Ellen Ullman on Code, Conscience, and the Museum of Me

For this episode of Fast Forward, I spoke with Ellen Ullman, writer of Life in Code, a series of essays that brainstorm in 1994 when she was a programmer in Silicon Valley. She at present makes her living writing by and large fiction, but Ullman remains a great observer of the region, the tech industry, and how the tools nosotros brand are irresolute us on a daily basis.

Dan Costa: Your programming roots go dorsum quite a ways. In 1978, y'all were an English major who decided to make the switch into programming. Why would yous exercise something like that?

Ellen Ullman: Well, I got involved with a grouping doing video, the Sony Portapak was one of those machines like the PC. These things that have been controlled by behemoth corporations suddenly were in your own hands. You could make your own videos, your own stories. You could get around and show them. There weren't restrictions. You could practise porn, etc., and it was a very exciting time. I learned that I liked working with machines. The possibility of working with these machines to do social change and art was intriguing. Eventually, I left Ithaca to get to San Francisco.

Fast Forward Bug Art One must leave the higher town or become lamentable. And one 24-hour interval, I was walking on Market place Street and there in the window of dearly departed Radio Shack was the TRS-80—affectionately known as the Trash-80—and I thought, "Oh, is this anything like a Portapak? What tin can you practise with this thing? Can you make art, social activism?

And then on an impulse, I bought information technology. And so in that location was the issue of programming. Information technology'due south a little harder than getting started with a Portapak, [where you just] press the button. I found information technology hard, but the skillful difficult. I remember that anyone who wants to go into software engineering has to experience, "Yeah, this hard, just also the pleasure of the hunt." If there isn't some sense of seduction near solving issues, it's going to be a very unhappy experience.

And people today don't quite understand that back then, when you bought a TRS-lxxx at Radio Shack, it didn't do a lot. You have to brand it do things. You take to learn how to use it, and that's sort of the invitation that you lot took it up on?

Yes, at that place was a blank screen. It was already an obsolete boob tube at the time. Information technology had a keyboard, and yous recorded programs on a reel-to-reel cassette. And I think information technology held maybe 4K. That was the maximum plan size. And it had the Bones language. And I don't know if people know the deviation betwixt interpreted and compiled linguistic communication. Interpreted means it just executes it right in front of you. So you, you lot know, enter two point, you know, two plus ii and information technology shows you iv. That's easy, getting the starting time footling things going. Doing anything pregnant was very hard, especially since early Basic had many traps. Like y'all could say "Go To" but it didn't automatically Go Back. So you could lose your way in the tangle, and it was known as the Geddy code. Over again, it was frustrating and intriguing.

When I was looking through the book and you lot talk about those early years, what strikes me is the people that you lot were working with. It about seems more like a ragtag drove of artists than engineers and hardcore programmers. Can you talk a little bit almost the camaraderie de corps, or the people you were working with?

Yes. If you look back and—you know who Stewart Brand is, the Whole World 'Lectronic Catalog, The Well—that was one of the offset online communities. And the people were attracted to it. You know John Markoff'due south book, What the Dormouse Said?

Sure.

You know, stoners and dropouts and people who were simply crazy, and wanted to have fun. That was the atmosphere that drew me in, similar the Portapak. These were people having fun, exploring. And the kickoff people I worked with when I had to take a chore and earn a living were like that also. A former Sufi dancer, a woman doing a dissertation and art history, some guy from French republic who smoked Gauloises, though smoking was not permitted. And he'd say, "A reckoner never told me to finish smoking." And these are the kind of people we worked with.

Somewhere around '83 to '86, it changed. Reckoner Science became a common degree, Software Engineering. And we were joined by a heavily male person, self-selected group who had studied Estimator Science as an undergraduate degree. And the atmosphere completely inverse. People were harder to talk to. In my experience, [they] were not also rounded. That is a very broad generalization.

I worked with men who were very well rounded. You could quote Shakespeare with the best of them. Simply that whole temper changed. I guess it's because the profession had to lose its innocence.

How were y'all inspired to practice coding?

Well, again information technology was, "Wow, what is this?" You lot know, when I got my outset real program running, there was a rush of pleasance. "Wow, it works. I did that." It went from nothing to something. And it was like the beginning time I fixed a carburetor. You know, you lot have things apart. Yous put them dorsum together and the auto starts, right? Then this is a rare pleasure, and it'south hard to get to. Once I had that feeling, information technology was a little bit like a drug, you know. You got that loftier and and so y'all lose it, and then you go it back it, or you try to.

You talk a lot nearly this in the volume. You lot say, everybody should larn how to code because it is difficult. And that difficulty is eventually going be the sense of satisfaction that y'all go from it.

I'm not telling everyone to learn to code. As I said, people need to be exposed to it. The point is to demystify code. We are surrounded by algorithms that control u.s.a., and this is no news to everyone who just got hacked past Equifax, for instance. 1-3rd of the adult population of the United States. So, the point is to know enough that it's written by people and it can be inverse past people.

There is a councilman in the Bronx who is proposing a beak in that location that the borough looks at all the algorithms that they are using, and they go from police force assignments to garbage pick-up schedules to what school the kids get to, and look for bias in them. This is the procedure that I'm hoping for, that people in the general public brainstorm to see that these things tin be changed. They accept bias and that bias tin can be addressed.

And then this brings the states to one of your major points, which is that the lawmaking itself contains biases in content. And I call up, with algorithms, information technology takes it to a whole different level. But depending on who wrote the lawmaking, it will reverberate certain political biases, cultural biases. Can y'all just talk a little bit about how that gets embedded into the code itself?

Well, software engineers don't decide what code volition get written. This comes from the height. And the people at the top are very rich, overwhelmingly white men, and their desire is what they think is great for society or what the assignments are.

For instance, a simple delivery app selected ZIP codes. What does that mean? That we tin't make money with people of marginal incomes? Or we don't desire to ship cars up there? It's a simple example. Information technology goes into, is it all shopping? Is that what people need? What kind of social viewpoint does that express?

It'southward all the way downward to algorithms, and even deeper than that. I just read a paper by Phillip Rogaway. He delivered information technology in 2022, and it's nigh the moral implications of cryptography. And he talks well-nigh cryptography as power. For whom are you writing information technology? For what are they using it? Then you lot can meet from the simplest app, down to the very lesser of what is considered mathematical, the bias is built in.

I think you've put your finger on it when you talk most the, "Is information technology all shopping?" because it's certainly all motivated by commerce now, and the reason you target certain ZIP codes is because yous can make money in those Nil codes. And that shapes the product and and so it becomes this reinforcing bicycle. The aforementioned thing happens with every institute. That'due south the reason why Amazon bought Whole Foods and not Stop & Shop. It's because of those demographic biases, and that's merely the marketplace at work. But those marketplace forces are pretty much determining the technological development every bit well now.

I have an example of something I think is simply wonderful. There was a grouping talking well-nigh, "This is coding from the ground up." No, people are not sitting in their glass-enclosed spaces that we work in, deciding that, "I'm a non-profit and this will benefit society and change the globe." This is a grouping working on the footing with undocumented laborers.

Ellen Ullman At present, these guys stand on the street corner, and they get picked upwardly from someone in a pickup truck or a van, and then they are dropped off later, and some of them don't get paid. So they are request, "How can we signal amid each other? Don't get into this. This is a bad guy." It'southward something similar the low era. Hobos who used to put these signs and saying you know, "A bad homo lives hither. A kind woman lives here."

So, they approached a group. Again, it came to the coders from the ground. And they said, "Tin you build u.s. something?" Most of these guys have mobile apps—the app keeps the anonymity, of course—and, they can put in this license plate. "Y'all look for that [license plate] and don't take the job." Y'all could ask, does anyone brand money? Somewhen, yes, because these people who are kind of desperate for piece of work are not cheated every bit much.

It's like a reputation organization. The same blazon of reputation, the aforementioned one that's built into Uber, where riders get rated and drivers get rated, and ideally the best drivers and the best riders get rewarded in some way. It's a fascinating example. Do you have any advice for people who are new to coding or are looking to explore it as a career option?

Advice people looking to explore it? First of all, there are run into-ups everywhere. No, I shouldn't say everywhere. They are in selected Zip codes besides. There are thousands and thousands [of meet-ups] in the Bay Area, where you can sign upwardly and meet with other people who volition teach you to lawmaking. At that place are some for women especially. That is very helpful.

I did a search for other places in the land. Buffalo, New York, Utica, New York. And there was nothing. So, what do we do with that? And then massively online courses seem a possibility. And one of the things that I wrote about, and what I did was, well, enroll to audit information technology. And I saw nevertheless bias in the teachers.

They were making assumptions about American culture. And well, you want to earn a billion dollars. A million wouldn't exist enough for you. All the same. I retrieve people can download these videos, become a group together, you know. You stick the tongue out at the guy who's telling yous social nonsense. And help each other and learn to lawmaking.

There are also code academies that price thousands of dollars, and they were actually closing. They guarantee people, you'll earn iii figures within eight months. I don't think that'due south a expert route. Ideally schools would teach it. And I don't hateful that coding joins the humanities every bit a bones literacy. I recall that people with humanities need to wait into coding to bring that knowledge of people who studied History and Sociology and Foreign Languages, how other people alive in the earth. Bring that into the coding loonshit.

I want to talk to y'all a petty bit about social media. Back in 1998, you wrote an essay called The Museum of Me, and I hear this phrase being bandied virtually as if information technology was brand new. Just you first wrote in 1998 and y'all said that the internet was going to enable us to build, "A private thought bubble," reading just those websites that reinforce his or her desired beliefs, which is exactly where we seem to be today.

You sort of called it, you were describing the net in general, but social media seems to have only taken this to a whole new level. I am and then sorry to exist right near that. At that place were things I saw happening that, I approximate in my heart hoped would be warnings. But, y'all know, I merely wrote for a small publication then-

I think this was probably in Harper'southward. So somehow, the selfie generation missed that Harper'south article, unfortunately.

Well, what are they reading? They are reading feeds, you know. So the feeds are telling them what's happening today. It'southward very difficult by the manner to sit down and talk with you today virtually engineering science, a twenty-four hours after we take this horrible slaughter in Las Vegas. And more than than fifty people killed, and hundreds injured. We really don't know the extent of it.

And you know, guns are technology. Industrial design is applied science. You lot know, we tend to remember of it as just cool phones and laptops and AI, servers, and algorithms. But people design these apparatus that allows a gun to shoot hundreds of bullets with 1 outburst. And that actually stops me. Yous know, in days like this, it is hard to get cheerful. I mean, I'yard the representative of gloom in full general, but this one especially stopped me.

It'southward hard for anybody to but become through the twenty-four hours on days similar this. Only ane of the points you keep bringing upward is that all technologies take these implicit furnishings and that they are not neutral, and a lot of the ethical conversations that surround all sorts of other debates nearly policy, about politics, social problems, tend to not happen in the technological space, considering engineers build it, the market asks for information technology. The designers build it, they send it, so we alive with the consequences.

But it seems at every plow, whether it'southward gun technology, software code, admission to services, we need to accept that conversation about the consequences.

That would be my hope. That'south why I wrote this volume, hoping to initiate a chat on different levels, you know. Non what is the goal of the futurity. The future is what you make of information technology now, having a conversation. Human desire drives the future. That is our prerogative. Also, to look to the past. Another reason I wanted to become back to the mid 90s is that nosotros learned from the past.

Just looking forrad teaches yous very little because that's all fantasy. Just if we get back and we see how nosotros got to where we are, and the trends that got the states here, the successes people had in the by, things that we learned. There are people who lived before us you know, and they've been working on calculating technology more than u.s.a.. We are going into 100 years, I guess you know, at some signal starting with the ideas nearly it. So I hope that people would stop and brainwash themselves. To look back, and endeavor to come across how they got where they are.

What exercise you think nearly the nature of the tech industry in terms of the size of the leaders? With Google and Amazon and Apple and Microsoft, you are talking nearly 4 companies that absolutely dominate entire industries. The web industry, the retail industry, the entertainment industry. Are these companies, is there something almost just being also big that creates its ain problem?

That's a good question. Yes, they are too big. Nosotros all have to live with them for a very long time. And a company like Apple portrays itself equally socially progressive. However, these companies are deeply libertarian. They don't want the authorities controlling them except when they want the government to brownnose to them.

Wait what'southward happening with all these cities begging to take Amazon come into their midst. "Oh, you won't accept to pay taxes for twenty years." Then, government helps them. That's what these companies want. They don't want regulation. I don't see in this political atmosphere, how nosotros are going to do that. We don't accept a Supreme Courtroom that's going to say, "These are monopolies. We are going to interruption them upward."

This is the right question. Yep, they are too large. Many people have to accost what we practice from here.

What is the trouble with them? Because there are companies, I've never seen a private company that has asked for more regulation. What's the risk of having these big giant companies? Is it that they are going to use their market power to exploit consumers? Is it that they are not going to innovate because they've got essentially monopolies in all these unlike industries?

Well, they are going to exploit consumers. We are going to have more than surveillance. There's more competition. I recall they can't but sit back and say, "I'm not doing R&D. I'm not reaching out." But, I'1000 afraid a lot of that is coming from ownership other companies. Startups have these ideas, and so they are purchased. And and so they dissolve their heady powers, if they take any, in this larger corporations.

Merely these corporations cannot sit back and practice nothing. Expect what happened to Microsoft. It really fell behind. Wait at Intel. Intel used to be one of the giants. Now it's behind. People no longer employ large computers and laptops have different chips, and they don't contribute at all to mobile. So they are going to have to innovate. I hope they don't practice it past just buying other people's companies and so diluting the effects of those companies, which seems to be happening.

You lot have an essay about the Millennium Bug, which almost people are going to have to Google to discover out what it was. But I was telling you lot before we went on air that I was really in a newsroom on New Year'south Eve in 1999 and no one really knew what was going to happen. W e knew personal computers would probably be okay, but anybody was sort of worried near what happens in the power plant with that computer in the backroom that nobody's checked on in ten years. But can you talk a trivial bit about what happened and then, and then what lessons that may hold for us today.

Ane of the things that was wonderful near that is why I met programmers who had written this code in the 60s who said, "I never expected this code would be living this long." The company Texaco was the only big corporation that was willing to get public to say, "This is not a hoax. This is what nosotros are doing. These are the problems we are facing."

And their efforts were heroic. And there were small local things that were not publicized. But I knew in my center, talking to these dedicated programmers and project leaders, that it was going to be okay because they were actually planning. The lesson for us now is we are still using all this former code. We've got this flashy forepart ends and then you lot've got the cyberspace, which is really nether stress. And that is very old code.

The reason we are having all these hacks is that the cyberspace was never designed for privacy and anonymity. It was designed to exist collegial and open. No one congenital in the idea that there'd be these thundering hoards. So nosotros have a vulnerability there. We go dorsum to the servers doing algorithms, that's on the operating system chosen Linux. Linux goes way dorsum.

If you go deeper and deeper into what's behind the pretty stuff, y'all detect old code, erstwhile code, erstwhile code. And programmers write stuff that rides on pinnacle of the other lawmaking, and don't e'er—oft practice not—sympathise the details of the stuff they are interfacing with. So the lesson lives on.

Fantabulous. I want to talk to you a piffling bit near AI. It seems like nosotros are in this position with the internet, which has all these unintended consequences that we are still trying to sort out—and now we are birthing this entirely new engineering, which has all these implications. You lot get to a lot of companies and they accept AI systems that are running, that are making business organization decisions and making recommendations, and they literally tin't effigy out how they got there. They tin can't identify the process that reached the determination. They just know it works. Can you talk a little chip most what that might mean?

They recollect information technology works. The trend right now in AI, the whole thought of a humanoid robot nosotros tin can't tell from a human has really gone by the wayside. It proves that it's very, very, hard to understand how homo beings function the way they do. It'due south now auto learning. The thought is that yous write an algorithm. Information technology looks at the results and information technology changes itself, and then looks at the results, and it learns.

Now what happens is, the people who wrote that original algorithm really don't know what happens subsequently several iterations. And the code moves farther and farther abroad from the people who understand information technology. So it is kind of a monster, you know, in the sense that it operates on its ain. And yep information technology works, but what is it missing?

It works on data that has bias. Information technology has a bias first of all, over the past. There are errors. So what information technology'south learning from is also biased and error prone. I do worry well-nigh AI in that regard peculiarly the idea of cocky-driving cars. And that we are talking about actual human harm. And I could talk as well long nigh that, then I ameliorate finish.

Almost of the forecasts say that the self-driving cars volition be better safer than human being drivers, which I feel like human drivers are a big part of the problem on the roads today.

Human drivers take a 100-year history of driving cars. So here's what humans tin can do - they don't merely look at the proximity effectually them. You lot know when you are driving, even if you lot are inattentive, if you have some experience, you tin see ahead of you, quarter mile, half mile up a hill that it'southward all congested, and you are going to have to get fix to stop.

Y'all can read the personality of cars. Someone comes on the far lane over there. You lot encounter the make, the model, the aggressiveness of the driver. When that car moves, you lot know information technology's going to cut you off. You don't expect for a proximity warning. Human beings know in ways that, in a flash that, either I'k going to have a crash or I'yard going to get into a road rage with this car. Or I'm going to say, "Okay, get ahead."

Now, let'southward become to the ways that self-driving cars can crash. Over again, they are relying on machine-to-machine interfaces. And there are going to exist bugs in that code, period. Then they are going to be communicating over the internet or over some wireless network. Big vulnerability there. We've seen that cars operating on wireless accept already been hacked past white hype-hackers to show, "Look, I can make your car drive off the route."

And so how is that protected? And then in that location is the interface between different makes and models. Now do y'all think that everyone is going to be driving a Prius? No, humans want different sort of performance. So there'll be contest once again into how cars operate, how the driver or the occupants want to feel when they're on the road.

At that place will exist some kind of standard API, Awarding Interface, which are famously mistake decumbent. But, at that place are many, many points of failure that can happen in that location. Also, they can't drive in the snow.

There is that.

There is that. I mean, you see, you'd be useless?

Do yous think that we are going to have self-driving cars that are but going to be very error decumbent and to security risks? Or that you don't retrieve that we can actually a build a safe self-driving car and that the engineering science is going to take much longer than people think?

I think information technology will happen. I call up in limited access roadways like freeways, it's going to work well. Unless it's snowing or there's heavy pelting. Information technology has to see the lines. And and then it has to see the break lights and the lights of the cars effectually it. And and then, with limited viewpoints, they are adding other LIDAR and then along.

Somewhen, it will work. It will take a piffling longer. Right now I remember there are some towns that are edifice an infrastructure in the boondocks eye for the cars. Only, we are here in Manhattan and we are having problem keeping our subways running. Practice you think we are going to have billions of dollars to turn Manhattan into a place that has sensors for self-driving cars? I don't know, hundreds of years from at present, if Manhattan hasn't been flooded from the rising seas?

It will be there. But information technology call back that the euphoria—and every bit usual I'yard the gloom representative—I feel like I'm on the seesaw, and at that place's everybody else that actually knows what they are doing, and I'm similar a flea bouncing on the other end, trying to push downward a little bit and level out the viewpoints. Temper the expectations. Anybody should know the difficult piece of work we demand to do earlier that wonderful expectation can exist fulfilled.

I'm going to ask yous a couple of questions that I enquire everybody that comes on the show. And this ane should exist an easy 1 for you. What technological trend most concerns you? Is there anything that keeps y'all upwards at night? Nosotros tin can do a whole another show.

All the above. Information technology doesn't keep me up at nighttime. I hateful, I spend enough time thinking about it during the day. I know, one of the questions I remember you intended to ask was, you lot know, should people take a pause from all these [gadgets and services]? Yes.

Today, I was almost thrown into the street past somebody coming by on a scooter.

And a phone?

With earbuds and a phone. So this really concerns me. People could really get injure.

People get hurt every twenty-four hours in New York Urban center because everybody is looking at their telephone and they are walking.

Yeah I live in the South of Marketplace area. I'm based in San Francisco on it's Startup Alleys, Second Street. Now, it's one thing to say everyone is looking at their phone, but when you endeavour to walk down the street headed Northward when everyone is headed S, and 100 people are looking into their phones, it really is like, I feel similar I'g living in an alien world.

For more Fast Forrard with Dan Costa, subscribe to the podcast. On iOS, download Apple's Podcasts app, search for "Fast Forrard" and subscribe. On Android, download the Stitcher Radio for Podcasts app via Google Play.