tv David Auerbach Bitwise A Life in Code CSPAN December 15, 2018 6:00pm-7:01pm EST
that'll happens tonight on c-span's to booktv. forty-eight hours of nonfiction authors and books every weekend. television for serious readers. here's david on how a rhythms and computers help us understand the world. good evening everybody. my name is billy, we are excited tonight. we have had a long and very fortunate experience, our relationship with the magazine. tonight will be no different. as we engineer a more integrated algorithms to translate and mediate our experience, and now the gap that divides us from machine, we willingly about that's what makes us human. this is why his thoughtful owed to computer linkages taps into
the imagination of his childho childhood, written algorithms that have come to tax size speech, knowledge and behavior. he's a writer and software engineer. please make our guests feel welcome. [applause] >> thanks for coming. you guys are great. this is my first time in the booklet store. i'm glad you're here. i'm going to give you a sense of
what the book is about. you may have read it already. i like to take a different approach to it. i think one of the things that stopped me as a writer and as a reader, his intellectual autobiography. this is a difficult genre. not a lot of people do. how does one write about how you came to learn things and how you came to on things and question what the certainty and truths that made you feel where your learned your professional trade, is a very fickle donor. i think david has done a wonderful job at it. we're going to talk about that. the thing about memoir, as usual in overcoming of the trauma or
crazy parents or external challenges, persecution. this is a story that can be positive in the early years, picturing his -- something who was nurtured by computers. [laughter] that's an important generational pale to tell as well but before we get into that, or to let david read from the book a bit and bring on a special guest. a great part of this book is about not so much what we can learn from computers but how we can learn to still be human and still be ourselves in the way in which humans are different from computers. >> thank you. this is from section about probing my child. it was integrated by maria, the
person who gave me the idea of writing in the first place. it was not in the original plan. thank you, maria. and inspired suggestion. in this chapter, is about child development and comparing and contrasting that with one what might call computer development which is a think what the machine learning techniques we have actually do. they are development to other than algorithmic. it starts off with a song was composed by my daughter, elinor. i like to have her come up and read it for you.
presenting eleanor. >> it can happen to you and computers as you are a computer. go into it. program yourself. the most powerful gun, do it. explore it, explore everything you have. if you can do that, not even houses because houses are us. bones push you, that's what i do. [laughter] >> she will return in a bit. because she was somewhat the
co-author of this chapter. i'm going to read this. a few years after leaving google, i started another product that is still ongoing. sent a progress report for the first 12 months. difference, my daughter spent two weeks away with her mother visiting family and receiving operates. i'm reunited with her, i'm examining her futures. the malfunction of the older model. the sun plenty of space but we want the quality upgrade. she wasn't simply changing on her own. really, you put food in one and then she has the ability to crawl. it doesn't work for computers and it doesn't work for babies. in fact, all these skills come with upgrades. due to monopolies lobbyist money
and antitrust mall, they have to come in packages. my daughter can be standing and turn book pages and says my mama upgrade, there's something about a luxury item in high demand. cheetos food in the floor, place with remote control and to two unprocessed. also the go to sleep feature which has her put her head on the ground, only to rise back up within seconds, whose idea was that? it certainly qualifies as an upgrade. so many of her expenses that we're going to be stuck up pretty before too long. she only now seems to be getting a hang of the current one. she thing steps forward without holding on. to gain more motor control. we paid for, like toys. she also seems to have figure
out for focal. she can do that yet. and incompatibility remains the. my daughter is not compatible with the card so we won't be able to educate her from my old laptop. no have my phone connected her over bluetooth. either would be more converted convenient been talking. >> these upgrades became more difficult to track. my daughters skills expanded and her comprehension of the world around her developed. in all these in comparison with fine transcendent, language, for anyone who's wrestled with the ambiguities and frustrations of howling what works and doesn't work, seeing a child maneuvering
both revealing and compounding. gradually, nationality starts to develop. by three and a half, elinor was modeling her motives. she didn't always do so flattering, would you like to come up and read this? this is -- okay. we will it off, so -- it's her blanket. >> you will like it. but i will do it. [laughter] >> she was able to determine everyone around her had sometimes these conflicted with her. she couldn't necessarily determine others motivations but she knew they were there. we are going to switch back.
at that time, my daughter began to understand things could be made out of other things. sometimes things she could not see. she knew bodies were made of bones and blood which led to this song. there will be dialogue after this. ♪ here he comes, here he comes. now he's wet. now he's wet. ♪ always walking through us, the now he's wet ♪ ♪. why did i get what? >> lead. you walk through us blood. [laughter] >> she created this explanation. thanks, we'll be back in a moment. she created this out of the raw materials for observation of the world around her.
the data is once of the most amazing in nature. many parent know how baffling this is. their limits, this implements the mysteriously finish replaced by new independence. the shared understanding we all possess. it cannot be rushed. she knew bodies were made of blood and bone, she cannot conceive the hierarchy of substances. she saw an atom in a book. we might need to adjust the microphone. i said to her, everything is made of atoms. even you. >> now, i am made of bones. >> your bones are made of atoms. >> no, our bones are made of muscles. >> your muscles are made of atoms. >> my muscles are made of muscle.
[laughter] >> remains a puzzle to me, how children can imitate the pre-association to reasoning. the brain grows and develops into neurons year after year. money, memory, and never gained any new reasoning capacity. there may -- many different types of reasoning. neural networks, and belief networks among others. always fall under the rubric of machine learning. today's most powerful learning machine, accelerate recognizing similarities. whether the patterns are made of words or soundwaves, they can judge by the pathways, what structures and word choices. they can say little about the text meaning.
more like a dog or cat. they knew nothing what i've cat or dog is. they can the humans they go, they cannot discern whether particularly our beautiful leslie train now for them. these machine learning networks can perform, they inherent context and judgment from human and they are unable to generalize a given task. they cannot reason about the application of labels by daughter did at age four. >> these are so soft. maybe they are made of polyester. >> maybe made of marginal is. you can eat them. >> you cannot eat them. ballet shoes. >> they are made out of marshmallows. >> nothing is made out of marshmallows. [laughter] marshall is a maid, made out of marshmallows.
that's why they are called marshmallows. used by people and others stuff. [laughter] >> this is verbatim. amazing living network cannot switch from visual to non- visual categories. needs excesses the coded directives given by humans. it can label pictures of marshmallows but it cannot confer marshmallows are made of mesh laws. if it's a coded in our genes, we have yet to discover it. i wonder how my daughter emerged to create this part. there are competing systems in play. nationality, narrative, visual injury and more. there's no explanation of how this occurs. many suggest that any certainty somehow rational human adults are most capable of coping with exploring the world.
they also create problems for us all, for ourselves and imposing it to the whole planet. another byproduct. i thought that becomes more abstract around the 85. my daughter who lost her memory of her early years as all children do, the age five, i asked her the meaning that she forgot she made up. that was? >> here he comes, he becomes. i was walking through us now he's working through us. now he's wet, now he's wet. >> i asked her again, what you think i was wet? >> water at that time? >> i said that, a logical explanation. she was perplexed to hear the real explanation.
she said, did i say that? the entire nature of the word wet, transferred. it was now a real concept. now she knew blood had nothing to do with how i could get wet. in moments like this, the whole of child development seems packed in a single word, the meaningful word causes human consciousness. she even came up with explanations to why she hadn't known things before. she was perplexed there was a time she didn't know the number that went on forever. >> once babies get to know 100, they don't really know there are so many others. one hundred is the highest of numbers. they hide secrets from babies. the secrets are 101 and the numbers of the above it. >> thank you very much. [laughter] [applause] just to finish that up.
the transition from association to rational excavation, ginger and unknowingly make is a mystery. the problems facing, moved from a specific to general inhumanly rational way. to take it from one clearly defined task and put it to new and different uses in general purpose thinking network. i suspect pushing this will require networks to engage with the physical world and a variety of different ways, processing visual and verbal information. learning slowly what approaches do and don't work. the network would need to be sufficiently powerful to apply the same techniques to novel problems. we still have a long way to go. they implied the behaviors to determine whether a feature can sink. i witnessed an increasingly sinking in ways they have not
done before. this is a process with our thinking, something i like it, that will be good enough for most. no wonder we are desperate to program it. as my older daughter asked me, you get the final word on this. her final question, -- >> computers life and we get to feel like what it feels like. >> thank you. [laughter] now that we've sufficiently exposed your child in her early, it's your turn. [laughter] you could not escape.
you're not as cute. [laughter] this book is also a story of your own cognitive development. you started out as a programmer. then you became a writer. or were you always the same? >> they were definitely not the same. if anything, the book is my attempt to explain how i tried to reconcile. they did exist with such disjoint part of my life's for so long. i was a happy kid until i became a moody teenager which then i became very obsessed with literature. i read before that point but it was more my nerd like, like
fiction. computer programs and then i moved to literature. i got it for my parents. my parents were like that. later on, where the teenager, brought home books because i thought it was this deep -- my dad was like, this is so difficult. why you like us? this. this is a realm of the book, my dad said, so weird and complicated. i think it was a final straw for my father. freaking 20 minutes, david. why are we watching this? this is worse than 2001.
something was pulling me in that direction. i got exposure to it through the experiment to literature people who experiment with that method. some of which required computer knowledge to create it. it remained a really a divided thing for me. graduated from college and i wanted to do computers because that was what seemed to be the more definable thing. i didn't think the thing that computers were going somewhere. so writing my free time. >> that was a good choice. [laughter] as somebody who took the other route. i know you can compare on that income. >> i think the thing is, i was presented with the option because i think for the programming for the whole time, is presented to me in a more
boldfaced manner. my wife with a poet. she went further than me. i sort of paste that, too. she ended up with -- she didn't burn out like i did. she still works at google to this day. shall i call her? an expert software engineer at google. she's making more than me. [laughter] >> are very alert. in a way in which -- you say the department with literary studies and literature, they really want to be tech. >> certainly the case. >> i your describing, in the 90s,. >> was becoming more of this
functionally oriented. a friend here who can speak to this better than i can because she went on and got her phd. i feel like i was being prepped to respond in certain ways and i had my own thing going on. so i drew from aquatic sources but i only connected with particular things. i took a writing class from one of the great authors of our time. i felt a sensibility third that i hadn't felt in other classes. i guess that since, if you want to put possibly, dissemination, maybe i should not constrain myself. i should try to get myself in a situation where i feel constrained. going to programming.
i found, once i got to new york, across the top i had been who was one of the original ones. a huge fan. it was very daunting, it still is. here is cycle. memory. he could recite them by memory himself. i had the influence coming and even while i was programming and ultimately, the acted as a firing something. i decided i wanted to pursue it. as a career, even then, after constrained by the rigidity of technology, i started to feel constrained by the loosey-goos
loosey-goosey. [laughter] that's what got me onto trying to figure out how the aspects, the techniques of one could be informing the other what was going on in my head. how this might apply to the world more generally. computers were having an increasing influence on every aspect of our lives. but something i lived with growing up. >> how you feel about the term technologically humane as him? is a label. >> if you put a label on you -- >> resisting.
humanism is a strange john. >> it's falling into a term. what is human enough. >> what baffles me, it's taken -- behind it on one side, human on the other, i think i go for humanism. it seems like a lot of people on the other side. i don't want to put myself in a position of snow. i think that the thing, ct snow was a midcentury english scientist philosopher, writer, novelist. i don't know. would be probably. he was saying, the huge debates with him.
the english critic. a lot of the more the role of how science should play a part in the term humanity. he brought a book, cachet called cultures about it. basically think people need to learn more about science. because what now is that things have changed because science is impacting technology and people in an everyday way. that couldn't necessarily be seen to that extent. back then. i'm not arguing for that so much is just a more circumspect look at the assumptions made in each of these names and are increasingly fractured. you think of the term?
i don't know. >> rejecting labels. i think what we want to -- why should we drive labels? you are a programmer and even when you are writing, you are, i think, in some sense still. what you play out, people who aren't consciously programming, program themselves by using technology. >> we have the systems that you mention it, i think it is one of the things, mirroring and projected. how one flips into the other. we apply these labels to themselves. to ourselves. when these labels are incorporated into societal or governmental taxonomy and then used to classify and regulate
them, they become descriptive rather than descriptive. i think the main danger is a nuance get lost, you have these categories that can be gender categories, personality categories, they can be categories of medical pathology and you're taken them as real things rather than has human constructs. i think one of the roles of technology is to amplify the effect because technology presents these objective quantitative form. if you see that everybody is in one of personality types, the washington post the other week, the claim for personalities. [laughter] that everybody fits into. very skeptical of these things.
what you start using these categories, a very widespread way, they become prescriptive. you are to anyone of the member of companies now and you work for them and they classify you by a union type to figure out what sort of work you are best suited that. >> are you allowed to disclose how they did this? >> i talked about in the book. if anything, i say google probably does less. it's more like management companies that are more insisted on figuring out, are you the and word?
i view the creative type, the leader? we're going to determine your career path based on this. there is skepticism toward the soft social sciences. within technology companies. so course, technology helps. categories become reified over and over. i was classified as an inwardly directed about three quarters of the people as well. >> how you sought out the peop people, color? almost like the humor. >> it comes before humor. you drip from union -- i'll leave for pax on our drive, no most are derived intimately from humors and i mean, it's not science. >> you are --
[laughter] you sought out claire people? or? >> i wasn't sufficiently mean to take the exercise seriously. i wanted to hang out with children, going to go with a nice, from the people rather than the intent, i'm going to argue amongst ourselves for five hours to figure out the with best way to do this. >> you think you're mis- categorized? >> i think the category system is invalid. so you can't be miss categories under and missed categorized system. it's like i if you play wrote, if you get the road -- wrong card -- [laughter] is not an answer to that the validity of the system is
something that is postulated rather than established. whereas with something like, if i break my arm and you tell me i have a headache, yes, i was missed categorized. that's the distinction i am drawing. so i was like back. [laughter] >> show us the picture. >> imac in the book but i forgot which one it was. so the green, the analytical logical theoretical diverted engineer, the blues with how many peacemakers, those of the people i chose to met with. they also said they air on blood. gold is traditional. structural of bureaucrat. you want many of those. finally, orange was the outgoing creative visionary leader which is firing -- stay away from
them. [laughter] they just want to tell you what to do. >> alive and well. you are prepared for, i think, some people -- one of the great things i love, i read this, apart from the site, the only writer of our generation born between 1968 in 1980, who writes about the importance of dungeons & dragons. beyond the circle of teenage life. clearly, i thought i had outgrown dungeons & dragons and i was wrong. it turns out the whole world is based on dungeons & dragons.
>> i found something interesting. some of you know jason who finished his graphic novel called birthing. an excellent work. i did not know this, he's a dungeons and dragons somatic. his created his own like role-playing game. he was probably, he was actually a role-playing game fanatic. you think it has been approved on. yet, not much you can see in this in his work. much of a quantitative impulse as i could see. i asked him about this becaus because -- he said something to the effect of, that's the job of the dungeon master. bring out qualitative factors. computers can do this. >> how did you get put into
computers? can you explain? structures, parts of our lives you wouldn't think facebook -- >> dungeons & dragons house a very simple set of mechanics for training the game world. harper train physical and mental characteristics like strength. you calculate whether your character you are playing, 60s or fails. it depends on whether they need a certain bar. euro that ice, is a higher than your strength? is it is, you failed to list the trapdoor that collapsed on you and kill deer. that ability to regiment the world into numbers is exactly what computers do. anything is on a computer at its
root. it is indeed exactly like that. it's required. all computers are doing it. a lot of symbols on the numbers but nonetheless, they are so precise. you need to have some form of quantification and labeling. not just arithmetic. it's healthy the aspect up, the same mental illness, here are bonuses and the types of insanity that he created for dungeons & dragons. he chose the corresponding once because -- >> victorian for capillary of mental illness. >> is on school. the same basket of any. they were involved, in the
driving you insane. >> crucially dungeons and dragons sexist. as it turns out. [laughter] >> anything from the trollop to the partisan. became invented -- >> this has made its way into computer ontology. >> i think what happened was, there were a lot of female program is -- a lot of computer scientist. if you look at the numbers, what happened was the percentage dropped off in the 1980s. my theory, a lot of work done on
this, but the amateur series, a lot had to do with the personal computer and affiliated consumer software coming from consumer hobbyist scene. it was much more male-dominated. you had an influx of a larger number of men who were interested in computers how'd an opportunity to express and study it. from scenes that were traditionally very male dominated greater system than women because there are a lot of female computer scientists. if you look, you can see they occur straight from the 40s when you have female programmers, all to the 60s, when you have women doing anchorage composition and only in the 80s that you see the
propulsion starts dropping. more men come in and remain more the same. i think there was asking when that took place on them. when i graduated, i think there were out of 50, there were two women. >> before we get to questions, i want to ask you, more thing. similars ways, you are skeptical of what computers can do and are doing, but you are not disillusioned in the sense of angry. let's go back a little, before
about how if we want to program a.i. to liles, we have to love them back. with implication. what should our relationship -- would help you to -- do you think some coding knowledge is basically -- we should know about it as much as we know about what's across the street -- >> the germ of this book lies an article that was real few years ago that was called stupidity of computers. i'm grateful for commissioning that in working with me for quite a while. struggling back and forth.
[laughter] i was a very top-down writer. he encouraged me to make it more bottom-up. it helps me write in a more acceptable style is supposed to my wife called what was bread like it was translated from german. [laughter] >> transited for them german? >> i learned how to you'd -- use verbs. [laughter] one essay was to say that he wasn't necessarily going to have those. se, it's more like saying, it doesn't need to be a writer in order to benefit from knowing more about how language works. is an understanding that language conditions how we see in the world, you have to be complete to appear -- to understand that language, the
conditions how categories are divided. conditions how people are treated. it can be harmful. it goes for computers. one doesn't necessarily need to be a coder but doing coding could be a good way to get familiar with it. benefit from understanding what it is that computers are doing and why it is we have problems with some areas of human life and why it is that they are so inedible to others. just having the understanding and make more attentive to what might be getting lost. to the extent that they prescribe our futures for us and describe who we are. we effectively have this digital fingerprint that people often see before they even know who they are.
sometimes it's not flattering. >> it can be life-threatening in some circumstances. if you identified or -- >> certainly. i talk about when governments do it, both in india and china, you are seeing increasingly popular, to classify people and china talking about reputational sports and others that are being used to serve in a over identification center. simple security security is one of the minimal problems they have. also understanding how these things transform human essences
into labels and categories is something that are very useful in understanding being seen in such a way by facebook, google or even by my son. >> figure. you have questions? this audience looks smart enough to ask. >> on anything really. book covers, everything. whatever you -- >> what is your favorite part of the book? >> probably the interactive fiction section and the second interlude. that's a personal level. was a way to give a certain
pleasure of my youth and thing with these games, working creatively, in that regard. that and writing about my daughter was great, too. >> if you have a question. [laughter] >> more of a comment. i been following your work for a long time. it's a real pleasure to read a book by you. thank you. >> thank you for reading a book like mine. [laughter] i had -- >> you said the problem today isn't the fact that there are
not enough facts we are being told lies but there are too many facts and that i think it's, no one understands the way for that data curated. that's not a question but our to say, that meant a lot to me when i read that. i think a lot of people had an education behind the fact that data in so many things are curated, i think surely that would be a good place to begin. i don't know. >> thank you. i think so much of today's the same political situation, rises from the ability smaller self organizing networks of people, they cannot curate their own narratives and not have to be joined together by a single and
bc nightly news or something like that. we have such an abundance and data and we now produce more data every day then what was computer data, the big text than words produced in the entire history of humanity up until 2000. you can find whatever you need. i think we are only beginning to understand what effect that is having. if you are a creative person, you can probably find whatever it is you are about to do, it's already been done. you have isolated scenes producing things in the middle of nowhere because it's harder to see isolated. people ask, where is the next funkadelic or something like
that? you always sympathize when someone, here is the authentic artist or something. i think part of that is because you have so much issues from, it almost seems like you can have computer-generated because there's 1 billion settings and you put it through a computer and say, keep generating, keep coming out with whatever it is you think is missing. there's an old comics to actually, that was a guy, goes three computer. we don't need pop artist anymore. check it out. put on a pop song and he says, i don't know, it's okay but it lacks so -- so. he says sorry about that, he flipped a label that said, still. [laughter]
>> excavated monkeys. >> so much of his evolutionary heart rhythms which is exactly that you combine until you get close to it. so much more is simply finding parameters between, other than ticking. basically given that we can run through these possibilities, if you can isolate the right categories and label they want to rate through, computers do work for you. you become that much more important to question labels in the stretch of liquid the way that it does. that's why james was so appealing that, you have these stretching's and whenever you want to call it that put ideas together in new ways rather than
accepted ways. certainly, by the suddenly everyone start using the word spoke a few years ago? i would anyone in the right mind is, given how fast i don't know, this is what happened. giving we contracting's, we're facing this increasing tension between free on the individual level predictable and boring significant levels. thank you. >> there was also a pathetic project. >> it was. >> it was one man said that it. >> i think that's the most interesting thing. it's not, he's the opposite of that. his genius was to say, i'm going to create my own scheme. i'm going to take every scheme
that's been created and pile them on top of one another. >> run through the machine. >> right. similarly created and it's an option. in a different way but yes, i think i haven't really connected with the romantics. i think, i read it and it's beautiful and all that but it is a singular vision. that's one vision, but what else that's a good one but -- [applause] >> you want multiple. >> you made a decision to include code in this book. we did this one republished, it would include code. i think this part of me that i kept thinking of this, there's this terribly tradition of the published was used in colleges and i was at columbia and it had
this blurb on the back, this is a wonderful translation. it's attractive. but you would think, you would understand that -- it is there for decoration. [laughter] was like, there's a code in the book, it picks up the text. when i just reading a block. >> much to the dismay, he was -- >> you want to put that in, i think -- >> you can push for it. >> we talked about this and agreed that it was, we wanted to because you're looking -- let's
break it up. we don't have enough ads. break up the text. it was kind of, in the margins attractive. i wonder if you want to do, how do you accept programs, you want them to keep to themselves -- >> you are too included. there is. >> it's different. i guess, it's in over writing reason for it. i guess it's because, to show in some ways how simple it is, it may look in code, the weird symbols throughout the if you look at it, often it's just either a set of constructions, it may not be clear what the instructions do but it's just same, despite a window. or it may just be, a mist of
things that, this is how we represent them. assign them numbers and 15th. to say that, when i look at papers and icy the greek symbols, there's a certain, this is going to take a while. i had to do that and i was like, would be nice if i could read this fluently but i hope to do is presented in say, even if it looks daunting, hopefully it will have a reasonable example. that show you at root, there's simple ideas and play. they get elaborated on but a lot of computer science does revolve around taking simple tools and putting them in complex structures. simple tools can clue you in and
also because i think i find the nature of separated abstract levels to be beautiful. that's what computers are. separate the abstract levels so that once you build up a certain level of complexity, you can work it from there. language is not there. [laughter] it's amazing and also maddening. i think that's what i struggle with, dry struggled with, we all struggle with. kids struggle with. our younger daughter, iris, was starting to say that she's pointed to food and said, too much. too much. we figured out that what she decided she got to watch meant,
a really good amount that i would like to have. that's too much. [laughter] more too much, more to much. [laughter] that's the opposite of what computers do. you don't work with large phrases that you breakdown tried to figure out the grammar. i think it's more logical. i don't think this is been too much. but more too much. more too much. we can lead you wanting more. thank you all for coming. [applause] double thanks to -- thank you.
>> thank you for joining us. >> thank you, that was really fun. >> steve and sarah purchase -- participated on a discussion about president trump. it's the fourth most-watched program of 2018 on booktv. see the full program and all of the most-watched books events at booktv.org. c-span launched booktv 20 years ago on c-span2. since then, we've covered more than 15000 authors. spanning more than 1000 weekends.
in 2004, john stuart was at the new york public library to talk about america, the book. >> i was looking forward to this to dispel some of the mythology surrounding myself in founders particularly the myth of our infallibility. moderns have a tendency to worship the author of the fathers. we will die to protect the second amendment. so dramatic. we know why we call them amendments? they amend. if asked mistake or correct omissions and made themselves can be changed. for the constitution to be written in stone, we would have written it in stone. [laughter] most things were written back in stone back then. i'm not trying to be difficult. it's bothersome when you blink your own influx ability and extremity on us. not that we weren't awesome, wrote it in the time that it takes them which is the a button in the name. we weren't god.
we were men. we had flaws. >> you can watch this and all the programs from the past 20 years, at booktv.org. the author's name and the word book in the search bar. at the top of the page. ... >> plan survived contact with the enemy, and sometimes no book signing survives contact with your university bookstore. [laughter] it turns out that we are without books today. however -- [laughter] and i'm very sorry about this. an unprecedented issue here at the national church and library certain, but