Darwin Grosse: All right! Today I get a chance to have a discussion with somebody who was introduced to me by my friend Tom Hall. His name's William Fields, and he has quite a remarkable music practice. He came to Tom's notice because Tom had been at the Algorithmic Arts conference in San Francisco a while back and was blown away by William's work, and he introduced it to me and I have been equally blown away, so I'm very anxious to talk to them. So if no further ado, let's have a chat with William. Hey William. How's it going? Great, great.
William Fields: Thanks for having me. It's an honor to be on here.
Darwin: Oh, it's an honor to have you. Thank you first of all for sharing some of your work with me so that I could dive in. I ended up diving in pretty deep because first of all, there's a lot there and secondly, it's really compelling. For listeners who might not really know your work, why don't you explain a little bit about [it]]?
William: Well, at the moment there's two sides to it. So I do a lot with algorithmic composition and generative music on the one hand, which really culminated in this project that I did called FieldsOS that you're referring to. And that music is all 100% generative. So it's completely hands-off. I program my music system to generate the music based on certain rules and I press record and I let it go. So that's the one end of the spectrum. And then on the other end of the spectrum is the music that I've released on various labels and that I do in a live performance setting, which is more improvisational. So, and in that case, it's very hands on. I have a controller where I'm controlling all aspects of the music in the moment.
And the music that I release is all just a live recording of these performances, basically. So I don't work with in a DAW and I haven't touched a piano roll in probably 15 years, or, you know, drawing envelopes or anything like that. I basically just sit down and play and record the results and the good ones that are the ones that I release basically. So it's two extreme ends of the spectrum. On the one hand I'm doing music that's very hands-on and improvisational. And on the other extreme it's completely hands off and algorithmic.
Darwin: So, yeah, that's really interesting because not only is it two ends of the spectrum in terms of it being how you describe your work, but it's two completely different versions of the work process as well, right?
William: Yeah, I think so. And you know, I think a lot of it comes out of the fact that I don't have a whole lot of time to spend on music and I'm a bit lazy. So I wanna make things as fast and easy and fun as possible. But on the other hand, there is some overlap in that in my live performance practice, on that end of the spectrum, I am also using these algorithms to dialogue with in some ways. So I can press a button and it'll throw me into a kind of random musical space. And then I improvise based on that. So, you know it, this is thrilling in a real life performance up on stage when you're about to push a button and you don't know what it's going to do, right.
And you just have to work with whatever it gives you. So that's kind of what like what I did at the performance at Algorithmic Arts Assembly in San Francisco that you were talking about. I hop on stage and basically I'm, I'm hitting a button that it just throws me somewhere random and then I'm steering it and working with it and then trying to make it into a musical arc or narrative. And then for I'll push the button again and it'll throw me somewhere else random. And then I work with that. So there is some overlap between the two.
Darwin: Sure. But that also seems like, if you don't mind me saying, that seems like sort of a crack-headed approach to things because - in my experience - a lot of people when they do utilize computers and algorithms in live performance, it's to provide a safety net for their work so that their live work is still within some parameters that that provide them a sense of ease. It seems to me like you are literally putting stumbling blocks on yourself and then part of the performance is like watching you sort of like survive that, that process.
William: Yeah. Well it makes it more exciting and hopefully, you know, for the audience too. But to be fair, the algorithms are not completely random. I use a tuned randomness. I mean, I do have the option of going completely random and usually that doesn't sound too good and it's hard to dig your way out of that. But I do have different buttons that have different algorithms that are tuned in different ways, so I can have some degree of control of what's going out. Like I might have a button that that generates a footwork type track. And so I know that I'm going to get something that's high energy, that has triplets or certain things, so I kind of know what's going on. But yeah, it, it makes it exciting. And, also, I'm a big fan of jazz, so that's also an influence where you're improvising in the moment. You're working with what you're given and trying to make something good out of it.
Darwin: Right. Well, it is interesting though, this idea of tuning algorithms to be useful. You're doing it in your live work, but that also is the concept behind the FieldsOS work now. That was originally a radio type show. Right?
William: Right. So there was a call for entries that went out for submissions to the radio station Resonance Extra. They were looking for radio shows that did something different, that were not a typical DJ playing records type of thing. And I had this system that I developed, it could generate endless amounts of music and I was kind of looking for an outlet for that. Like, how can I take advantage of this system, this abundance that the system has to generate so much music. And so I came across this thing for Residents Extra and I put together a proposal, submitted it and it was accepted. And so yeah, it was a weekly radio show where each week, I released an hour of new music that was generated by my music system. Each week was basically a different genre or some kind of algorithm or approach.
And I would set up the system, press record, and then press play. And then the system would run for an hour. And that would be the radio show basically. I would do some, you know, talk over and do some introductions so forth. And I of course I had to sometimes do multiple takes if the first couple of tracks were stinkers or something, I would start again. But, but I did not edit really. I mean, if there is a bad note or something that was harsh or difficult in the middle of it, I would just leave it in. I was not curating the results at all.
Darwin: Really? That's really interesting. Now I'll also say that there's some intrigue just in looking through the titles. So the FieldsOS, file set is 24 tracks and it starts off pretty understandable: Electro, Hip Hop, techno. But eventually you get into Breeding Music or Brownian Techno or Unconstrained, which I was one that I actually liked. My favorite was number 20, called Variety, which I wasn't sure what I was going to get into. And I got into it and I just found it a really charming, a sort of mish-mash and very eclectic sounding thing. But what's amazing to me is that while generative, it didn't have what I had come to think of as generative music. This would have one of two aspects: either it would be very random or it would be hyper-predictable.
And especially as people would tend to try and make it sound like a particular style of music, it would tend to get more and more predictable - almost like, and I have to admit a bias here... If I imagined what "Machine Learning Techno" would be like, I was scared it would be that. And it was very much not, it was very much alive. It kind of breathed, it was transforming in a way that moves it forward. And I found it really interesting.
William: Thanks. Yeah, I mean there's a lot of probability involved. So at the core of it is a probabilistic sequencer. And so it's not going to repeat the same thing. Every bar, there's some subtle randomness and there's this kind of a scripting to it. For example, I'll program the system so that it'll generate something, play that for, I don't know, four or eight bars, and then it'll generate a variation on that where 10% of the parameters are different. Then that sounds like the next section of the song - you can tell it's still the same song because it's mostly the same, but 10% of it is different. So it sounds like, "Oh, this is the B section." And then I might jump back to the A section and then after eight more bars that I might create a variation on the variation. So it's again, it sounds somewhat familiar but a little bit different. And so through that I was able to create song structure and variation on the original. It's not like I'm generating something and then it just sits there and runs for whatever, 32 bars or something. It's generating variations within that structure.
Darwin: Sure. That makes a lot of sense. And when you say it that way, I can kind of hear it in my mind and how that's working. So one of the things I like doing on my podcast is talking to people about their background and how they got to be the artists that they are. And in your case, it seems like there's gotta be some really interesting background, because to come to things and have the kind analytical mind it takes to come up with algorithms to do these sort of genre-oriented works comes from one area. And, and conversely, the kinda chutzpah it takes to say, "Okay, I'm going to play live and have stuff blurt stuff out and be able to dance on a wire to make that happen" is in a different space. And I'm curious to know where you came from.
William: Yeah. I've found telling the story is a good way to explain how my music system and FieldsOS works, also. So I guess the beginning for me was when I got a Commodore 64 as a kid and I got really hooked on computers and programming. I would get some magazines, some programming. It was Commodore 64 magazine or something like that. They wouldhave code in the back of the magazine and you'd have to type it in, in order to run the program or the game or whatever. And, that's when I learned that if a single character is out of place, that's not gonna work.
So I got hooked on computers and programming early. it just resonated with my brain is just... I just love it naturally. And then I got into bulletin board systems, so I'm dating myself here. This is, before the internet really. And there was... some of the listeners might be familiar with [an artform], it's called ANSI Art. So I would make this artwork for the bulletin board systems where you'd have to dial in with a modem and you would connect and then you'd see this intro screen, very blocky, pixelated kind of artwork. I got into that scene of making ANSI Art, and then from there I discovered there was a way this music thing that people were doing with trackers - they're called Mod Trackers.
And that's when I first got into the music side of things. And I got really hooked on that. You know, people would create these mod files and upload them to the BBS is and share them and stuff like that. You could make your own samples, it was fantastic. And I actually still have some of the old mod files, believe it or not. And the earliest ones I have are from 92-93, iwhen I first started making electronic music on trackers. And so then things progressed. I made music on Cakewalk and FL Studio, regular dDAW type programs. And for a long time I made music and this of traditional way of using a DAW: drawing in notes on the piano roll and slaving over a track for hours and hours and hours, weeks and weeks, slaving over the mixdowns and driving myself crazy.
You know, I was trying to get it to sound good. And, so for many years I worked in that way. It took a lot of time and was challenging and not always fun. And then I had kids and that changed things a lot because I just didn't have the time anymore. I was also working a full time job and, for a while I just wasn't really making music much - just because I didn't have the time. But then I started playing music with a friend of mine, Jeremy Beck, and we still make music together and release it as a Beck/Fields and he's a amazing singer, musician, pianist, classically-trained. He's incredible. But we started just improvising together. And at first I just had a very simple... I was just running him through effects, basically.
And then I started to develop that system more and more. I got a controller (to make it easier). I ended up Lemur, that iPad app now is my main controller, and that makes it really easy to change things around and evolve the system over time, you know? And so then at that point, I'm improvising a lot with my friend Jeremy, and I have this music system. But over time I develop it more and more to the point where I know I can make beats in it. And now I have since developed this step sequencer thing and it gets more and more sophisticated over time. And then I started performing, making music solo with it. As I've mentioned earlier, doing this kind of improvisational thing where there was no algorithmic stuff involved at this point.
So it was just me sitting down with my music system and making something, which is lots of fun. And this system, I'm a strong believer in iteration and practice. So, my playing music with my friend Jeremy is really valuable in that we would sit down and we'd make music, record the results. I go home the next day and listen to it in my headphones and take notes, basically like, "Okay, this didn't sound good. I need to adjust this or I want to be able to do this. I want to be able to do that." And then I'd make adjustments to my system and then we'd go play again and then repeat over and over and over and over again. And that allowed me to evolve my music system.
So it became more and more flexible and, and sounding better. So now I have this music performance system which is really fun. But I feel like I'm always falling into the same kind of rut with it. I have certain musical habits, like: I always end up putting a bass drum on the one and snares on the three or whatever. And so I realize, "Ah, okay, I have these controls I can; with this Lemur thing, I can randomize the controls, I can make a button. If I push it, it will just throw all my controls into a random position. So I did that. That was the first step towards the generative and it's really fun. you end up finding these musical spaces that you would never find on your own.
Because we just, we're human beings. We have certain habits. We, we tend to do the same kinds of things, but this just kinda just throws it into some random place and it's really fun. It's like a slot machine or something you just push. I could sit there and push the button over and over again and see what comes out. So that was great fun. But as I mentioned before, pure random does usually doesn't sound too good. It's too chaotic. There's not enough structure to it. So, from there I started to tune the randomness. I'd make certain things more likely, certain things, less likely, certain things I don't want to actually randomize at all. And then I started to get better results. I get more interesting, more musical results and that, that that was working really well.
But then I realized I could actually make different algorithms: you could imagine multiple random buttons where each one works in a slightly different way, or each one has the randomness tuned in a slightly different way. Just to take a simple example, like a techno button. So if I press the techno button, it's going to randomize a lot of the controls, but there's certain things it's not going to randomize. The tempo is always going to be 135 or whatever. There's always going to be kick drums in certain places. So you have a 4/4 rhythm, things like that that give it some structure. But within that there's still a lot of variation and randomness. And then I might have a, I don't know, a drum and bass button that structures the beat a little bit differently and so forth.
And so you can see how that could lead to this FieldsOS thing where I can basically develop different randomization algorithms - where the randomness is tuned in different ways to create these different genres. Or I want to work in different ways and I basically use that technology for the radio show. So I also created this kind of scripting language, where I could execute these randomization and algorithms on the beat, as I mentioned before, or create variations. It's really fun, really powerful because I basically have the state of the system is just a set of numbers. It's just data, but all values between zero and one. And when you have that, you can not only can you save the state of the system and recall it later, but you can also do fun things like you mentioned Breeding Music, where I would generate two different compositions or songs, whatever you want to call them, two different States of the system.
I would play the first one as the first parent, so to speak. And then I'd play the second parent. And then I would breed those two together to make a child. And then you would hear the child. And if you listen, if you pay close attention, you can really hear the similarities that you can hear a little bit of both parents in the child. So it's really fun, once you have everything stored in this way as data, you can do all kinds of fun things with it. And, as you can tell, I did end up going into computer science and programming, and I work in it for my job. So this kind of coding stuff is second nature at this point. If I have an idea, usually I'm able to realize it because of that.
Darwin: I have like 900 questions, so I have to pull my shit together here to be able to ask them sensibly. Before we get into the algorithms and stuff, you talk about your music system; what is it? I mean, is it a coding thing, like Tidal[Cycles], is it a patching system? What is the music system that you base this off of?
William: So it's kind of this Rube Goldberg machine. People have asked me if I am going to open source it or share it in some way, but it's really hard because it's this patched together collection of stuff. Basically I have a controller, as I mentioned, is Lemur, running on an iPad. I'm actually using JavaScript which is running in Google Chrome as this sort of data processing layer, so to speak. So that's what's generating the randomness now. Originally I did the randomness inside Lemur itself, but it's kind of limiting so I moved it out. So I have the control layers Lemur, this kind of data processing and generation control data layer, which is a JavaScript writing inside of Chrome. And then that in turn, is sending control data to a Reaper, which is my main audio and sound generation engine.
So it's Reapers. I love Reaper. It's really stable and customizable and it has its own programming language built into it. I have some custom plugins, mostly MIDI processing plugins that I've written inside Reaper, that are also involved. So it's just evolved that way because, as I explained in the beginning, in my history it started out as just Reaper and then I added Lemur as a controller and then I said, "Oh, well, it would be nice if I had this layer in between where I could manipulate and control the control data." So that's how Chrome got involved. I also do visuals stuff, responsive visuals, but that's another reason why I use Chrome in the middle. So it's a collection of technologies that are talking to each other, but I do everything in the box. It's all software. I don't even own a keyboard or a synth at all.
Darwin: It's really interesting, though, that a lot of your work is done in JavaScript on Chrome. I mean, first of all, when you say that you have reactive visuals, are the visuals just like WebGL or something in another Chrome window? How are you doing visuals off of this?
William: Yeah, and you can, if you check on my Instagram or Twitter, you can see what they look like. I do have a couple of different approaches that I'm working with, but I mean I'm using Three.js, which is a JavaScript library for 3-D graphics and that's running inside Chrome. And so I'm using Web MIDI to get the MIDI data coming into Chrome and then I'm generating visual events, objects, motion, things like that based on the MIDI data that's coming in from Reaper. So it's very responsive. It's an area that I... the music is primary for me, but I'm also very interested in audio-visual music and creating visuals that are very tightly correlated to the sound. That [way], for example, you can see every single high hat separately, every single drum hits separately. You can, when the pitch goes up and down, you can see that visually so that it's not like an FFT audio analysis where you kind of can figure out when there's a bass drum or something like that. It's very, very precise. That's another aspect of, of what I do.
Darwin: That's amazing. And I can see where, in comparison to some of the bespoke musical tools, if you're comfortable working with JavaScript, there's a lot of things there, especially in terms of multiple state saving and stuff like that, that would be really conducive to using JavaScript as the tooling.
William: Yeah. I mean, I love it. It's a kind of a weird language and it's always changing, these days. But I really like it. And as you said, there's a lot of stuff out there. I also use P5.js, which is - there's this language called Processing for doing visual stuff, and P5.js is the JavaScript version of that. So I do also use that sometimes. There's a lot out there that you can [use], a lot of libraries and stuff that you can just download and plug in and start using and yeah, I did have a period in my life where actually the first version of my kind of live system was written in PD. So I did have a period where I worked on these kind of control or flow data flow languages, but I just found that I'm more text-based. Programming is more natural for me. I'm able to do what I want easier that way. So this is kind of a personal preference.
Darwin: Well, I that certainly connects though with your history, too. I mean coming from typing in code listings in the back of the magazine, it seems like code - creative or not - is going to look like text anyway.
William: Yeah. At least for me, that's, that's what works the best. Yeah.
Darwin: Now you talk about sthis iteration process and it's really fascinating to imagine the history of these things being developed. Starting off pure random and just being giddy to hear the effect of that. But then eventually having a desire to pull it into a more musical direction. But it makes me wonder to what extent did you have to take a break and say, you know, I really need to study, what is it that makes Hip Hop [be] Hip Hop, or what makes a Footwork track work at a really elemental level? What was the process of breaking down the music? Or did you just randomly do stuff and be like, "Oh that's kind of footwork-ish. I'm going to put a button relates to that!"?
William: It was an adventure every week based, because this was a weekly radio show, so I had time pressure, right? And I'm just doing this on the weekends. And so I'm just...
Darwin: So this stuff wasn't, this stuff wasn't precooked ahead of time? I mean, you didn't already have the UK garage button set up?
William: Oh, I'd have to sit down and code it out and figure it out. I mean, I've also been listening to electronic music since, the mid-90's or something like that. So I have a lot of experience data in my head and I analyze it and think about it a lot. So I don't know if I always really hit the mark, I think it's always my interpretation of a certain genre and I'm not sure a purist in any one of those genres would listen and be like, "No way! This isn't real UK Garage!" or whatever. But, you know, I would try to just listen to the genre and analyze and think about, okay, what's going on here? And, like with, UK garage, that's a ton of swing.
It's at a certain tempo. There's a certain beat structure to it, things like that. So yeah, I mean each, each week I'd just have to sit down and figure it out. And sometimes it was a challenge to come up with something, you know? And that's when I started to do these more unusual things like the Breeding Music or Brownian Techno, which is a funny one. So, Brownian is a reference to this kind of random noise, right? So basically that one starts out with a kind of a techno type thing for maybe eight bars. And then it basically starts to slowly fall apart because all of the parameters start to drift in random directions. So you hear it the techno track will just slowly break down and fall apart until it's like, just struggling and it ends up sounding like free jazz or something.
And then the whole cycle starts over again. That's actually one of my favorites. It's fun to listen to. But it was a really interesting project and creative challenge to come up with something each week.
Darwin: Sure. I'll bet that that, like you said, that had to push you pretty hard in a pretty cool way. So I'm curious as you started working with this, I mean there must have been surprises for you along the way: finding that certain kinds of parameter changes or certain kinds of variations surprised you as being artistically valuable or maybe artistically useless or, or even anti-valuable. What are, what are some things you found that either were pleasant surprises or kind of real disappointments?
William: Yeah, I mean, I guess one example I could point out is that for each kind of instruments, or each sonic element, I have two different envelopes on it. So, for example, for the bass drum, snare drum, the bass synthesizer, whatever, each one of those has its own filter envelope and a pitch envelope. This allows for a lot of variation. Yeah, so there's the starting [of] the envelopes, very simple. It's just the starting point, the ending point and how fast does it go to get from point A to point B. That allows you to have just a simple element, but it can have a lot of variation by just using those two envelopes. One thing that I really have to restrict often is just the pitch bending - the drums are fine. It's fine if those get pitched in strange ways, but if I want it to be actually musical or interesting - musically coherent - than I have to make sure that the pitched elements are not getting re-pitched or pitch bent too much.
Because then it all sounds crazy and microtonal and weird sounding like algorithmic music. And I think in the Unconstrained episode and maybe the Ambient episode, I do let that run free a bit and you get some kind of weird microtonal sounding stuff. So sometimes it can be interesting, but that's one thing that we humans like: we'd things to be pitched and harmonious in certain ways most of the time.
Darwin: Now in going through the FieldsOS stuff, one of the things I notice is that generally you would pick a tempo and stick with it. Is thatvbecause tempo variations were kind of hard to manage within these variation models?
William: No, no, it's, it's not really a problem. I think it's more [that]] I saw a lot of times tempo is a big part of the genre. So techno, of course, is at a certain, usually at a certain tempo, and a Drum & Bass or Hip Hop. So that, to me, was just a core part of the genre. I mean, again, it's like the Unconstrained experimental type of episodes. The tempo did jump around a bit more.
Darwin: And then, you talk about having these sounds, having different envelopes and stuff like that. The sounds themselves, are those held in tracks in Reaper? Or where does the sound exist where you can sort of edit it? And do you have different sound sets that you use for different kinds of stylistic purposes?
William: So I'm using all software synthesizers, so it's mostly synthesized. There's no samples or loops involved? Well there is some samples in the drums for the, the transients when I want to get a certain quality to the snare or the bass drum or something like that. In those cases, they're just single shot samples and it's encapsulated in a soft synth in some way. And for the most part I kept the synths and the samples, kind of the sound generation engine part of it the same through through FieldsOS episodes. I think I made some some changes in the Hip Hop episode - I use a different snare cause I wanted a certain quality or I might make a few changes here and there. But for the most part, it's the same sound engine and yeah, it's all all synthesized.
My main synth is called Surge. And it's an older soft synth. I actually bought it when it was a commercial product, but then it went defunct for years and I just loved it and I was so used to it and I just continued to use it. But then a few months ago, I discovered that it was open sourced and it's alive again and they're developing it. So I recommend checking that out. It was developed by one of the guys that went on to found Bitwig. I think Vember Audio was the name of his company. But yeah, he went on to do Bitwig. And so, this got abandoned, but now it's happily alive again and thriving. So I recommend checking it out - it's a great synth.
Darwin: Is that surge like S-U-R-G-E? OK, cool. That sounds really interesting. I had never even heard of it, so that gives me something sniff around abpit. Now. I noticed that, recently I notice that you're going to be coming to my neck of the woods. I live near Minneapolis and I see that you're going to be playing at a festival at the beginning of December called In Situ. When you go on, "On The Road" to do these kinds of performances, do you bring anything other than your laptop and Lemur on the iPad? Is that it? Do you keep that tight of a system or do you bring peripherals just to make it feel more at home?
William: Well, no, I mean I'm pretty much a minimalist. So it's just my laptop and my iPad for Lemur and my audio interface, that's pretty much it. Unless I'm doing visuals and I need to bring a projector as well. But I really like keeping it simple from that perspective and minimal. And you know, I'm not obviously not into modular or analog or anything. I'm not a purist in that sense. Maybe my ears aren't good enough - I don't hear the difference, but it's just so convenient and so flexible to do everything in software. So I like to work that way.
Darwin: Sure. Now it's interesting that you talk about being not being a purist from the standpoint of analogs or modulars, but what I do notice is that you find yourself in a lot of festivals with people who are live coder folks - for whom their performance, a real specific part of the purest part of it is to be projecting their code on screen while they're developing it, all that stuff. I mean, it, it's funny that while the coding part of it is a part of your practice, that that's also not a thing that you're a purist about either.
William: Yeah. I mean, I have fallen into this community, which is wonderful. It's a great community. But I do also feel a little bit strange about it because yeah, I don't do live coding. Obviously algorithms are a big part of what I do and programming is a huge part of what I do, but my approach is different. I think of it as more of a kind of gestural approach to performance. Sure, I'm using a controller and it is to me live coding. I mean, I don't have much experience with it and some people are absolutely wizards with it. You're Kindohm, Renick Bell. These guys are amazing - what they're able to do. But to me it's, it feels like you're trying to build your instrument on stage in front of audience and play it at the same time.
And to me, I would rather have my instrument, practice it a lot, get really good at it, and then, bring my instrument on stage and play it. It's just a little bit different mindset.
Darwin: Sure, that makes a lot of sense. Well, William, I can't believe our time's up already. We burned through that really quickly. But before I let you go, why don't you give people a few hints about where to go to experience some of this music, and to check out some of the FieldsOS stuff. Or maybe to get a chance to hear or see some of your live performance work or some of the live visuals. You mentioned Instagram and Twitter have some stuff. What's your handle over there?
William: Oh, it's "williamfieldsy". So William Fields with a Y at the end. I on both Instagram and Twitter. I don't do Facebook for my artist stuff. I have WilliamFields.com, there're links to everything there. And but I'm most active on Twitter, @williamfieldsy and on Instagram also @williamfieldsy. And I have... FieldsOS you could check cows on Bandcamp. (WilliamFields.bandcamp.com). And I've got this festival as you mentioned in Minneapolis. I'm a little scared of the weather.
Darwin: It's December isn't now. It doesn't get bad yet. It's worse in February - that's when you'll cry.
William: But yeah, I'll be performing and giving a talk there on augmented creativity. So that should be fun. And I have a release coming out soon, hopefully in the next month or two on Conditional Records called Shackamaxon, and I've got an amazing set of remixes from some incredible artists that are going to be on that as well. So keep an eye out out for that too. But yeah, feel free to get in touch, if you have any questions, send me an email or on Twitter or anything, I'd be happy to talk.
Darwin: William, I want to thank you so much for taking the time to have this chat. It was super revealing. It actually makes me want to start working on my own algorithms because what I see as you talk about this iterative thing is like, well, "You're never going iterate if you don't get started!" Well, thank you so much for the time. It was fabulous to talk to you. Bye!
Copyright 2019 by Darwin Grosse. All right reserved.