Robin Sloan
The Media Lab
June 2021

Dreams of discord

The Wild Apples of Caucasus, 1930-1932, N. I. Vavilov and J. S. Lawson
The Wild Apples of Caucasus, 1930-1932, N. I. Vavilov and J. S. Lawson

Below, you’ll find a request for ideas and reports! First, some con­text.

I wondered aloud the other day: could you oper­ate a group Twit­ter account with a Discord channel? If you’ve never used it: Dis­cord is a ubiq­ui­tous chat app, each “server” (not really a server) divided into myr­iad chan­nels. Think: Slack except not for work, or IRC with many more features.

The group Twit­ter account would work like this:

This led me to the vision of a kind of meta-Twit­ter where ALL the accounts oper­ate like this: not a network of individuals, but a net­work of col­lectives, large and small, find­ing ways to talk to each other. Silly idea, and I’m not sure I want to take it any fur­ther than the pre­ced­ing few sentences, but it did get me think­ing about Dis­cord and its capabilities.

A while back, sev­eral newslet­ter-ers banded together to estab­lish a shared Dis­cord for their paid sub­scribers. I’m a paid sub­scriber (and superfan) of one of them, so I joined. The actual chat turned out to be too snarky for me — which isn’t to say it was par­tic­u­larly snarky, just that my snark tol­er­ance is close to zero — but I loved the idea, and still do. I think it’s smart, fun, and genuinely new.

Discord is highly programmable. Using its API, you can engi­neer bots that oper­ate in/on the chat: mod­er­at­ing messages, respond­ing to spe­cial commands, man­ag­ing access to dif­fer­ent chan­nels, post­ing mes­sages of their own, even mes­sages with cus­tom “action buttons” attached … there are a lot of possibilities.

I’m a mem­ber of a few Dis­cords focused on spe­cific soft­ware and hard­ware projects, and they are: fine. They’re mostly some fla­vor of “cus­tomer support”, along with the occa­sional announcement, etc. EVEN SO: the famil­iar dynam­ics of social spaces emerge! Even these tiny Dis­cords each have their own “loud” users, the ones who take up all the (dig­i­tal) oxy­gen.

I wonder if you could use a bot to help bal­ance that out — to gen­tly shush the loud users and/or encour­age the quiet ones? Of course, the first thing that occurs to me is some kind of daily quota (mes­sages? characters?) but that’s prob­a­bly too blunt.

How might you go even fur­ther, though? Maybe mod­u­lat­ing the “amount of chat” is play­ing the wrong game. Could you recon­fig­ure the uses for, and expec­ta­tions of, a “chat room” in a very deep way?

What is a chat room FOR?

Whenever I think about ~social infrastructure~ related to the Soci­ety of the Dou­ble Dag­ger, I think about tap­ping into the experience, eru­di­tion, and sen­si­tiv­ity of the group of peo­ple who receive these emails. Like: I am able to ask ques­tions of this group, and it’s an enor­mous asset; a privelege. It might be cool if the group could do the same!

Beyond that, I don’t know. Is a chat room, even a very good one, just another trickle of mild dopamine hits, another social slot machine lined up beside the rest? Or is it an oppor­tu­nity to make deeper connections, the kind that might be truly gen­er­a­tive? (You start chat­ting with an inter­est­ing per­son on the Dag­ger Dis­cord, and six months later, you’re build­ing a web­site together!)

Is it just … a hangout? Noth­ing more, noth­ing less?

You might detect above my gen­eral unease with like … all forms of social inter­ac­tion online. For me, this is at least seven lay­ers deep, and explain­ing them all would require a whole other dispatch — a long one. Suf­fice it say, the moti­va­tion behind both (1) my exist­ing sys­tem of newslet­ters and (2) this new line of think­ing is the same. It’s a sense of like, I am so dis­at­t­is­fied with all avail­able options that I guess I will make my own.

To that end: I am start­ing to feel like I might (might!) start a Dis­cord some­time this year, so here’s where I pose my ques­tion, which is two-pronged.

Prong the first: what can you imag­ine a bot doing in a Dis­cord to make it health­ier and more inter­est­ing AND/OR weirder and more gen­er­a­tive? (Assume, for the sake of spitballing, that the bot is omni­scient and omnipotent: it reads every­thing and it can oper­ate on mes­sages and users how­ever it likes.)

Prong the sec­ond: what ~social infrastructure~ have you encountered, bot-related or otherwise, on Discord specif­i­cally that impressed you? (I’m not inter­ested in hear­ing about dig­i­tal social spaces gen­erally; that scope is too broad, and I’m think­ing, at the moment, spe­cifically about Dis­cord.)

Whatever you choose, Robin will see that this response is coming from . Please also sign it with your preferred name ✌️

(There’s a response field in this space for subscribers to this newsletter. If you’re seeing this, you’re either not a subscriber — alas — or you reloaded this page in your browser, which broke the link to your email address. Just follow the link from your email to re-establish it.)

I’ll loop around and add some of your responses to this newsletter (respecting your quotation preferences, indicated above) as I receive them. Check back in a couple of days!


Casey writes:

Can I plug my Discord server-as-art-gallery? Scrumwave is a read-only Dis­cord server (with the excep­tion of a guest­book channel … I should really make a gift shop channel, I suppose, to recre­ate the museum experience) that dis­plays gen­er­a­tive art. Right now it’s more of an alter­na­tive to Twit­ter/Mastodon post­ing, a place to view the work in its own con­text instead of blended into your own timeline, but I’ve been pon­der­ing adding to it by mak­ing some­thing more interactive. [ … ]

Needless to say, I love this — exactly the sort of boundary-pushing, chat-deconstructing kind of thing I was hop­ing to discover.

Thomas reports in:

One of the most intrigu­ing uses of Dis­cord I’ve seen is the DFTBA Life’s Library book club, where users self-select a “shelf” based on their personalities, and while every­one has access to the same core “read-only” chan­nels like announce­ments and Q&A, the main dis­cus­sion of that month’s book only takes place on your par­tic­u­lar shelf. The same dig­i­tal oxy­gen-sucking occurs here, of course, but it never feels suffocating. More like sit­ting in the corner of a dark bar lis­ten­ing to strangers talk intelligen­tly about a shared experience. It’s really a delight­ful lit­tle cor­ner of the internet, and I would imag­ine the per­son and bot mods do a LOT of work to make it feel that way.

Bob reports in with a neg­a­tive (but still very use­ful) example:

A Dis­cord for stu­dents that I lurked about in (I was their teacher, spying, play­ing antropologist) had a bot that auto­mat­i­cally answered the basic fre­quently asked stupid ques­tions (about things like where to find the syllabus). This seemed like a good idea at first (no need to tell the newbies all the basics over and over), but it stim­u­lated ask­ing those low-com­mu­nity-value ques­tions, obviously. Plus it removed a chance for the oldies to wel­come the new­bies with a friendly answer. My point, I think, is to reflect not just on what cool and use­ful things bots could do, but to think what prac­tices it may stim­u­late or cultivate.

Here’s an inter­est­ing thought from Vanessa:

Bots need to be better teachers! In education and parenting there are lines of thought about how to encourage and discourage various behaviors by the ways we interact and the words we choose to use with kids. Not by enforcing behavior (e.g., posting limits if you are too “loud”) but rather by facilitating spaces and encouraging states of mind in which community members make good choices. What if a bot could message individuals and groups, triggered by various types of action, and in the message use language that might prompt positive behavior in those receiving the message?! [ … ]

Usually, it’s impor­tant to develop rela­tion­ships first to make this sort of thing effective. But maybe the nature of opt­ing in and want­ing to be a part of a spe­cific type of com­mu­nity (helpful to be explicit about what this means up front) can serve in the stead of that relationship. AND, the bot could be affir­ma­tively trig­gered by other users (?) who could use it to send more per­sonalized mes­sages.

It makes me think of Red­dit Gold, a gift bestowed upon par­tic­u­larly great posts (and their authors) by other users on that platform. Imag­ine a whole host of “golds”: dif­fer­ent gifts and thank-yous that could be given by one user to another. Fun to con­sider what form they might take … 

Here’s Stephen Carradini, help­fully nudging me towards curation:

[ … ] I think that curation is a thing you may be more inter­ested in: instead of qui­et­ing the loud people who say too much or the wrong thing, curate and pro­mote the work of people who say the right thing or the right amount. While this is to some extent “sets of rules applied to sets of peo­ple”, it is a much more cel­e­bra­tory idea. [ … ]

I had an inter­est­ing back and forth with my friend Dan Bouk in which I clarified:

For me — and I acknowledge this is idiosyncratic, per­sonal, maybe stupid — I can recog­nize a recur­ring impulse to just “share a thought”, or maybe a para­graph of what I’m read­ing, or an idle ques­tion … and Twit­ter is presently the only place to do that, which feels like a shame. Some­times when the impulse strikes I will send it instead as a text mes­sage to one of a few friends, but that has its limits: “Sloan, ENOUGH with the inscrutable snap­shots of book pages! Yes, I see where you’ve under­lined it!”

Maybe it’s the reflex that’s the problem, not the platform; maybe I need to just learn to read some­thing great and appre­ci­ate it qui­etly and keep it to myself??

(You can tell I don’t actu­ally believe this.)

Here’s a sweet report from Jacopo:

This is some­thing quite small but that I appre­ci­ated a lot. In the now defunct Learning Gardens Slack, there was a bot that would reg­u­larly and pub­licly ask: “Hey, what are you work­ing on and how is it going?” This Slack was a space for self-learning and play­ful exper­i­men­ta­tion (very much aligned with the vibes of: “I guess I’ll make my own”) and the bot allowed everyone to just share their projects and get some feed­back without mak­ing-it-a-thing. In a way, the bot was a (dig­i­tal) oxygen injector: it cre­ated the space for peo­ple to share with­out hav­ing them take up the oxy­gen for every­one else.

“A dig­i­tal oxy­gen injector”; what a nice way of think­ing about how a bot might oper­ate.

Here’s Meg on bot moderation:

I wonder if some kind of sentiment analysis would be use­ful for a bot mod­er­a­tor (boterator?). If the con­cern is keep­ing Dis­cord a healthy space, a big piece of that has surely got to be mod­er­at­ing tones and lan­guage. I’m sure a Python or R script for sen­ti­ment analy­sis could be mod­i­fied to treat the server as its corpus.

Even as I say that, though, I have to admit that I find the idea of bots mod­er­at­ing human inter­ac­tion and lan­guage to be vaguely dystopian.

Here are some solid, struc­tural thoughts from Kim:

I’ve been on a few dis­cords that encour­age you to send lots of mes­sages, by users gain­ing lev­els via bot the more they engage with the discord. I don’t enjoy it because even in chat rooms I hate small talk. I talk with the pur­pose to com­mu­ni­cate not just to fill the air.

So, I wonder if you could make a bot that would encour­age thought­ful dis­cus­sion by say, gain­ing points when some­one replies to your response.

I also think it would be inter­est­ing to have bot throw out prompts like this when chat gets slow.

I’ve seen on Discord implement Reddit-like upvotes/downvotes for ques­tion submissions, so maybe you could also imple­ment that in the bot and then assign roles that let you say, talk more often when you’ve proven your­self to be some­one who gen­er­ates use­ful comments. (So really I guess I am sug­gest­ing Red­dit-ifying a Dis­cord.)

It’s inter­est­ing how many responses referred in some way to the speed of the chat, all say­ing some ver­sion of, “it always goes too fast!” Peo­ple have very dif­fer­ent “ideal chats”, obviously, related to the pat­tern of their lives, the num­ber of other chan­nels they’re man­ag­ing, their read­ing speed, their com­fort with chaos — all sorts of things. Maybe one pre­con­di­tion for a healthy online com­mu­nity is that most of its par­tic­i­pants “oper­ate at the same speed”? (I’m not sure if that’s true; just a thought.)

Here’s Alek with a thought that I have to con­fess I find appealing, even if I know it would be tyrannical:

Make it a “Dis­corded Repub­lic of Letters”—the bot should enforce only mes­sages that meet a min­i­mal epis­tolic standard. No rushed remarks writ­ten with your thumbs. Nice thought­ful form.

(Okay, I would make an excep­tion for dialogues, if they read as if taken straight from Plato. The bot could enforce that too, after feed­ing it all of [his works] and gpt3-ing them like crazy to estab­lish a point of reference.)

Here’s a use­ful note from Jon:

My issue with these plat­forms when they become social spaces, rather than work spaces, is that its easy to miss the truly good con­tent (the same can be said for social media in gen­eral)—it’s why I like email newslet­ters so much, they sit there patiently and wait to be read.

So my bot sug­ges­tion is a peri­od­i­cal (weekly?) col­lec­tion of the best posts/conversations, this could be based off the num­ber of inter­ac­tions, or per­haps there’s a bet­ter met­ric out there?

I totally appre­ci­ate this is tak­ing it full circle, but I know for me I’d get a whole lot more out the spaces like this.

The abil­ity to “sit there patiently and wait to be read” is absolutely an under­rated feature, and one that shouldn’t be taken for granted where it exists: in newslet­ters, and, of course, in books.

Here, Carl slices the Gordian knot:

There’s some­thing to be said for using ran­dom­ness when true fair­ness is impossible. We don’t know who should be chat­ting or how much, but we do know that more than N peo­ple pro­duc­ing M mes­sages in one day is unread­able and shreds con­text because you can’t be a completist. So how to fairly share the M? Maybe just ran­domly allo­cate a daily quota with the goal of ensur­ing that no more than M mes­sages are produced. Some days you have a lot to say and hit your quota, some­times you have noth­ing in par­tic­u­lar to say but a lot of tokens jin­gling around in your vir­tual pocket, and some days are slow and no one hits quota, so the cap is lifted all together. (Of course, like all lotteries, you could also secretly rig it to favor … hmm … )

Here’s Billy with a fas­ci­nat­ing report on a Dis­cord bot/game called Mudae, which is new to me:

There’s a good chance you’re already famil­iar with this, but the Mudae bot impressed me (mostly in a bad way lol). It’s just a gacha game where people roll to get pic­tures of anime, video game, etc. waifus, but peo­ple get really into it.

You get a cer­tain num­ber of rolls every hour, and the instruc­tor in the Youtube tuto­r­ial I was watch­ing explained that he was sleep­ing in incre­ments of less than an hour to make sure he got all his rolls in.

Mudae started some huge fights in the Dis­cord I was in, even break­ing up a cou­ple friendships. I think the peo­ple who fought over Mudae were high drama anyway — as you can imag­ine — and they were prob­a­bly just look­ing for some­thing to fight over.

But still … Mudae just gives you pic­tures of waifus. It’s like a somehow-even-less-inter­est­ing NFT. And yet peo­ple go wild over it.

Whether or not this really answers your sec­ond prong might depend on how you define social infrastructure. But if you’re just talking about the ways that bots on Dis­cord can bring peo­ple together, Mudae would define the flow of conversation. Peo­ple would come to the server every hour to make their rolls, and we’d all end up talk­ing about all sorts of stuff as we took turns rolling.

(Kinda funny that you’re talking about the eru­di­tion of your readers, and I’m talk­ing about anime waifus, but such is life!)

I have to say that this was prob­a­bly the most inter­est­ing response I received, in the sense that it points in the direc­tion of weirdness, “wait REALLY?”-ness.

How about a Mudae for public domain art?

Here’s my takeaway, for now:

Tap or click to unmute.

Sent to the Media Lab committee in June 2021