart at Alan Turing. And he says, I don’t have time to explain myself as I go along. I’m afraid these men will only slow me down.”

So he builds this com­put­er single-handed, lit­er­al­ly. There’s a mon­tage: Turing builds the com­put­er. And it’s not actu­al­ly a com­put­er. And in real­i­ty, the bombe… There were more than ten thou­sand peo­ple involved in the Bletchley Park code­break­ing effort. The things were man­u­fac­tured on an indus­tri­al scale by real man­u­fac­tur­ing com­pa­nies. They weren’t even built at Bletchley Park, still less by Alan Turing with his bare hands. But there’s some­thing about that myth of the lone genius that we’re very attached to.

That brings us to Walter Isaacson. And me attack­ing Walter Isaacson is like a gnat attack­ing a bat­tle­ship. It real­ly isn’t going to have very much effect on the world. But there are some good things about this book. He does stress team­work, live­ly writ­ing. He has foot­notes and ref­er­ences to schol­ar­ly his­to­ry. Some of my col­leagues actu­al­ly received it sur­pris­ing­ly well. But to me, that sub­ti­tle there, How a Group of Hackers, Geniuses, and Geeks Created the Digital Revolution” real­ly tells you every­thing that is des­per­ate­ly, ter­ri­bly wrong with the book.

Do you get a sense of quite how Walter Isaacson is dom­i­nat­ing the world? This is Amazon’s top 10 in Computer Industry History. You’ll see the first five places are dif­fer­ent edi­tions of Isaacson, as are two more of them. But the only thing that gives us any hope is up here in the cor­ner, Track Changes: A Literary History of Word Processing. And he was schmooz­ing with the schmooz­ers at Davos. He is in kind of elite circles.

The cover of Isaacson's The Innovators set next to the movie poster for Avenger's Assemble

So when I read this, I thought well it’s inter­est­ing he’s stress­ing team­work but he’s also basically…all his char­ac­ters are pre­sent­ed as super­heroes. And I thought, Where have I seen that before?” It’s basi­cal­ly they’re still super­heroes, but they have to work togeth­er to change the world. And I mashed things up in a col­umn that’s avail­able for free down­load in Communications of the ACM to make Innovators Assemble: Ada Lovelace, Walter Isaacson, and the Superheroines of Computing.” Some of the points that I make here are devel­oped there.

Several men and women working inside the room-like ENIAC machine

Similarly, ENIAC is one of a suc­ces­sion of Great Machines” that peo­ple tend to some­what fetishize. So, it’s basic life sto­ry, because I know not all of you will have mem­o­rized this. It starts in 43 with the idea. Detailed plans had been drawn up in 44; pro­to­type and work in con­struc­tion begins. By the end of 45, it’s fin­ished and debugged. It’s in exper­i­men­tal use through 46 at the Moore School at the University of Pennsylvania. But in 47 it’s moved to the place that actu­al­ly paid for it and will use it for its pro­duc­tive life, the Ballistic Research Lab at Aberdeen, Maryland, where it’s decom­mis­sioned in 55.

Now, one of the things that you don’t get in the exist­ing his­to­ry at all, basi­cal­ly, is any sense of what hap­pens to ENIAC after it’s unveiled. It does appear in the com­put­ing his­to­ry. This is a tree from the 60s show­ing basi­cal­ly all the com­put­ers built up to that point. And where is ENIAC in this? ENIAC is the root of the tree from which every­thing else grows. And that’s how it’s tend­ed to be seen in com­put­er history.

The oth­er thing, when his­to­ri­ans decid­ed we real­ly did­n’t want to take part in these vio­lent and bloody bat­tles over what was the first com­put­er, we basi­cal­ly sent all the pio­neers home with a prize. So we agreed on a string of adjec­tives to insert between the word first” and the word com­put­er.” So ENIAC got the first elec­tron­ic, dig­i­tal, general-purpose com­put­er. And that’s kind of remem­bered as a step for­ward to the next big inno­va­tion, the first stored-program computer.

This is a nice chart from Arthur Burks, kind of how it all that fits togeth­er. A bunch of things go ENIAC, which pass­es them on. But before you get to the mod­ern com­put­er we all care about, you have to add this oth­er stuff. Von Neumann and delay line archi­tec­ture, etc.

A classical painting of a robed woman holding a platter with a man's severed head on it

So, ENIAC’s place in his­to­ry kind of reminds me of this guy here, John the Baptist. It’s an impor­tant place in the sto­ry, but its job is to her­ald the thing that you’re all real­ly real­ly to find out about, which is either the stored-program com­put­er or Jesus.

So to sum it up, con­ven­tion­al com­put­er his­to­ry is obsessed with firsts. It reduces all the mate­ri­al­i­ty and rich­ness and prac­tice of each com­put­er to its sin­gle date of first oper­a­tion. It’s pre­sent­ed as a series of bril­liant ideas, the more abstract the bet­ter, which is why every­one loves Alan Turing despite the fact he had essen­tial­ly noth­ing to do with the inven­tion of the mod­ern com­put­er. And it does­n’t care what com­put­ers were actu­al­ly used for. And what we’re try­ing to do in the book is to fill in, in one way or anoth­er, basi­cal­ly all those miss­ing pieces.

Building ENIAC

I’m going to sketch some of that his­to­ry for you here. Actually build­ing it. I mean, it’s a real machine. It has to be built. It does­n’t just—, Oh! Good idea.” Look, there’s the computer.

Three women working at a large, complex piece of machinery, the differential analyzer

It’s built at the Moore School at the University of Pennsylvania, which had had strong ties to the local elec­tron­ics indus­try. Something no one remem­bers is Philadelphia used to be the cen­ter of radios, and radios were the main con­sumer use of elec­tron­ics. So it was real­ly a hub for elec­tron­ics man­u­fac­tur­ing at that time. It had already part­nered with the Ballistic Research Lab to build this thing, an electro-mechanical dif­fer­en­tial ana­lyz­er that was oper­at­ed by women to car­ry out com­pu­ta­tions. It was a rel­a­tive­ly small engi­neer­ing school, at least com­pared to MIT which was just huge­ly big­ger and better-funded.

The project ini­tia­tors were John Mauchly, a physi­cist who had seen more oppor­tu­ni­ties con­vert­ing to elec­tron­ics, and Presper Eckert, the star elec­tri­cal engi­neer­ing stu­dent who’d been recruit­ed to run the laboratory.

A firing table showing values for a 75-millimeter gun, and chart illustrating the trajectories for various launch angles and velocities

The spon­sor, the Ordnance Department, what they want­ed to do with it was one very spe­cif­ic thing. The whole project was paid for to do one thing, which was to cre­ate fir­ing tables. So, most peo­ple in World War II died basi­cal­ly from muni­tions artillery. And when you fire the shell, you need to know where it’s going to land. You can’t just point it, you have to look up in the table what angle you need. It’s basi­cal­ly Angry Birds.

So they need­ed to work these tables up and they could­n’t com­put­er them by hand fast enough to send the weapons to the war in Europe with them. So it was jus­ti­fied in terms of the urgent needs of World War II.

Now, these guys are rea­son­ably well-known. You find their names in the his­to­ry. They’re the engi­neer­ing team.

But there’s oth­er kinds of work involved with the project that tend­ed to be for­got­ten that I think are essen­tial to the work of inno­va­tion. From the Dean to [inaudi­ble]. The fac­ul­ty mem­ber who’s the offi­cial Project Director. But also peo­ple like the sec­re­tary, the longest-serving draughtswoman, a Professor who helped them work out the math­e­mat­i­cal meth­ods to know how they should build what pre­ci­sion, how many dig­its the machine need­ed, how to avoid round­ing errors.

The BRL, Herman Goldstine was the BRL per­son respon­si­ble for over­see­ing work at the Moore School, but he real­ly became a core part of the project. He kind of went native and was pret­ty much root­ing with the team. And his boss­es at BRL, the users at BRL who were doing com­pu­ta­tions and who the machine was being built for also kind of helped to shape what it would be like.

As I men­tioned, it was con­struct­ed to a degree that real­ly has­n’t been appre­ci­at­ed from care­ful math­e­mat­i­cal analy­sis of the fir­ing tables prob­lem. But it could tack­le many oth­er kinds of prob­lems. So it had a degree of gen­er­al­i­ty there that was unprece­dent­ed. And it could achieve this because of its unique archi­tec­ture. So, the talk’s called Working on ENIAC,” it [could also be?] called Working in ENIAC” because the machine was basi­cal­ly a set of room dividers, and they formed an inner room inside the real room, which the oper­a­tors stood inside. When it moved to BRL, it was the only air con­di­tioned place in the Center, so in the sum­mer peo­ple would move their desks inside the com­put­er just to avoid the head and humidity.

And it had the flex­i­bil­i­ty because con­nec­tions weren’t hard-wired between the dif­fer­ent elec­tron­ic parts. They were wired as need­ed for a par­tic­u­lar prob­lem. So set­ting the machine up to run a pro­gram essen­tial­ly involved using a mod­u­lar kit, which reminds me of the Radio Shack 201 elec­tron­ic kits, to run the wires togeth­er to make the special-purpose com­put­er you need for a par­tic­u­lar job.

It fin­ished up cost­ing like half a mil­lion dol­lars. Anyone who’s been involved with a project that promised to build some­thing may under­stand how that could hap­pen when the bud­get start­ed out as $15,000. It filled about 2,000 square feet, weighed 30 tons, used 150 kilo­watts, com­pa­ra­ble to mod­ern large-scale data cen­ter equipment.

On the oth­er hand, its mem­o­ry rather small at 200 dec­i­mal dig­its of read/write mem­o­ry, 4,000 dig­its of change­able Read Only Memory. And it could do about 300 mul­ti­pli­ca­tions a sec­ond, which was absolute­ly unprece­dent­ed at the time.

Two photos:a close-up of the electronics for a column of vaccum tubes; four women side by side, holding smaller and smaller generations of electronics

To give you a sense of the mate­ri­al­i­ty of the thing, you can kind of see all those beau­ti­ful joints there, that kind of elec­tron­ics. People talk only about the vac­u­um tubes because that was an unusu­al thing, but those oth­er com­po­nents, the kind of work that went into them, were equal­ly impor­tant to the suc­cess of the project.

This is a 1962 pho­to, fre­quent­ly mis­la­beled, show­ing sev­er­al gen­er­a­tions of com­put­ers at BRL. How big one dec­i­mal dig­it was. So this lady here is kind of man­ag­ing to smile while stag­ger­ing under the weight of a sin­gle dec­i­mal dig­it. And by 1962, you could kind of just hold a dig­it like this, and obvi­ous­ly they’ve got­ten a lit­tle small­er since then.

Also, with my inter­est in busi­ness and labor his­to­ry, this made me think about some sides of the sto­ry that real­ly haven’t been talked about at all. Like, dur­ing wartime how do you pro­cure this stuff? And it turns out that there was basi­cal­ly a Soviet-style kind of econ­o­my going on, where dif­fer­ent projects were assigned pri­or­i­ty num­bers by the gov­ern­ment and were sup­posed to go through offi­cial chan­nels to pro­cure, because there just weren’t enough elec­tron­ic com­po­nents to go around.

One of the things I was sur­prised to see that even involved inno­va­tion, they had the find just the right kind, was wire. Early in the project, they found the kind of wire they need­ed some­where in MIT, but no one knew where it came from. So this is in the archives, a let­ter they taped a piece of wire to, sent it out to sup­pli­ers and said, Do you make this wire? Can you help us get some?”

And we talk in the book about some of the oth­er chal­lenges. The pow­er sup­plies turned out to be some­thing that almost killed the project. No one thinks about pow­er sup­plies as this great tech­no­log­i­cal inno­va­tion, but it turned out they need to gen­er­ate 78 dif­fer­ent volt­age lev­els, and the com­pa­ny they ordered it from turned out to be a fly by night com­pa­ny that just com­plete­ly could­n’t do what it had promised.

Precision resis­tors. That was some­where where the Dean came in very use­ful. I’ve got all this cor­re­spon­dence with the mil­i­tary pro­cure­ment where they’re try­ing to explain they need these resis­tors and the gov­ern­ment guy is not under­stand­ing what they mean. Fortunately, the Dean of the Moore School was also the founder of the com­pa­ny that made them, and is still the Director, so I think they were able to go through backchan­nels on that one. And that kind of work was essen­tial to make it work.

The oth­er thing peo­ple for­get is that the com­put­er actu­al­ly has to be built. Walter Isaacson, it’s clear from his text, real­ly has no idea what engi­neers do, has no idea that they are not in fact build­ing the machine them­selves. Engineers, they design things. Who builds it?

It turns out that by the end of 1944, there were sev­en design engi­neers. There were also mechan­i­cal design and draft­ing groups, mod­el mak­ing teams. There were defined pro­ce­dures for when design had been signed off and could be moved from one of those stages to the next one. 

But the pro­duc­tion team was the equiv­a­lent of thirty-four full-time work­ers. So the largest part of the ENIAC team by far were the peo­ple that were actu­al­ly build­ing the thing. And it’s inter­est­ing they’ve been for­got­ten by his­to­ry, because although their job titles were wire­men, tech­ni­cians, and assem­blers, being a busi­ness his­to­ri­an I looked up the account­ing records, and some­times they spell out the pay­roll. You sud­den­ly see all these wom­en’s names like Ruth, Jane, Alice, Dorothy, Caroline, Eleanor show­ing up.

So when I dug through it— You may have heard of the women of ENIAC, the six oper­a­tors. But it turns out before they were ever hired in 1944… And all we could real­ly do for them was make them a lit­er­al foot­note in his­to­ry. We could find the names of almost fifty women. And those are just the ones with giv­en first names. There are a bunch more that I’m sure were there, but they only have ini­tials so we don’t know if they were men or women. In 1944 alone. And I as I say, I wish we could do more for them, but at least they got a footnote.

Another inter­est­ing angle of it. If you’ve ever worked on a grant to sup­port a project, I think you can prob­a­bly iden­ti­fy with this. The work that as the end of the war was loom­ing, and there were spe­cial con­tract pro­vi­sions that would let wartime con­tracts be can­celled when the war stopped. The work that Goldstine was doing spin­ning to his boss­es how it was coming.

So there was about eigh­teen months where ENIAC was con­sis­tent­ly three months from being fin­ished. In May 1944, it was going to be relieved by October 1. In August, it would be vir­tu­al­ly com­plet­ed” by the end of 1944. In September, work was done while it was on the fair­ways.” In December, they were in the throes of com­plet­ing the pro­duc­tion with­in the next two months.” But some­how by May 1945, they were still on the home stretch” and test­ing was [inaudi­ble] to start about two weeks from now.” And test­ing of the full machine real­ly only began in December, although there was unit test­ing before that.

Anyway, it final­ly came togeth­er. There the launch day. Basically the mil­i­tary offi­cial and some spon­sors meet­ing with a cou­ple of the senior peo­ple involved with the team. Good job. They had a nice din­ner, filet mignon or broiled steak. Desserts; fan­cy cakes, which I guess…I think desserts maybe have got fanci­er since then.

I [inaudi­ble] some great pub­lic­i­ty work. Who thinks about the PR side of it? But Goldstine had been work­ing very hard for a long time to spin this to the press, to get the embar­goed press releas­es ready. They actu­al­ly did an ear­li­er demo just for jour­nal­ists on February 1, so that the machine was on the front page of The New York Times the morn­ing before the evening when it was ded­i­cat­ed, which is great PR work, and may con­tribute to the fact that ENIAC became so famous and well-known down to this day. Front page of the times, that’s not bad at all.

Operating ENIAC

As they start­ed to get the thing ready, they need­ed to have some peo­ple to work the thing. I mean, it’s an auto­mat­ic com­put­er, but there’s still an awful lot of labor involved. They hired six women who’d pre­vi­ous­ly been doing the cal­cu­la­tions man­u­al­ly or with oth­er kinds of equip­ment. So they thought the exper­tise that’s need­ed is to under­stand the cal­cu­la­tions. If you can do it with the old tech­nol­o­gy, we can train you to do it with the new technology.

Their duties includ­ed con­fig­ur­ing and wiring units from paper plans; help­ing to diag­nose and cor­rect prob­lems; feed­ing cards in and out of ENIAC—I’ll say more about that in a sec­ond); work­ing with the punch card equip­ment; and work­ing with the sci­en­tif­ic users, the physi­cists and engi­neers who had jobs to run, to help devise the setups that would it.

There’s a pic­ture here of it in oper­a­tion. One that peo­ple don’t appre­ci­ate is ENIAC actu­al­ly had a remote con­trol on a wire that could start and stop it. That’s what’s being held in Arthur Burk’s hand right here. So there would be sev­er­al peo­ple work­ing the machine basi­cal­ly con­stant­ly when it was in operation.

It also worked along­side with punch card machines, because the only input and out­put the ENIAC had was a punch card read­er and a punch card punch. So any infor­ma­tion came in as these holes on a card, encod­ing dec­i­mal num­bers. It need­ed to work togeth­er with a sorter, a col­la­tor, a punch, and a tab­u­la­tor, exist­ing kinds of punch card machines, to do any­thing use­ful. To get the data ready, to get the data out. And in the more com­pli­cat­ed kind of jobs, they real­ly had to do a lot of work with dif­fer­ent [runs?]

This goes with the job on the poster, the Monte Carlo cal­cu­la­tion. You see here a lot of steps are being han­dled auto­mat­i­cal­ly by the pro­gram inside ENIAC. But there’s also a lot of steps that require the women to phys­i­cal­ly work the cards, to sort decks, to add some cards, remove some cards, run them through, tab­u­late them, get them ready, put them back in for the next step.

A complicated flow chart diagram showing many steps across four columns labeled function tables, ENIAC operations, punch-card output, and punch-card operations

This is one of the most com­pli­cat­ed jobs ever run on the ENIAC, the first numer­i­cal weath­er sim­u­la­tion. You’ll see here, this col­umn are the ENIAC oper­a­tions, but they’re doing punch card oper­a­tions that some of them are very com­pli­cat­ed to pre­pare and sort these out­put decks from one step to be ready to turn into the input decks for the next step. So it took them twenty-four hours to run a sim­u­la­tion that would cal­cu­late twenty-four hours’ worth of weath­er. And pret­ty much all that time was spent doing things with punch cards.

ENIAC as a Material Space

There’s also an inter­est in ENIAC as a mate­r­i­al space. When it’s moved to BRL, they are basi­cal­ly build­ing the first ever com­put­er data cen­ter. And there’s oth­er kinds of work and inno­va­tion involved with that that no one thinks about.

At the Moore School, con­di­tions were, to say the least, not very good. There were floods in October and December [1945], and on December 25, Mauchly does­n’t go home until 3AM and accord­ing to the log book, he left behind five men still work­ing, mop­ping water and emp­ty­ing buck­ets with catch drips,” which must’ve been a great Christmas for everybody.

On the oth­er hand, in October the thing caught fire, and for­tu­nate­ly they’d engi­neered a shut­down to the blow­er fans that ven­ti­lat­ed it, oth­er­wise the fire would’ve spread and destroyed the whole machine. As it was, they lost one pan­el and had to make an insur­ance claim.

The move to Aberdeen, there’s all kind of stuff in the files about how they did it. I’d assumed that mil­i­tary trucks would kind of swarm up and sol­diers would run out… But what they actu­al­ly did was hire a civil­ian com­pa­ny and make a hole in the wall to winch it out through.

And for the new space, they had to make archi­tec­tur­al plans to install the equip­ment. So we see a lot of the front of ENIAC. This is the back of ENIAC, and you need a space between that and the real wall so you can get in and fig­ure out what vac­u­um tube has gone wrong and change it. They had to make the ven­ti­la­tion sys­tem, and these are all con­tract­ed to dif­fer­ent kinds of engi­neer­ing com­pa­nies. They had to design the test room. The elec­tri­cal ser­vice. They had to put in a 150 kilo­watt elec­tri­cal service.

And a sus­pend­ed ceil­ing. That is some­thing we take for grant­ed now. It was very cut­ting edge in those days, and they weren’t quite sure if they want­ed to spend the mon­ey on it. They were like, Well, it would be nice but it’s kind of a lux­u­ry…” So in the end they only fit­ted it after ENIAC had already been there for a year, when they thought, Yeah, we’re get­ting so many peo­ple com­ing to look at this. We need to make it look nice.”

As a show­piece it was very suc­cess­ful. There’s President Truman vis­it­ing. Even before ENIAC was fin­ished, there were so many peo­ple com­ing through that that Ordnance Department, the offi­cials in the Pentagon sent a let­ter to the Moore School, Stop show­ing peo­ple around.” Like, let’s get this thing fin­ished. Even with it was tech­ni­cal­ly secret. After it was pub­licly unveiled, they were get­ting del­e­ga­tions through sev­er­al times a week to look at the machine.

Partial photo of a newspaper article with headline "Mechanical 'Brain' Has Its Troubles"

On the oth­er hand, it was­n’t work­ing very well. The con­ven­tion­al sto­ry is ENIAC resumed oper­a­tion in July 1947 and worked around the clock for­ev­er, until it was decom­mis­sioned. That’s lit­er­al­ly what Wikipedia says. In real­i­ty, at the end of that year, six months after they first tried turn­ing it back on, it was get­ting two hours of work done a week. All the rest of the time was spent either try­ing to make it do things or fix­ing problems.

Audience mem­ber: Here it says mechan­i­cal brain,” and it says math­e­mat­i­cal,” but the elec­tron­ic phrasing—and Dianne Martin had com­ment­ed strong­ly about that image-shaping of what—

Haigh: Yeah. The brain thing has been mis­un­der­stood in ret­ro­spect. John von Neumann and some of the oth­er peo­ple involved with the project were very much into cyber­net­ics. And the whole idea of cyber­net­ics is basi­cal­ly peo­ple and machines do the same thing. There’s kind of a com­mon lan­guage, a com­mon math­e­mat­i­cal descrip­tion that works on both sides. Now, part of that is yeah, humans can have brains, machines’ con­trol sys­tems can be brains, too. That sounds weird to us now because that part of cyber­net­ics did­n’t stick. 

But there’s some­thing else von Neumann did, which does­n’t seem weird at all. He talked about com­put­er mem­o­ry. At that point, it would’ve been equal­ly weird to describe a machine as remem­ber­ing some­thing. But that vocab­u­lary stuck and we’ve got no prob­lem think­ing of RAM chips as mem­o­ry, even though the brain metaphor, which is essen­tial­ly the same metaphor, did­n’t stick, and now seems real­ly weird and kind of cranky. So that’s a lit­tle inter­est­ing exam­ple of his­to­ry in action. We talk about that in the book.

Now, Frank Grubbs, this was a great lit­tle sto­ry. He was a PhD stu­dent when he went to do wartime work at BRL. He had to put his dis­ser­ta­tion on hold. He fin­ished up get­ting a month of time on basi­cal­ly the only func­tion­ing general-purpose com­put­er in the world to do his dis­ser­ta­tion with. And he was able to cal­cu­late tables for sta­tis­ti­cal out­liers. So that sounds good, except his month of com­put­er time basi­cal­ly con­sist­ed of two eight-hour shifts when the machine was work­ing. And it took the first three weeks to get it to do any use­ful out­put. Intermittent pow­er sup­plies dumped; They found errors in the math­e­mat­i­cal treat­ment; the trun­ca­tion was mess­ing up the results. They lost time to hard­ware upgrades. They had to put every­thing on hold when the Secretary of the Army was vis­it­ing, and then they heard at lunchtime he could­n’t make it. When they did get some results, they ran it again, which was essen­tial in those days, and they got dif­fer­ent results. Oops.

We relied on the [inaudi­ble] this great source, the ENIAC oper­a­tions log, which had nev­er pre­vi­ous­ly been used by his­to­ri­ans. It’s basi­cal­ly the diary of ENIAC, day by day in this peri­od of operation.

One of the peo­ple who emerges as a bit of an unsung hero is Homer Spence, basi­cal­ly the lead hard­ware main­te­nance guys, orig­i­nal­ly assigned to ENIAC as an enlist­ed man in the Army and then return­ing as a civil­ian to BRL. He was with the machine for almost its whole life­time, fix­ing, sol­der­ing joints, and just real­ly under­stand­ing how the hard­ware worked. And thanks to that kind of effort, and obvi­ous­ly the tac­it knowl­edge that the oper­at­ing team was build­ing up, it went from 25% uptime in the sec­ond quar­ter of 48, to peak in the fourth quar­ter of 1950 at 70% uptime.

So you think of the tech­nol­o­gy as hav­ing the one day it becomes oper­a­tional on, but almost two years after that, ENIAC basi­cal­ly was still a white ele­phant that did almost noth­ing use­ful. And thanks to this kind of work from the oper­a­tions team that has been com­plete­ly ignored, it turned into an extreme­ly use­ful and prac­ti­cal tool.

Upgrades to ENIAC

A num­ber of upgrades were made. ENIAC was kind of lit­er­al­ly made and remade as a hard­ware arti­fact. One of the inter­est­ing things that this poster goes with and Mark is going to very quick­ly click through some slides for you with is that it was com­plete­ly changed to a new pro­gram­ming method in March 1948. And that became what this flow dia­gram here cap­tures, the design work done for what was the first mod­ern com­put­er pro­gram ever run on any machine ever. By mod­ern pro­gram we mean some­thing that express­es cod­ed instruc­tions with oper­a­tion codes and argu­ments that are stored in address­able mem­o­ry. So you can do jumps and con­di­tion­al branch­es and that kind of thing.

So I’m going to switch over to you Mark to just show them— He’s going to impress you with the inscrutable… If any of you were media archae­ol­o­gists, you’re going to love this.

Mark Priestley: The real­ly cool thing about this is that we’ve got this project we start­ed around this pro­gram that was run in 1948, which was like, we claim the first, in some sense the first mod­ern com­put­er pro­gram. And the real­ly cool thing about it is we’ve got a com­plete pro­gram list­ing for it, which one lit­tle bit of looks like this. It’s actu­al­ly thir­ty pages long like this.

What I was going to do was show you a few pic­tures that illus­trate the rich­ness of the doc­u­ments that illus­trate how ENIAC pro­grams were planned and writ­ten and put on the machine. So that text was the 1948 pro­gram. Before that, ENIAC pro­grams were tables or dia­grams, and they kind of looked like this.

It’s a com­plete­ly dif­fer­ent nota­tion­al for­mat. I’m not explain­ing what any of this stuff means. I’m just show­ing you the rich­ness of the doc­u­ments that we had to work with.

Haigh: Yes, if you want to under­stand it, buy the book. It’s a bargain.

Priestley: And then there’s a com­plete wiring dia­gram like this. This is the orig­i­nal ENIAC in 1943:

What these doc­u­ments do is tell you how to actu­al­ly set up the ENIAC to run a par­tic­u­lar prob­lem. It’s not real­ly a pro­gram in the mod­ern sense. It’s almost a wiring dia­gram, set­ting switches.

And this pho­to­graph shows you some of the 1946 famous ENIAC Women pro­gram­mers putting in plugs and turn­ing switch­es to set up one of those pro­grams on the ENIAC, to run.

That’s the only orig­i­nal doc­u­ment we found from one of the ENIAC pro­gram­mers Douglas Hartree, a British math­e­mati­cian. That’s what the nota­tion had evolved to by 1946. And that tells you how to set up the ENIAC for his par­tic­u­lar problem.

Then as ENIAC was being devel­oped in the mid-40s, of course, the EDVAC got devel­oped and com­put­er pro­gram­ming took a lin­guis­tic turn, large­ly due to the fact of John von Neumann dur­ing the famous First Draft of a Report on the EDVAC” came up with the first thing that we would rec­og­nize as an order code, a machine pro­gram­ming lin­guis­tic cod­ed instruction.

Haigh: And we have a whole chap­ter on that report. Where it came from, what it actu­al­ly said. It’s very fre­quent­ly cit­ed and very lit­tle understood.

Priestley: So, what hap­pened in 1947 was that the ENIAC team took these ideas and with John von Neumann said, Okay, basi­cal­ly what we’re going to do is con­vert the ENIAC into a machine that runs like one of these new von Neumann archi­tec­ture EDVAC machines.” And the lin­guis­tic turn, mid-1947 they were com­ing up with plan­ning order codes for the ENIAC which they were describ­ing as “[51] order vocab­u­lar­ies.” So you know, machine code, the orders are…it’s a com­plete turn­around in the ideas about how you pro­gram machines.

So at the end of 47, begin­ning of 48, ENIAC is set up to run this code, and these are old-style ENIAC dia­grams telling you how to set up an inter­preter for the new code.

The actu­al pro­gram that I showed you, there’s a col­umn of two dig­its, which is the only thing that was set up on the ENIAC. The rest of it is all meant for human read­ing. This doc­u­ment is not just a code list­ing, it’s actu­al­ly a com­plete work­ing doc­u­ment writ­ten by humans, meant for humans to read and under­stand. One of the oth­er bits of infor­ma­tion on this dia­gram which is inter­est­ing is these fig­ures in the left-hand mar­gin which are cross-references to the flow dia­gram, which we have on the poster. And every­one’s famil­iar with flow dia­grams. The sto­ry goes that they were invent­ed by von Neumann and Goldstine in 1947.

But the gen­er­al idea of using a direct­ed graph to show a com­pu­ta­tion was in wide­spread use in the ENIAC project even before this hap­pened. So this is a very messy direct­ed graph by John Mauchly design­ing some aspect of the ENIAC. There’s a tidi­er ver­sion from Douglas Hartree in 46, describ­ing the struc­ture of an ENIAC com­pu­ta­tion as a direct­ed graph in what they call the master-programmer diagram.

There’s the first offi­cial flow dia­gram from John von Neumann’s 1947 report, and then these flow dia­gram evolve into the one that’s on the poster, which is the flow dia­gram for the March 1947 pro­gram which was the first machine code mod­ern pro­gram to run. 

Haigh: Not that we care about that.

Priestley: Not that we care about that. [crosstalk]

Haigh: We’re above all these firsts. Just, if any of you hap­pen to care about that at all… The first com­put­er pro­gram has been a com­plete­ly for­got­ten thing. We could­n’t be able to stop you.

Priestley: Slightly more leg­i­bly, a re-engineered flow dia­gram in the style as the orig­i­nal one for the actu­al code that we have. I’m only putting that up here so you can get an idea of the com­plex­i­ty of the pro­gram. This is not a 10-line pro­gram that adds two num­bers or prints out a table of squares. This is a major, 800 instruc­tion pro­gram with nest­ed loops and branch­ing state­ments with sub­rou­tines doing a com­pli­cat­ed Monte Carlo nuclear dif­fu­sion sim­u­la­tion. The first pro­gram ever run is a big, com­plex, dif­fi­cult program.

The oth­er thing on the list­ing that I want­ed to draw atten­tion to very briefly is all the stuff on the right. This is Klára von Neumann, the author of this pro­gram, going through step by step, work­ing out what the pro­gram does, doing a kind of dry run by hand of the pro­gram as part of the test­ing process, part­ly with num­bers, part­ly with math­e­mat­i­cal notations.

And this is what Arthur Burks and Betty Jean Jennings are doing in this pic­ture, is step­ping through prob­a­bly an old-style ENIAC pro­gram, step by step to see how it executes.

Screenshot of an ENIAC simulator application

The rich­ness of the doc­u­men­ta­tion we have is such that we can repro­duce step by step exe­cu­tion of this Monte Carlo pro­gram on an ENIAC emu­la­tor, run­ning the inter­preter, run­ning the code. And if you look, the num­bers there are the same as the num­bers on the list­ing that I showed you a moment ago.

So look­ing at this one doc­u­ment of pro­gram code, that’s three aspects of it. But there’s an awful lot more you can say about this. It is a pro­gram which runs a Monte Carlo sim­u­la­tion of chain [reac­tions in nuclear mate­r­i­al]. The first Monte Carlo pro­gram, it’s run­ning a big appli­ca­tion. How did it do it? It’s a large com­plex pro­gram, includ­ing a sub­rou­tine. The first sub­rou­tine to gen­er­ate pseudo-random num­bers. It’s just an extra­or­di­nar­i­ly rich pro­gram. The whole ENIAC project has just revealed such an extra­or­di­nar­i­ly rich amount of doc­u­men­ta­tion, and it’s been a real eye-opener, actually.

Haigh: Yes. I want to stress we have in the book a chap­ter on the con­ver­sion, two chap­ters on the Monte Carlo. This is by far the best-preserved pro­gram, best-documented, and I think the most com­plex sur­viv­ing pro­gram from the 1940s. So any­one who is inter­est­ed in soft­ware stud­ies or code stud­ies from a his­tor­i­cal kind of per­spec­tive, Mark is the one who real­ly delved in-depth on that, and it’s real­ly a quite remark­able accomplishment.

Another inter­est­ing lit­tle his­tor­i­cal thing, when ENIAC moved on to BRL, not all of the oper­a­tors went with it. One of them, Jean Bartik, was hired under a sub­con­tract to stay on Penn cam­pus at the Moore School and head a group of peo­ple to devel­op pro­grams for ENIAC. So this, I think, was the first time that any­one had been hired specif­i­cal­ly to pro­gram, the first time a con­tract had ever been giv­en specif­i­cal­ly and exclu­sive­ly for pro­gram­ming rather than say for com­put­er devel­op­ment or com­put­er oper­a­tion. And that in a way is the begin­ning of pro­gram­ming as an occupation.

In terms of the mal­leabil­i­ty of the hard­ware, we talk about a lot of the lit­tle changes and adjust­ments they made. One of the most impres­sive was fit­ting a core mem­o­ry unit. It’s one of the first core mem­o­ry units to be used any­where. Core mem­o­ry was the stan­dard mem­o­ry tech­nol­o­gy from the mid to late 50s, real­ly into the ear­ly 70s, for com­put­ing. There was actu­al­ly inter­est­ing mate­ri­al­i­ty of that. When core mem­o­ry pro­duc­tion ran, it was being hand-woven by women in India.

Data Processing Operations Work in the 1950s & 60s

One of the things that you can see the his­to­ry of begin­ning here is the his­to­ry of some­thing that’s been almost exclu­sive­ly ignored, which is the oper­a­tions work of com­put­ing. So Eckert and Mauchly them­selves in fact leave Penn, found the com­pa­ny that becomes UNIVAC, and start the com­put­er to busi­ness. I love these busi­ness­peo­ple here, inside the vac­u­um tube.

And as much as this is an [i‑school?], I’m very inter­est­ed in where this whole kind of rhetoric of infor­ma­tion comes from and how it’s con­struct­ed to make an area of pro­fes­sion­al exper­tise. And you’ll notice here a tiny lit­tle thing. No one was call­ing this infor­ma­tion tech­nol­o­gy at that point. They did­n’t have all this kind of infor­ma­tion rhetoric. And so they were try­ing [inaudi­ble] was a fact-troller,” which is kind of the same thing, but some­how it does­n’t roll off the tongue quite so easily.

In some of my ear­li­est work, The Chromium-Plated Tabulator”, I talk about this ini­tial peri­od of adop­tion. The remark­able thing is that by the late 50s, at which point basi­cal­ly no com­put­er oper­a­tions were actu­al­ly pay­ing for them­selves for busi­ness admin­is­tra­tive pur­pos­es, there’s nev­er­the­less sev­er­al thou­sand com­pa­nies that have installed com­put­ers So they real­ly bal­loon in a huge kind of way. 

Data pro­cess­ing is what cor­po­rate com­put­ing depart­ments used to be called. Now what’s the data pro­cess­ing work­force in 1971? Again, we kind of think of every­thing in terms of the his­to­ry of men doing pro­gram­ming. But we see almost a third of the whole work­force, the biggest job cat­e­go­ry, was women doing key punch work. Basically data entry onto punch cards. One of the things that ENIAC oper­a­tors were doing.

25% is com­put­er oper­a­tions. Working the hard­ware, load­ing and unload­ing tapes, etc. Some of the work that the ENIAC oper­a­tors were doing. Only 17% of peo­ple are called pro­gram­mers, and anoth­er 11% are called analysts/programmers. So that’s under a third of peo­ple in the data pro­cess­ing depart­ment are doing pro­gram­ming work, even though that’s the only kind of work that tends to get remem­bered in terms of com­put­ing applications.

Conclusions

And of of course what we think of now as pro­gram­ming work is a piece of what the ENIAC oper­a­tors were doing, but it was­n’t the only thing that they were doing. So what’s inter­est­ing to me in wrap­ping this up is some big­ger points about the his­tor­i­cal mem­o­ry here, and how it comes to be that the ENIAC oper­a­tors are both famous, and famous as the first com­put­er programmers. 

So there’s def­i­nite­ly a trag­ic under­rep­re­sen­ta­tion of women in IT in gen­er­al and com­put­er sci­ence in par­tic­u­lar. And it seems like the best idea peo­ple have had to tack­le this is to find some women to be his­tor­i­cal fig­ure­heads, to say to girls, Hey look, they did this. You can do it, too.” Ada Lovelace has a day. Grace Hopper has a cel­e­bra­tion of women in com­put­ing. And the women of ENIAC have increas­ing­ly become the third leg of this tri­pod, as the first pro­gram­mers and as proof that women can program.

We talk in the book a lit­tle bit about how that hap­pened his­tor­i­cal­ly. Until the 90s, they real­ly were com­plete­ly for­got­ten. There was a great arti­cle by W Barkley Fritz, who worked on ENIAC him­self lat­er on, where he pulled togeth­er frag­ments of mem­oir and was in con­tact with all the women and got them to write about their own experiences.

Kathryn Kleinman has worked for a very long time to make a film, and her efforts to pub­li­cize the film and pub­li­cize the women did a lot to bring them to broad­er atten­tion, espe­cial­ly into a Wall Street Journal col­umn in 1996. And there’s a clas­sic his­to­ry of tech­nol­o­gy paper by Jennifer Light, When Computers Were Women” that did a lot also in aca­d­e­m­ic cir­cles to draw atten­tion to what they had been doing.

A large stone block with a meta plate wrapping around two surfaces commemorating the six Women of ENIAC

But I have some wor­ries about how nar­row­ly this con­cept of the women of ENIAC is being described now. Their names are now etched onto this thing at Penn. I think it’s just been put up. And the women of ENIAC is say­ing to mean exclu­sive­ly the first six oper­a­tors. Not the dozens of women who actu­al­ly built ENIAC. Or Adele Goldstine, who was work­ing from the begin­ning of the project; wrote the man­u­al, trained the oth­er women in how to car­ry out com­pu­ta­tions. It’s very con­tro­ver­sial whether she also trained them how to set up ENIAC. And was involved in recruit­ing women to do the fir­ing table com­pu­ta­tions. Or Klára von Neumann, who did the 

[The Isaacson quotes are cit­ed are cit­ed on the ENIAC in Action web site’s ENIAC Errors in Isaacson’s The Innovators” page.]

…I showed you min­utes ago is being done by women. But the women are doing key punch work and oper­a­tions. And it’s not some great fem­i­nist break­through to cel­e­brate women only for pro­gram­ming and to say that we have no inter­est in cel­e­brat­ing the kind of work that was typ­i­cal­ly or more rep­re­sen­ta­tive­ly done by women, because that work is not worth remem­ber­ing. That is not any kind of fem­i­nist posi­tion. This has been point­ed out by Wendy Hui Kyong Chun, who wrote 

reclaim­ing these women as the first programmers…glosses over the hierarchies…among oper­a­tors, coders, and analysts.
Wendy Hui Kyong Chun, Programmed Visions p.34

And this is some­thing we very much see in our analy­sis of it. 

So to wrap up, there’s a kind of increas­ing loss of aware­ness of mate­ri­al­i­ty and work in com­put­ing, in the age of the cloud. The cloud metaphor hides from view the actu­al phys­i­cal infra­struc­ture and chal­lenges of com­put­ing. And I think the same in a very sim­i­lar way, that his­tor­i­cal­ly this focus on genius, con­cep­tu­al break­through, and pro­gram­ming has hid­den the his­tor­i­cal real­i­ty of ear­ly com­put­ing as this kind of very mate­r­i­al, very labor, very spe­cif­ic kind of activity.

Now, every­one loves inno­va­tion. It’s sci­ence, progress, the future. I mean, Yahoo is buy­ing you food, what’s not to like about that? So mak­ing that kind of alliance with the Silicon Valley bil­lion­aire ver­sion of inno­va­tion as Isaacson does and back­ward­ly projects into the past is clear­ly instru­men­tal­ly a very savvy thing to do. But it leaves his­to­ry in trou­ble, because his­to­ry by def­i­n­i­tion is about the past, and I could go on for­ev­er about how his­to­ry of [?] might not find a place in the [?] world. But it’s very reveal­ing that the famous Silicon Valley cap­i­tal­ist Vinod Khosla just wrote

If sub­jects like his­to­ry and lit­er­a­ture are focused on too ear­ly, it is easy for some­one not to learn to think for them­selves and not to ques­tion assump­tions, con­clu­sions, and expert philoso­phies. This can do a lot of damage.
Vinod Khosla, Is major­ing in lib­er­al arts a mis­take for students?”

Now, in response to that kind of thing, my friend and suc­ces­sor as chair of the SIGCIS group, Andy Russell, who’s worked on the his­to­ry of Internet infra­struc­ture, said that what we need to do instead is to have a project The Maintainers: How a Group of Bureaucrats, Standards Engineers, and Introverts Made Digital Infrastructures That Kind of Work of Most of the Time.” So that’s his his­to­ry of the Internet, hav­ing done the his­tor­i­cal work to see what actu­al­ly hap­pened, and we have real­ly I think a par­al­lel kind of under­stand­ing of ENIAC.

Closing Thoughts

So, I think his­to­ry mat­ters, even though IT is always about the future and no one wants to believe that his­to­ry mat­ters. And there’s more to his­to­ry than the firsts and lone genius­es. The Imitation Game is ter­ri­ble. Successful IT inno­va­tion has always depend­ed on exe­cu­tion, oper­a­tions, logis­tics, and doing those lit­tle things like the wire.

So what’s our kind of wrap­ping up les­son about what this shows us about the work of inno­va­tion that we see from the work of ENIAC. Well, try­ing to sum up what was hap­pen­ing sev­en­ty years ago, I think I would say ENIAC is the sto­ry of smart (to to very smart), hard­work­ing (to obses­sive), flawed, men and women who came togeth­er to do many kinds of work more or less collaboratively. 

And if you want to think about what hap­pens next, things get real­ly real­ly ugly, and you talk about the decades-long legal bat­tles that fol­lowed the cou­ple of years of very suc­cess­ful col­lab­o­ra­tion. They were in the right places at the right time, sup­port­ed by big­ger insti­tu­tions. I mean, the Army was giv­ing them mon­ey. They had the local resources to draw on. They were able to col­lab­o­rate with Bell Labs to get the relays that they need­ed, with the Dean to get the resis­tors they need­ed. This was a sto­ry basi­cal­ly of a suc­cess­ful kind of bureau­crat­ic insti­tu­tion pulling things togeth­er to make some­thing hap­pen, as much as it’s a sto­ry of indi­vid­ual genius.

They did their jobs, I think, well enough. Not every­one has to be a genius. And even with­out super­pow­ers, they still changed the world. All of them. Even the sec­re­tary, and the draughtswoman, and the wire­men, whose names are forgotten.


Audience 1: In the last twen­ty or so years, we’ve real­ized that soft­ware devel­op­ment works bet­ter as an iter­a­tive process rather than a sort of water­fall one. You men­tioned about the prob­lems that they had with ENIAC that required them to rebuild the thing con­stant­ly. Would you say from your stud­ies of this thing that devel­op­ment itself, whether it’s hard­ware or soft­ware, is always an iter­a­tive process?

Thomas Haigh: Uh…yes. And our poster, you’ll see…the real­ly cool thing is this flow dia­gram. We had space for one whole flow dia­gram, but we thought one whole flow dia­gram won’t tell the sto­ry of this prob­lem. So we start out with John von Neumann’s plan of com­pu­ta­tion that was intend­ed for the orig­i­nal pro­gram­ming mode, and was actu­al­ly essen­tial­ly small­er in scope. Then we have the first full flow dia­gram in John von Neumann’s handwriting. 

Then we actu­al­ly talk in the book about a suc­ces­sion of draft flow dia­grams, and we actu­al­ly man­age to cap­ture that process where they’re find­ing idioms that we take for grant­ed, like using a loop to iter­ate through a lookup table. They’re actu­al­ly dis­cov­er­ing that, and it’s so cool to see peo­ple dis­cov­er­ing this for what may be the first time, in the dif­fer­ent drafts of the flow diagram.

And then we have lat­er flow dia­grams, and we have pro­gram code. We actu­al­ly have basi­cal­ly anno­tat­ed repro­duc­tions of many of this things on our project web site. And then here’s the report sum­ming it up. So we felt enor­mous­ly priv­i­leged that this was so well doc­u­ment­ed that we could actu­al­ly get inside and basi­cal­ly see this first iter­a­tion through the soft­ware devel­op­ment life cycle.

Mark Priestley: And you can see many of the sub­se­quent iter­a­tions on this code dia­gram. There are red cor­rec­tions where they’ve gone back and cor­rect­ed some of the code. There are inser­tions where they’ve kind of—because this was like a [?] pro­gram which ran for like six dif­fer­ent prob­lems. So they have spe­cial code for some of the prob­lems. You’re absolute­ly right. And one of the cool things about this thing is the pro­grams are set up with switch­es on the ENIAC, so they could­n’t keep a per­ma­nent record of it. So the hand-written paper thing was the per­ma­nent record of the pro­gram. And because of that kind of glitch in tech­nol­o­gy, all the iter­a­tive devel­op­ment process is here in this one doc­u­ment. It’s kind of beautiful.

Haigh: And what was dif­fer­ent about soft­ware in those days (and by the way, the word soft­ware did­n’t exist at that point), it was not about abstrac­tion. Because they have to be so close to the hard­ware. They have a very def­i­nite math­e­mat­i­cal appre­ci­a­tion of what’s going on. But they also have to be incred­i­bly aware of the real­i­ties of the hard­ware that they’re deal­ing with. And we kind of see that in the process, too.

Although there is the inter­est­ing sense in which, after the con­ver­sion in 1948, ENIAC hard­ware is only ever run­ning one pro­gram. And the pro­gram is an inter­preter that reads instruc­tions one at a time from the change­able lookup mem­o­ry on the func­tion table. So that you could say is a vir­tu­al machine. You could also call it microc­o­d­ing. But from a com­put­er archi­tec­ture view­point, it’s fas­ci­nat­ing to see those prac­tices emerg­ing so early.

Audience 2: Thank you. Love the mate­ri­al­i­ty of the pre­sen­ta­tion, too. It’s just so cool to see all the lit­tle bits and pieces. I was won­der­ing if you could talk a bit more about the divi­sions with­in the gen­dered divi­sion of labor that you kind of end­ed on. Like, who were the women doing pro­gram­ming work ver­sus oper­a­tions work. What were edu­ca­tion­al back­grounds? For whom was it a career ver­sus a one-time thing, or that kind of stuff.

Haigh: Yeah. Well, pro­gram­ming work and oper­a­tions work, at least in the orig­i­nal year of oper­a­tion at the Moore School, were the same women. And they’re remem­bered only as pro­gram­mers. They did all the oper­a­tions work for all the prob­lems, and work­ing the machine required at least two peo­ple hands-on constantly.

They also helped… In some cas­es they gave assis­tance, in some cas­es I think they took the lead, in devel­op­ing the con­fig­u­ra­tions for the ten or so jobs that were in the Moore School. Some of them, peo­ple just turned up with every­thing planned out. In oth­er cas­es, I think they gave a lot of assis­tance to peo­ple who knew what they want­ed to do math­e­mat­i­cal­ly but did­n’t know how to con­fig­ure ENIAC.

So the point is, in those days, and we have a nice quote from Adele Goldstine… Is it in the man­u­al, Mark? Where is Goldstine describ­ing the work of the oper­a­tor in a way that is group­ing togeth­er stuff that we would think of as pro­gram­ming and real­ly quite high-level work, with stuff that we would think of as being very kind of blue col­lar work. At that point, those are seen as tasks that all go togeth­er. So that sep­a­ra­tion takes place later.

Priestley: Yeah. These women were hired as oper­a­tors, but with­in about six months they were the machine experts. So a sci­en­tist like Hartree comes in with a prob­lem. He kind of roughs it in as the algo­rithm. He writes out that pro­gram I showed you, and he makes some mis­takes, if you read through it. He’s mis­un­der­stood some aspects of the ENIAC. So pre­sum­ably that gets cor­rect­ed, and then in the acknowl­edge­ments of the paper he then pub­lish­es, he acknowl­edges Kay McNulty, the oper­a­tor who helped him. 

It was a whole list of things, run­ning the machine, set­ting up the machine, help­ing him plan the com­pu­ta­tion. They kind of evolved as real­ly com­plex, multi-faceted job descriptions.

Haigh: Yes.

Priestley: —does­n’t kind of cor­re­spond to lat­er ter­mi­nol­o­gy at all.

Haigh: And so they absolute­ly deserve to be remem­bered. But the inter­est­ing thing is com­put­er sci­ence has been inter­est­ed in only remem­ber­ing them as pro­gram­mers, not as oper­a­tors. And no one has had any inter­est in remem­ber­ing the women who actu­al­ly built the thing.

Matthew Kirschenbaum: Question for Mark, actu­al­ly. So we’re look­ing at the doc­u­ment here, and can you just… Where do you start, as a media archae­ol­o­gist? What is your Rosetta Stone? How do you actu­al­ly go about under­stand­ing what we’re look­ing at?

Priestley: In this case, we start­ed out with a flow dia­gram. So the start­ing place was the flow dia­gram. And we also had a large doc­u­ment they’d writ­ten which sum­ma­rized the behav­ior of the program.

Kirschenbaum: So you had documentation.

Priestley: There was a flow dia­gram, there was doc­u­men­ta­tion which described both, and it cross-referenced the flow dia­gram and the code. Armed with that, we could then go through the code and lit­er­al­ly a line-by-line anno­ta­tion of what every line of the code did.

Haigh: I just want to show you a few of the things we’ve put online that you can see for your­selves. We’ve got the orig­i­nal dia­grams in 1943 for the tra­jec­to­ry com­pu­ta­tions. There’s a myth that these were devel­oped very short­ly before the pub­lic demon­stra­tion in 46, but a great deal of the prepara­to­ry work com­plet­ed very ear­ly and shaped the design of the machine.

We’ve got dif­fer­ent ver­sions of the design for the con­ver­sion. And the first step in under­stand­ing those instruc­tions would be to know what the instruc­tion set is, and we’ve repro­duced the tech­ni­cal report that describes that. 

We’ve also repro­duced the pro­gram code, and we’ve repro­duced the flow dia­gram. So if any of you want to try this at home, basi­cal­ly all the key mate­ri­als are now online. And we would like to thank John von Neumann’s daugh­ter Marina, who gave us per­mis­sion to repro­duce those mate­ri­als from her father’s papers at the Library of Congress.

Audience 3: My ques­tion was won­der­ing if you looked also on how peo­ple at that time were describ­ing work in gen­er­al. Not by com­put­ers, but by say an army of peo­ple. If you’re going to build a pyra­mid, you’re also going to have loops like, a pile of bricks, and have one per­son come and pick it up until it’s over. So there must have been oth­er peo­ple in oth­er domains that were not com­put­ers that were describ­ing work in some work­flow fash­ion, and some instructions.

Haigh: Well, sure. That’s basi­cal­ly the his­to­ry of man­age­ment. I mean, there’s Taylor and Gilbreth. That’s in some ways where the basic ideas of man­age­ment are com­ing from. By this point we start to get into the whole thought exper­i­ments and human rela­tions era. It’s kind of on a cusp. But what’s hap­pen­ing in labor that’s very impor­tant there, obvi­ous­ly, is The New Deal, the CIO, labor unions, and we’re more or less up to Taft-Hartley. (You see, I’m also a labor historian.) 

So it’s kind of a peri­od where you could say that con­cep­tions of work are in flux. A big thing, obvi­ous­ly, is is this like Rosie the Riveter? Are women only doing this work because it’s wartime? Or are they doing it because there’s a prece­dent for women doing this kind of work? And if you think, as some peo­ple have done, of this being pro­gram­ming and an unprece­dent­ed thing, you can’t answer that question. 

But that’s not how they thought of it. They thought of it as being women oper­at­ing machines. So, women had been oper­at­ing the dif­fer­en­tial ana­lyz­er. Women had been oper­at­ing cal­cu­lat­ing machines. And women had been doing the cal­cu­la­tions by hand. And it turns out that while some areas like physics and engi­neer­ing were extreme­ly hos­tile to women, women did rel­a­tive­ly well in math­e­mat­ics, par­tic­u­lar­ly in applied math­e­mat­ics, which not coin­ci­den­tal­ly was the lowest-paid part of math­e­mat­ics. And par­tic­u­lar­ly with­in applied math­e­mat­ics, doing the grunt work of com­pu­ta­tions, which was the low­est sta­tus of work. But at this point, women were actu­al­ly get­ting PhDs in math­e­mat­ics, par­tic­u­lar­ly in applied mathematics.

So I think one of the things we try and do in the book is sit­u­ate this not as the mag­i­cal start point of pro­gram­ming, but as a con­tin­u­a­tion of an ear­li­er tra­jec­to­ry of female labor with­in applied math­e­mat­ics, which I think is the spe­cif­ic con­text in which it makes most sense to con­sid­er it.

Likewise the women build­ing the machine. On the one hand, they might have got men to do it if the men weren’t all doing the war. One the oth­er hand, women were work­ing in radio fac­to­ries, and women were work­ing with­in the Bell sys­tem to pro­duce equipment.

And one of the things that labor his­to­ri­ans know is that the con­struc­tion of what’s low- and high-skilled work is com­plete­ly sub­jec­tive, and that any work that women do tends to be seen as low-skill. So you say, well they can do that because they have such great lit­tle fin­gers, or some­thing. And this oth­er work men can do because they have skill. And I think that kind of work with doing the wiring and so on (although unfor­tu­nate­ly we have no doc­u­ments that prove this) to the extent to which it was a con­tin­u­a­tion of labor prac­tice in those indus­tries, it would be based around the idea that women had the kind of del­i­cate hands need­ed to do the work, and also you did­n’t have to pay them very well.

Three women holding sheets of paper, standing in front of ENIAC panels comprising hundreds of rotating switches, making adjustments to the program settings.

Priestly: I’ll just make one very brief com­ment about mate­ri­al­i­ty, because this slide got missed. The point where the abstract dig­i­tal machine code, the two-digit machine oper­a­tions meet the met­al of the ENIAC is on these tables here. This is the ENIAC’s old func­tion tables. And the pro­gram was actu­al­ly set on these things by turn­ing these switch­es, and that’s the point where the abstract Von Neumann machine meets the met­al of the ENIAC.

Haigh: And if any of you are old enough to have used a com­put­er where you’d debug with switch­es on a front pan­el, you’ll appre­ci­ate that with the ENIAC all the pro­gram and all the data was imme­di­ate­ly appar­ent with its own switch­es. Plus, you had lights on each accu­mu­la­tor that would show you the con­tents of every dig­it of writable elec­tron­ic stor­age. So when the machine was work­ing, actu­al­ly, it must’ve been very easy to debug. And if you want­ed to change a pro­gram when it was run­ning, you’d just turn the switches.

Further Reference

Event list­ing at the MITH web site, with sum­ma­ry, bios, and com­plete slides.