Can tweets be considered poetry? And can robots write poems? A conservative reader of poetry might baulk at the idea of answering these questions in the positive. But we are not conservative. Nor, for that matter, is poet, performer and activist Harry Giles. Over the last few years, Harry has ‘cultivated’ a whole ‘garden’ of poetic, political and poetico-political Twitter bots. Here they talk about the joys and frustrations of the medium, how they got involved in the Twitter bot scene, and how we can too.
C: Harry, thank you for talking to All These New Relations this afternoon on the subject of Twitter bots or wherever the conversation may take us. So let’s start with your journey into bot making. What was it initially that drew you to making Twitter bots?
H: It’s genuinely just seeing other people’s Twitter bots and thinking, ‘I want to do that’. I think that’s a problem I’ve had throughout my life – I’ve never seen an artform that I didn’t want to try. Sometimes I manage to restrain myself. But I’ve always been a dork. I was never quite fully the sort of child computer programmer type person, I never got that far, but I’ve always been playing computer games. I grew up on the internet as a teenager, I lived in a tiny Scottish island, so the internet was my social life, so I’ve always kind of been in tune with that, with that world. And when I was younger, when I was a student, I made an interactive fiction game, so I used to make a lot of text adventures essentially, so yes, on literature, on digital literatures, I’ve just been interested since I was writing I suppose, and Twitter bots is just a new form of that. But I started seeing the wonderful, creative and very poetic Twitter bots, and seeing the potential of the medium, so I started teaching myself how to do it.
C: Cool. And you see it as a kind of way of bringing together a sort of poetic imagination with a, in your own words, ‘dorky’ tradition? [laughs]
H: Hmm, that’s interesting. I suppose it’s with dorky interest, but there’s been a huge shift in my lifetime, of like when I was a teenager, socialising on the internet was weird, do you remember when it was weird? I met people from forums and people were worried I was going to get stabbed. And now everyone socialises on the internet, and that’s a huge change in my lifetime, so actually it’s not dorky to be on Twitter, it’s like it’s not even a strange cultural experience, it’s a very mainstream cultural experience to be on Twitter, so to make a poet-like – the skills of making Twitter bots are dorky skills, but the aesthetic isn’t dorky – it’s actually very populist. Like it’s the most populist poetry I do, I think, because it’s taking place in a massive popular social space and it really needed that space.
C: I wonder if – because to my mind there’s something, that maybe with that shift from the internet being a subculture to the culture, it seems they’ve lost some of the magic, or the innocence – because I had a similar experience and I remember just, being into really weird, niche genres of music that nobody in my town liked but all of a sudden there were all these people you could share it with. Whereas now I suppose because it’s so ubiquitous, you know it loses some of that, I don’t know… Do you find that the Twitter bot as poetic medium is a way of keeping some of that magic, or that kind of countercultural energy?
H: I think you’re right – I think you’re right to put your finger on that, and I think that’s partly as a result of the people who are involved in it, because a lot of them are people like me who miss some of the early internet stuff. The artistic Twitter bot community tends to have a very ambivalent relationship towards Twitter as a social space, and as a platform. I mean I think lots of Twitter users… Twitter users hate Twitter a lot more that Facebook users hate Facebook. Like Twitter users hate Twitter. And that’s part of being on Twitter, which makes it very contemporary. So I think a lot of Twitter bots are there to inject beauty, calmness, weirdness, strangeness, magic into a social space that’s often, like, really frightening, abusive…
C: Polarised in a lot of ways.
H: Polarised… Tedious, banal, all corporate, blah blah blah blah. So there is that. And the thing about this dorky thing again – because the other thread that leads into Twitter bots is the history of generative literature, procedurally generated literature, pre-digital – like it’s really old – there’s an amazing essay, there’s one by Holly Gramzio, there’s another by Katie Rose Pipkin I think it is, who’ve both done genealogies of generative literature that go back hundreds of years, and both of them pick entirely different examples and give hundreds of years of genealogies of generative literature, it’s really interesting.
I think a lot of Twitter bots are there to inject beauty, calmness, weirdness, strangeness, magic into a social space that’s often, like, really frightening, abusive…
C: Wow – I don’t know those essays, I’ll have to check them out.
H: Yeah yeah yeah, well if you remind me I’ll find them and send you links. Because there’s things like poem-squares, code-squares, poem-wheels, and like coming into the modern era, stuff like Raymond Queneau’s…
C: Is it ‘A Hundred Million Billion Sonnets’ [Cent mille milliards de poèmes] or whatever…
H: Yeah, yeah.
C: He invented the ‘Choose Your Own Adventure’ as well.
H: The what?
C: Choose your own adventure story, at least as far as I know.
C: But then perhaps there’s a whole other genealogy out there.
H: Yeah, that’s another form. So yeah, threading into there, there’s generative literature, there’s gamebooks, RPGs, that feeds into that cultural stream of Twitter bots, there’s hypertext literature, there’s uncreative writing like Kenneth Goldsmith’s, all of that feeds into Twitter bots.
C: So do you see your work as quite consciously in that tradition?
H: For me, yeah, those traditions, yeah.
C: That also raises something that I wanted to ask you about, in that you use games as a model quite a lot in your work, both online and live work and so on. I wonder if you could say a little bit about how you see the relationship between games and poetry and how Twitter bots play into that?
H: Mhm. There’s two ways into that – what’s the connection between games and poetry – and the easy and simple route is just to say that for me poetry is play in the space of language. And play can be very undirected, like when we look at children playing, sometimes they’re following sets of rules to make something happen. But the other kind of play is just being in a space with lots of possibilities and stimulants and, like, moving around a bit, and most poetry is that kind of play.
C: I agree, and it goes back to that sort of, ambiguities and polysemy of language, it gives it that space for wordplay.
H: There is a difference between play and game. So I could say that play is central to my work, but actually – you’re right – it’s games, because what games are are like unnec- – I can’t remember what the definition is, but, like, unnecessary sets of rules voluntarily followed, that’s kind of what a game is, it’s voluntarily following a completely arbitrary pointless set of rules.
C: In a way Twitter does that as well, giving you 140 characters.
H: Exactly, exactly. So Twitter has – it has more rules than that. It has social rules, and social conventions, as you can win points essentially. Well you can literally – there’s faves and RTs, you literally win points, and you know the kind of things that will win points, and deliberately tweet to win points.
C: That’s like Charlie Brooker in his rundown of the top 25 games ever or whatever, and I remember I was watching it and I was like oh there’s Monkey Island and there’s Doom, now what the hell’s going to be number one and then he said Twitter. I initially thought well that’s a bit of a cop out, but actually the argument he put forward was pretty good, it’s that ‘gameification’ or whatever, that point scoring.
H: Yes – but also, formal poetry is a kind of game, it’s a language game. And given a set of constraints and you have to find – it’s like half a crossword, it’s like half a puzzle, half a game. But then also looking more deeply at the structure of it I would say, not in the Wittgenstein sense, but that language is a game. And it’s also social conventions of games, like there are these sets of rules that we follow like languages, and there’s different sets of language you can play, different registers, different accents, different dialects that we speak in, and a game is both a means of exposing the rules that are there, and also pushing out and testing those rules. Like, whenever somebody – well, most people – when they play games look for ways to break rules, and get around rules, and trick the rules – and that’s what games are. It’s not following them, it’s like how can I break through this system in this system, and that’s the approach I take a lot of the time to writing poetry.
C: It makes me think of the Dadaist poem, you know Tristan Tzara’s instructions for a Dadaist poem, cut up a newspaper and blah blah blah, and there’s one word in that, in the instructions (which kind of looks like a poem itself with the line-breaks and so on), there’s one of the instructions which is ‘Copy conscientiously’, and it’s that word ‘conscientiously’ that I think the secret of the poem is in because it’s the only thing in the instructions that a robot can’t do, you know, there’s no conscience….
H: Well… [laughter]
whenever somebody – well, most people – when they play games look for ways to break rules, and get around rules, and trick the rules – and that’s what games are. It’s not following them, it’s like how can I break through this system in this system, and that’s the approach I take a lot of the time to writing poetry.
C: Go on!
H: Well I just – I like thinking through the agency of em… Okay, so there’s two ways we can think about the conscience of a robot, a bot, a Twitter bot – and something that characterises the artistic Twitter bot community actually, is that it’s highly conscientious. I mean we’ll talk about conscientious bots, they’ll talk about ethically good bots. So a lot of the prominent makers of Twitter bots will put language filters on their Twitter bots, so if they’re using, if they’re generating from a corpus of words they’ll make sure that certain words aren’t in there. A good rule of thumb is that you don’t make a bot that would tweet anything that you wouldn’t say, that’s a good rule of thumb. And people – Dairus Kazemi thinks through this a lot, Leonard Richardson has written on it well as well. And it’s really quite subtle thinking on how to do that, to the point that when – do you remember Microsoft’s Tay disaster, the Tay bot?
C: Yeah, where it turned into a racist Nazi in like five minutes…
H: Yeah yeah! And the artistic bot making – which is a quite, it’s interestingly, it overlaps but it’s a bit distinct from the corporate and functional bot making community – we saw that happening and we were like you complete numpties!, if you’d have had one conversation with any of the artists who make Twitter bots we could have told you what was going to happen, and told you how to fix it. Like it was a solved problem, so easy, we could have told them how to fix it, because we’d thought all this through, we’d developed the theory and the techniques, and they just didn’t think, they just didn’t think.
C: So they were perhaps playing a different game?
H: I think they hadn’t worked out what the rules of Twitter were. They hadn’t really thought through what the rules of Twitter were.
C: Interesting. So yeah, following on from that, this idea of agency, or responsibility I suppose. Do you ever – I mean your bots are pure in the sense that umm… Do you ever tweet as your bot?
H: Oh, umm…
C: I mean, do you ever break the generative rules and just, you know…
H: Mhm, I don’t think I do. No, I have one or two cyborgs, so there’s a couple of my propaganda bots, I have Anarcoo [@AnarchyCoo], which tweets anarchist slogans and pictures of highland coos, and a daily anti-fascism, which is a half-made bot I haven’t finished it yet. The ones that have a propaganda purpose, I’ll sometimes – if they’ve got followers – I’ll sometimes login them to just tweet something about a current issue, so in the bot world we tend to call those cyborg bots, ones that have human and generative in them.
C: I see, yeah.
BAEH. OUTREACH FAILS SOMETIMES, BUT SILENCE FAILS ALWAYS. REACH OUT. pic.twitter.com/E3jIveGL4n
— Anarcoo (@AnarchyCoo) May 21, 2017
H: Some of my bots are propaganda rather than poetry. And the agency thing is interesting – I mean the reason I paused with the ‘can a machine be conscientious’ is, a lot of us tend to talk about our bots as separate agential entities. Yes, we create them, yes, we write all the code or copy all the code. But each bot, you do think of them as having a life of their own, they do have a kind of life or mind of their own and they always surprise you. A well-made bot should make things that you weren’t expecting it to do. And because you made it, it produces a result which is alien and surprising to you, you think of it as… When my bot tweets something bad because I haven’t done the programming right, when it tweets something and I think oh god you shouldn’t say that, I don’t think – part of me is thinking I failed in the writing, but part of me is like bad bot!, like, I have to train you! And thinking about bots as having agency, that’s the language that I use.
C: And I suppose it’s – is there often a, do you try to make your bots as finished as possible before you make them go live, or does this, you know, nurturing process, to pick up on the parental metaphor that’s seeping into your discussion…
H: It differs from artist to artist, because all of my art practices I do develop in public – I’m always releasing drafts of things, like I did a thing last year where I made thirty bots in thirty days, and I think they cannot all be finished, so that’s quite explicitly…
C: Yeah, I wanted to ask you about that!
H: And I just haven’t got back to them for a while, but I need to do a lot of maintenance on them.
C: So how was that process?
H: It was thrilling, and exhausting, and it forced me – because I wasn’t satisfied with just churning out bots on the same model, I wanted to learn – it forced me to learn a lot of new programming skills, hard and fast. So my programming ability just doubled in capacity over the course of that month, which is really satisfying, but I also, similarly, just learnt a lot about the aesthetics of bots at the same time. It was a charge through, figuring out everything I could about the aesthetics of how bots worked, and it was kind of an interesting month to do it, because it was the month of the election of Donald Trump, and there was some really interesting interactions with the project around that. Because it was producing weirdness and beauty and occasionally bits of propaganda, because a lot of the bot makers are based in America, and most of them tend to be sort of progressively aligned – like, there are a lot of very sad traumatised scared people – and they were saying things about the bots, like oh God at least something good is happening on the internet right now! Because Twitter was a vile space, and it was churning out all this random generative poetry for a month! And anti-fascism. So that was nice.
C: So, you were talking about how the Twitter bot, is in a way a kind of antidote – maybe antidote is too strong a word…
H: Yeah, yeah it’s not an antidote. A respite!
C: A respite, yeah. From the tumult, from the view… Do you think more generally… Do you think Twitter bots are inherently… In terms of the form, do you think there’s a potentially disruptive way they are positioned in relation to, you know, post-truth, online discourse, blah blah blah…
Every technology and every artform can be used by anyone, could be turned towards any end, so what matters is also the social technology and the social expectations that you build around the artform.
H: Yes and no. I mean I wouldn’t say inherent. It’s… [sighs]. Every technology and every artform can be used by anyone, could be turned towards any end, so what matters is also the social technology and the social expectations that you build around the artform. And, as it happens, artistic Twitter bots are, for the most part – like, there’s a lot of sort of leftist or progressive activist Twitter bots, like there’s an incredible Twitter bot called Trump2Cash [@Trump2Cash], that uses Donald Trump’s tweets to predict the stock market, and donates profits to Planned Parenthood.
— Trump2Cash 💰 (@Trump2Cash) May 4, 2017
H: And it succeeds. Because when Donald Trump tweets about a company badly it falls in stock price, like that’s just what happens, and it’s turning his Twitter into cash. So there’s a lot of that kind of thing. But also, like, the internet, the autonomous internet far-right, makes extensive use of Twitter bots…
H: Oh God yes! Look into Trump bots – Hilary had bots too, Clinton I should say, but – but yeah, the people who turn up when you mention any of the keywords of any given American political issue, like, a lot of them are bots if you look into their accounts, and a lot of them are what I called cyborgs earlier. So people will have little armies of bot accounts that just tweet abuse, or like spam. Fake news, out into the world, but they’ll also go in there and type other things to make it seem real. You can see them co-ordinating the stuff on 4Chan live, you can see them doing it.
C: It’s amazing, and it’s quite a remarkable contrast between these sort of villainous bots pretending to be humans, and masquerading a sense of human agency, and the way that you and other artists approach it…
H: And there’s also, like, very conscious… war…. And there’s, there are arms races that go on. There are people – so for example, the first reply to any tweet that Donald Trump makes will be valuable digital real estate, because if you get in that, whatever you link to gets loads and loads of clicks. And so there have been warring bot armies on different sides, to get replies in to Donald Trump’s tweets as quickly as possible – I might be able to find the articles on that. [It’s] being used in lots of different directions. There’s a brilliant bot maker called Nora Reed, best known for Thinkpiece Bot [@thinkpiecebot], if you’ve seen that, it’s beautiful. She’s made many good bots and she’s pioneered a form called ‘honeybots’, which tweet, just tweet platitudes on keywords to annoy reactionaries on Twitter [laughter]. There’s, like, a Christian mob there to annoy like reactionary atheists, of which there’s a lot. So the kind of people that search for keywords and then tweet abuse at them, she’s made bots to deliberately attract them and waste their time, and argue with them in a completely infuriating way. So there is like an arms race that goes on at the political end of Twitter bots.
Will Interruptions Survive Without Thrill-Seeking?
— Thinkpiece Bot 🌹 (@thinkpiecebot) May 22, 2017
C: Cool – so actually your bots are engaging with all this activity that’s going on, it’s not just like….
H: My bots are taking place in that social sphere. I’m not that sophisticated a programmer and most of my bots are not particularly interactive, and that’s partly because I’m more of a poet than I am a programmer. So most of my bots are non-interactive texts, some of them have some very mild interaction with them, or some of them have a form of – I’m better at designing social interaction than machine interaction, so I have one, I have Autoflâneur [@autoflaneur] that’s designed to be interacted with in real space, and I’ve thought through the social space around the text of that bot. There’s another one called DebateCoachBot [@debatecoachbot], which you can @ into any argument that you’re having on Twitter and it will send you like friendly messages, and I made that to try and keep people being nice and constructive in arguments.
Fellow worker. Check in: the person you’re talking to is a human.
— Debate Coach (@DebateCoachBot) May 22, 2017
C: Yeah! And that’s like countering the, as you say, the rules of Twitter, in a really kind of inventive way…
H: And that’s kind of rooted in my theatre background, my performance art background, in that when I’m making theatre and making performance I’m thinking about designing social interactions, like that’s what I do, so I’m doing that in the Twitter bot space as well. Because I know how to design those but I don’t know how to program very well. I mean I can program a bit, but…
C: Let’s talk about those kind of practicalities now. I wonder, can you talk us briefly through how you go about… [laughter] how?!
H: I mean if you Google ‘how to make a Twitter bot’ you’ll find a hundred tutorials and the best collection of tutorials which explains things better than I ever can is botwiki.org, which I highly recommend. There’s a general programmer, artist, a wonderful person who tweets as @v21, who made a platform called ‘Cheap Bots Done Quick’.
C: I know that platform, yeah!
C: So is that a collective thing, where several people are making bots, like NaNoWriMo?
H: So NaNoWriMo is a really long-running thing where people try and write a novel in a month, in November. And then, in response to that, a couple of people who are into generative literature, Darius Kazemi is one of them, I can’t remember who the other person is – Hugo something I think [@hugovk]– created NaNoGeneMo, which was National Novel Generating Month. So people had to write a program that would generate a 50,000 word novel in a month, with a very very loose definition of novel, because natural language processing isn’t really developed enough to write, like, actually interesting 50,000 word novels, so most of them are bizarre and wonderful experimental texts. So NaNoGenMo was happening, and then we were like oh, why don’t we do NaBoMoMo?!, ‘National Bot Making Month’, and the challenge was to write a bot every day, and I think I was one of only two or three people who achieved it, or possibly I was the only one… I’m not quite sure.
H: I’m really obsessive!
C: So you mentioned botALLY…
H: Yes, botALLY or botALLY…
C: Tell us a little about that.
H: It was a community of makers, that started just as a Twitter hashtag, and there are various places that it gathers on the internet. And it’s an ethic and an aesthetic as much as a community. It’s people who make artsy interesting political Twitter bots, and who are interested in the politics of Twitter bots, and discussing generative literature, generative texts, the role that bots have in our world and pushing that forward, aesthetically, ethically, politically. That’s all it is. But it’s very friendly and welcoming community, so anyone who wants to make Twitter bots, if you just tweet a line like how do I make a Twitter bot? and then the hashtag ‘botALLY’, somebody will pick that up and be like oh right, have a look at this link.
C: That sounds great, and also refreshing in terms of – I think a lot of people who aren’t part of these communities might feel a bit intimidated by the kind of knowledge and so on – but you’re saying that’s not a prerequisite to…
H: No, I mean I came into that community with no programming knowledge, as a poet, having made, like, two very very technically simple Twitter bots, being interested in Twitter bots, and I watched a conversation and I chimed in every now and then and I picked up from what they were talking about, and then when I needed help I asked for it, and people helped. And yeah, they’ve taught me a lot, loads of, yeah…
C: Cool, so it’s a good place to start then.
H: Yeah, just have a look at the chat that’s on #botALLY, and botwiki.org is a very good place to start as well.
I’m not going to go so far as to say machines have consciousness. I do think they have agency, but I’m not going to go so far as to say they have consciousness. And because there’s a distancing of a non-conscious machine that is not making aesthetic decisions, they can break language or remake language in a much more dramatic way.
C: I wonder if you could talk about, maybe take an example of the bots you’ve made, and some of the challenges you faced, which you feel you kind of surmounted and it was a success…
H: I’m just looking at my phone… Pick a bot. I don’t know, are there any in particular you wanted to ask about?
C: Umm, not necessarily – wait, what’s the one you do with the train map? I like that one, it seems quite elaborate as well, using images and so on.
H: Yeah that’s kind of interesting, in that – so that is, um, Spotrail… Botrail! I think it’s called Botrail [@BottishRail].
C: I see what you did there!
H: Yeah, that wasn’t my name actually. I think it was Henry Bell’s…
C: Oh yeah…
Tidal Tram pic.twitter.com/UAOUTKOdZm
— BotRail (@BottishRail) May 21, 2017
C: Oh right!
C: So do you then find yourself, does the idea come first and then you’re like, okay now I’ll have to learn how to do x, y and z so that…
H: Yep, usually the idea comes first, but in the process of figuring out how to do the idea the idea gets better. Because you have an idea and you think, ah, I’ll try and implement that, and as you’re implementing it you see the possibilities around that idea, so you can reach into there, and reach into there.
C: In that respect it does sound like writing a poem.
H: Yeah, exactly! But also like – if I’m doing a poem that has a performance element where I think it would perform well (not all my poems perform well), it has to be performed to an audience a few times, at least before it’s done, because I need to see how the lines and the rhythms land in those people’s ears, and then modify it accordingly. And similarly, usually when I make a bot I need to run it for a few days to look at what three days of the timeline looks like, and think, oh I need to tweak this and rebalance that and shift that.
C: That’s a really interesting idea as well, that it’s a sort of drawn out nurturing or cultivating or sculpting, whatever kind of metaphor you use…
H: Cultivating’s the metaphor I usually use. I think of myself as having a bot garden, or maybe a bot ant farm. Yeah, I have a garden of bots, that live autonomously and grow autonomously and do their own thing, but I do need to tend them, prune them, repot them sometimes, yeah.
C: Nice. So there’s a quote, I can’t remember it exactly – I should have made a note of it – Christian Bök, the Canadian poet, said that – I’ll paraphrase – the poets of the future won’t be poets as such. The great poets will be revered not for writing great poems but for creating a robot that will write great poems for us. To what extent do you agree with that statement?
H: I’ve written a whole essay about it. It’s called ‘What Can Poets Do About Robots?’ which says anything that I could say better, I think I was quite pleased with that bit of writing.
C: Is it online, can we link to it?
H: Yeah yeah. It’s on my blog somewhere. Um… I sort of agree. And I sort of don’t, in that lots of artforms have taken this trajectory – from craft to conceptual art, from creating things to making conceptual art, and that’s actually sort of circular, in that conceptual art like Damien Hirst, boring as he is, now have studios of craftsmen – craftspeople, craft workers – to make all of his stuff. And that’s also what Rembrandt did. So it kind of goes circular. They come up with the concepts and other people do the execution. And I think poetry to an extent has now discovered you have to do the same thing.
C: Mhm – so we’re outsourcing…
H: Yeah… well yeah yeah yeah. And like generative poetry has a long history and, but, there’s a particular – and this is also where Christian Bök comes from – there’s a particular moment with the Oulipo, who are often misunderstood as writing lots of weird texts, but what they actually define their work as is…
H: Is writing the rules. That’s what they do. That’s how they defined their work at the start. They were like let’s come up with constraints for making poems, and the work is to come up with an interesting constraint and to exhaust that constraint. That’s why you have things like Eunoia, which is exhausting the constraint of writing univocalism, to do the maximum you can do with that, so it almost never needs to be done again, and then people misunderstand it – including me – and are like oh it’s so fun to write univocalisms! But the whole point is you don’t have to anymore [laughter].
C: That’s quite similar to conceptual art in some respects as well, in that it’s, you can only do conceptual art once.
H: Certainly making bots to make great poems, that is what’s happening now, that’s what we’re doing. I mean I do think there are bots – I don’t think I’ve quite managed it – but I do think there are bots that make genuinely great poems. My favourites are poem.exe [@poem_exe]…
a young couple argue
— poem.exe (@poem_exe) May 22, 2017
C: I know that one yeah, it’s brilliant.
H: Makes these like beautiful… And then I think there’s, it’s either Tiny Minimalist or Autominimalist. It wasn’t announced with great fanfare, but it’s very quietly beautiful, which is appropriate. I’ll just see if I can find it. Yeah, Auto Minimalist [@AutoMinimalist], and it’s tweeting endless variations on various techniques, pioneered, and variations of techniques, pioneered by Aram Saroyan.
C: He’s great.
as a new
— Auto Minimalist (@AutoMinimalist) May 21, 2017
H: He is great. And that’s made by Allison Parrish.
you don’t have to enjoy the magic, you can also enjoy the craft. Some magicians, some people want to enjoy the magic show and some people want to enjoy the craft, and both are legitimate.
C: Yeah, that’s cool – and it’s interesting you should mention that, I don’t know where poem.exe gets its [language]…
H: I don’t want to know! People have threatened to tell me but…
C: That’s a really interesting point in and of itself. Is it exposing the man behind the curtain, or in this case robot? Does it spoil that kind of magic? What do you think?
H: I don’t know. Yeah, I don’t want to know. Yet, if you want to – you don’t have to enjoy the magic, you can also enjoy the craft. Some magicians, some people want to enjoy the magic show and some people want to enjoy the craft, and both are legitimate.
C: And in a sense that kind of goes back to what you were saying about how when people play games they’re trying to find ways of breaking the rules, or in video games you find the glitch that lets you jump forward four levels or something…
H: But back to the can robots take our jobs as poets thing, can robots take poets’ jobs? I do think we’ve now got the technology and the aesthetic movement to write robots to make poems. They’re no [inaudible], poem.exe is the closest that I know. They’re nowhere near being able to write good poems for most forms. Like, haiku it’s just about got there. I don’t know, any other form that poet, that poet – limericks, there’s some good limericks.
C: Yeah, and it seems to me, the act of retrieval in Pentametron [@Pentametron] for example, is using form, but it’s not writing is it…
H: No, it can’t write genuinely great poems. It writes good poems, but it can’t write great poems.
C: But it’s interesting you mention this Saroyam connection with Auto Minimalist, because a couple of my favourites, I can’t remember the names of them, but there’s one that uses the syntax of Ezra Pound’s In a Station of the Metro [@BlackBoughBot], and there’s another that uses the syntax of The Red Wheelbarrow [@DependsUponBot], and they’re great, and often very funny, and absurd, which is what Twitter bots do so well I think. But yeah, we’ve spoken a bit about the tradition of that kind of generative work but how do you see it as – I mean you’ve got Saroyan, you’ve got Ezra Pound, you’ve got William Carlos Williams, all innovators in modern, or modernist, or late modernist poetry – how do you see Twitter bots as responding to or following from that page-based tradition?
so much depends
a wry height
locked with slight
beside the gay
— So Much Depends Upon (@DependsUponBot) May 22, 2017
H: Mhm. So I might – there’s a few different lines to draw. One might be to go to Gertrude Stein, breaking language, reforming what language can do, going into that. Asemic writing or asyntactical language, creating new forms of expression – and because they’re distanced by a non-conscious machine – and I’m not going to go so far as to say machines have consciousness. I do think they have agency, but I’m not going to go so far as to say they have consciousness. And because there’s a distancing of a non-conscious machine that is not making aesthetic decisions, they can break language or remake language in a much more dramatic way. Although I think Tender Buttons is a masterpiece in terms of creating aesthetic effects that go beyond meaning. Yep – another line that we could draw is to people like Ezra Pound, but I think that line’s been drawn, I think that line’s been superseded long ago by hypertext. Like, I find The Cantos boring, because we live in a hypertext world, we live in a world of constant collage, shuffling, reference, and there are way better – Twitter is one of them – there are way better formats, media, for doing that kind of collagey, referential work than the page. Another line we could draw is back to Baroness Elsa von Freytag, the early sound poet. I made a bot for her. It’s called Auto Else – Auto Else, I should say [@autoelse]. There are lots of bots that are playing with sound in various ways, and that are just fully exploring the expressive capability of sound. Like, I’ve got quite a few sound poetry bots, but I’m not the only one.
Pft! Any baselesspoet has sufficient tapping
Shibboleths in their Freudian pegboard to loom all
Repurchases. I’ll hire one.
— Auto Else (@AutoElse) November 20, 2016
C: What about concrete poetry then, in that context?
H: Concrete is harder for bots. Concrete poetry’s really hard to come up with new means of expression in, and this relates to the, can bots supersede, can they take all of our jobs. Because they can. I think eventually, I think it’ll take a couple of decades, but we’ll be able to do most forms well – robots will be able to do most forms well, they’ll be able to come up with interesting – because they’ve long been able to come up with very interesting absurd poems, that’s a magnificent history. So there are types of poetry they can do really really well.
C: Yeah, like have you seen that The Policeman’s Beard is Half-Constructed?
H: Yeah yeah yeah, exactly exactly. So there’s loads of poetry they can do extraordinarily well. But what they can’t do at the moment is personal expression, individual expression. Political expression that’s tailored to a specific moment, because they’re not conscious. So until we have very advanced artificial intelligence, at which point there isn’t really any distinction between them and us, they can’t take our jobs! [laughter] Because the job isn’t always to innovate and make new, that’s not the only job of a poet. The job of a poet is also to express yourself – you are a self which is unique, snowflake [laughter] – and to be a bard, to respond to the existing community in a meaningful way.
C: Yeah, absolutely yeah.
H: So they can’t take those jobs until they’re so much like us that they may as well be us, or we’re so much like them that we may as well be them. And I also think that will come. But you were talking about concrete poetry – because concrete poetry’s so specific, every concrete poem is so specific it’s very hard to generate, in a continually satisfying way.
C: Yeah, and it’s specific in the sense that its syntax is singular…
H: Exactly, that’s…
C: That’s the essential property of a concrete poem.
H: So I have a couple of very basic generative experiments around concrete poetry but they’re not aesthetically satisfying, they’re exercises.
C: I mean to pick up on the – you mentioned The Cantos and it just raised the question in my mind – what’s the relationship of the Twitter bot to poetic genre i.e. fragment and epic?
H: You might have to unpack that question.
C: Okay – umm, do you see the feed as an ever-expanding poem, or do you see each tweet as a discrete entity, or, well I suppose those are not mutually exclusive…
H: A wave particle! [laugher] I think it depends on the bot, that obviously some bots – so for example autoflâneur is explicitly written as an infinite poem, that it’s an endlessly unrolling set of instructions, and I designed it to be read as an infinite poem.
C: That’s interesting as well, because you were talking about how that bot uses social spaces and social conventions and so on, so that its infinity comes up against our finitude.
H: Yup, and another one of those that works reasonably well like that is – it’s not finished, but, well, it’s getting there – is one that I do called ‘Ubersonate’, which is based on…
H: Yeah Kurt Schwitters’ Ursonate because it’s infinite, so it’s an infinite Ursonate. And it’s just using the syntax of Ursonate, and generating it so it’s like an infinite sound poem. But there are others where – so for example, both poem.exe and Auto Minimalist, each of those tweets is very and clearly intended to be an individual poem, so if you look at the whole feed it’s a book of poems, but it’s not an infinite poem, it’s an infinite book of poems, if that makes sense. But there are bots that can do both, and you have to understand the bot in its whole context. And there are lots of bots, including many that I’ve made, where the joy of them actually falls apart when you look at the feed, because the magic’s important…
C: That’s something that’s interesting as well. Would you ever consider publishing the output of your Twitter bots on the page, in a book?
H: Oh yeah, I’m going to do it.
C: Oh really? Now that’s exciting! [laughter]
H: Next year it’s, yeah, I can’t really reveal more than that, but yes. It’s active recontextualisation.
Every artistic movement responds to the technical and social conditions of its time, and Twitter bots might not be interesting in twenty years time. And that’s okay. Like, I’ll obviously be sad.
C: Well let me ask you two more questions because I’m conscious of time. Firstly, it’s interesting what you’re talking about in how at the moment these poems are good at doing absurdity but natural language processing is at such a point… For example, I did a riff on Ursonate as well, when I was listening to, I think it was Christian Bök’s reading of it on YouTube, and I accidentally clicked the closed caption thing, and it thought that he was speaking Italian, so it came up with these subtitles in Italian. Then I used the automatic translate to get them in English, and then I put that text in the frame of the Ursonate score, kind of like a translation by Google. And for me we’re almost in this golden age of natural language processing for lovers of the absurd, because the poetics of error you get with Google Translate are just brilliant.
H: Yeah, the uncanny valley.
C: Exactly, and it also makes me think of – a bit of a preamble for this question I apologise – of early cinema, it was quite avant-garde until the talkies developed and it became more conventional – maybe conventional’s the wrong word but you know what I mean…
C: Yeah! Is there then a danger that if natural language processing and generative literature becomes more intelligent, and better, could it potentially lose some of the magic, much as the internet has…
H: Well yeah! But more than that, I think it shifts. Every artistic movement responds to the technical and social conditions of its time, and Twitter bots might not be interesting in twenty years time. And that’s okay. Like, I’ll obviously be sad. And like, I might even be one of them, like individual creators, who are still obsessed with early noughties style Twitter bots, and because their obsession is so outlandish they’re still creating genuinely interesting work. In the same way that there are people writing genuinely interesting sonnets! Why the f… Why should sonnets be interesting in the twenty-first century? They do not speak – they shouldn’t speak – to the people…
C: But they’ve never been more popular!
H: Yeah yeah yeah! So that will happen, it will go in and out of vogue, but it should, and the social meaning of that poetry form will change, and as you say the potential of natural language processing will change, it’s all in the, like, it’s in the downward slope of the uncanny valley at the moment, where it’s still very transparently machine made, it’s not at the bottom and it’s definitely not rising up the other side, so I think we’ve got a long time…
C: Exciting times…
H: And I think it’s going to take a lot longer than people think.
C: Yeah, yeah.
H: I think AI is going to progress. Like, I think people are expecting really impressive AI in five years. I think it’s going to take twenty or thirty.
C: Yeah, especially in terms of language.
H: People just have no idea how hard it is, and how little, what a small advance we’ve made.
C: Yeah. I wonder then, to close things up, and you kind of answered this question already, but where do you see the future of Twitter bots vis-a-vis the poetry world? Do you think they might – you know, it’s kind of niche – do you see it becoming… Much as concrete poetry, once upon a time, was ‘it’s not poems’, whereas now everyone is happy to have a concrete poem in their collection, how do you see that developing?
H: Well, I mean I’m trying to push it, and this is one reason I’m putting print Twitter bots in my next book. To be clear, that recontextualises the poetry, and that makes it a very different thing, which is important. So that’s partly why I’m putting that in there, just to try and introduce generative literature to new audiences. Because I was shortlisted for a prize last year – the Forward Foundation commissioned a poem from me and I gave them a Twitter bot just to see what would happen – I also gave them a static poem, just as a static poem, as well, but it was just, like, picking the best of the Twitter bots. But also, progress. Stanza – somebody at Stanza – got in touch this year and said ‘we love autoflâneur, can we do something with it at Stanza’, and I was like yeah sure, so I made them a new version of it.
C: A kind of St Andrews special!
H: Yeah yeah! So I made them a Stanza-specific version of it. Because I want mainstream poetry to see this world and see the potential, and I think it’ll come, yeah, I think it’ll come quicker than, yeah, poetry changes really quickly. And it’s, because it’s a very populist form it gives more people an in. And like, I think there’s much more resistance from British lyric poets to, like, American-style post-internet poets. There’s a bit of a generational war going on there, which I’m trying to sidestep by just writing the weird stuff that I write. I’ll take your word, I’m not really into either of you! [laughter] That’s not true, I like reading it, I just don’t write either of styles.
H: Haha, yeah don’t quote me as saying I don’t like either of those, I actually do like them. No, yes, that’s what it is – I love reading all kinds of poetry, but I like writing games, forms, generative work.
C: Yeah, I can sympathise with that [laughter]. Well thank you very much, it’s been a pleasure.
H: Thank you, it’s been a lovely chat. I hope this was the kind of material you were looking for.
C: It was exactly the kind of material we were looking for, thank you!
WE DEMAND FREEDOM FROM THE ASHEN IDEOLOGIES OF THE BOURGEOIS NATION-STATE pic.twitter.com/HJwTDZ2X2D
— Anarcoo (@AnarchyCoo) October 16, 2015
Harry Giles is a writer and performer from Orkney, Scotland, now based in Edinburgh. Their work gets everywhere from page to screen, from happening to performance or, as they put it, ‘in the crunchy places where performance and politics get muddled up’. They’ve performed poetry and theatre across the globe, winning several slams and putting on critically acclaimed shows at the Edinburgh Fringe including Everything I Bought and How It Made Me Feel’. They also make games, working with Book Week Scotland, Nevis Land Partnership and Stanza (see above!), among others. Their first collection Tonguit, Freight (2015), was shortlisted for the Edwin Morgan Prize (2015) and the Forward Prize for First Collection (2016). Be sure to check out Harry’s work at harrygiles.org.