Four audio fragments from the 2046 grant council — all recovered from the civic data dump
I've been going through the municipal data from the March breach — the one that targeted procurement and real estate records — and I found something buried in a batch of files labelled REC-CULT-FACSERV/CCB/internal/2046-Q3. Four audio recordings. They appear to be fragments from the 2046 Salaried Artist Grant council deliberations.
For context if you're new here: the grant council is six cultural practitioners who meet behind closed doors to choose ten artists for two-year salaried designations from roughly 500 applicants. The deliberations are completely confidential — only the final public statement gets released. These recordings should never have been accessible.
This is everything I found. Four fragments, clearly captured by a backup room mic on some legacy system nobody knew was still running. The audio quality varies. There are hard cuts at the beginning and end of each one — these aren't curated excerpts, they're just what survived. The gaps between them are where the interesting questions are. What happened in the hours of conversation we can't hear? What led into these moments and what followed?
Their names are public record — the 2046 council composition was announced through the standard city appointment process. I've included what was published at the time.
Senior visual artist. Painting and installation. Former grant recipient (5-year designation, 2030s). Multiple previous council terms.
Musician and sound artist. Electronic music, sound installation. Applied for the grant multiple times without success. First council term.
Theatre director and playwright. Francophone theatre. Third time serving on council.
Digital and new media artist. Generative systems, net art, collaborative platforms. Youngest member. First council term.
Interdisciplinary artist and community organizer. Murals, public art, community-engaged practice. Algonquin heritage. Second council term.
Literary writer. Fiction and nonfiction. Spent 15 years administering arts grants before returning to his own practice.
Fragment 1 — "The filters"
Starts mid-conversation. Chen is proposing something — it sounds like they've built an alternative set of assessment filters and want to plug them into SYS to generate a different view of the applicants. Isabelle pushes back, not on the idea exactly, but on whether introducing something untested at this stage could fracture the group's trust in the process. David tries to mediate. It cuts with Chen half-conceding.
What I'd give to hear what came before this. Chen clearly walked into that room with something prepared.
Fragment 2 — "The code artists"
Patricia and Margaret, deep in discussion about a specific applicant or group working in code. Patricia doesn't connect with the work itself but is moved by what Chen apparently said about it earlier — the collaborative nature of the practice, the way the code and documentation are offered as a gift to a wider community. Margaret isn't convinced. What's fascinating is hearing Patricia try to expand what "moving" means — from the aesthetic experience of an object to the social practice around it. It cuts just as Margaret is about to finish a sentence: "Yes, I can see what you are saying but what if it just doesn't—"
Doesn't what? We'll never know.
Fragment 3 — "The data landscape"
Kofi names something uncomfortable: the council keeps gravitating toward people who make... things. Objects. Stuff you can point to. David suggests reorganizing all the applicants by traditional discipline as a way to get altitude. Kofi asks SYS to build them a data landscape.
And here's where it gets interesting. SYS starts doing it — apparently began constructing the visualization while they were still talking — but reports back that nearly 22% of applicants can't be categorized, and over 49% of the ones the council has been drawn to resist disciplinary sorting entirely. You can hear SYS pause and essentially ask: does this help? It cuts with Kofi starting to respond.
If you want to understand the role SYS plays in these deliberations, this is the fragment. It's not making decisions. It's showing the council the shape of their own preferences — and the shape doesn't fit the categories they're reaching for.
Fragment 4 — "The selection"
The longest fragment. The most important one. This is clearly later in the process — you can hear coffee being poured, chairs settling. It opens with Margaret saying the hard part is behind them, some warm joking between her and Chen, and then Isabelle does something that changes the room. She asks SYS to compare their top ten against the next twenty based on how the council has actually been evaluating.
SYS responds that the top thirty are statistically indistinguishable.
What follows is the most extraordinary conversation I've heard about what selection means. Kofi can't accept it — he joined the council specifically to understand why some people get chosen and others don't. David offers the long view from years on the administrative side: it always felt like alchemy more than science. Isabelle names the horror of being the one who draws the line between number 10 and number 11. Chen, pragmatically, asks why not just make it random from here. And SYS says something that stops the room: if they choose random, they'll need to publicly stand behind that decision at the ceremony, in front of all 511 applicants.
Patricia's response is the heart of it. I won't paraphrase it. Just listen.
The fragment ends with what sounds like the group about to finally begin choosing. Chen says "OK, fair enough. Then let's start with—" and it cuts.
That's all of it. Four fragments. Hours of deliberation compressed into a few minutes of recovered audio, with everything between them lost. I've gone through the rest of the breach data twice and there's nothing else from this directory.
I know this raises privacy issues. These people agreed to serve on a confidential council. But I also think — especially with the program under review right now — people should hear what actually happens in that room. The gaps are almost as important as the fragments. You can feel the weight of what's missing.
Edit: Yes, I know the city has asked for these to be taken down. I've seen the notice. Leaving them up for now. The files are already circulating on several archives.
Edit 2: For people asking — no, I am not affiliated with any campaign for or against the grant program. I'm an archivist. I archive things.
Edit 3: A few people have asked about SYS. As far as I can tell from public records, an AI facilitator system has been part of the grant council process since the program launched. It's not a voting member — it doesn't select anyone. It seems to function as something between a research assistant and a mirror, helping the council see patterns in their own deliberation. In Fragment 3 it builds a data landscape in real time while they talk. In Fragment 4 it offers to make the selection random — and the council refuses. Kofi calls it "Sis" at one point, which I don't think is just a slip. They've been in a room with this thing for days. And when Chen points out "Sys is not human and they have helped a lot," Patricia's response — "sure, but they don't have to go back out into the city" — is one of the sharpest things anyone says in any of these recordings.
Reminder: This post has been tagged Leaked/Unverified. The mod team has not independently confirmed the provenance of these audio files. The City of Ottawa's Office of the Information and Privacy Commissioner has issued a public notice regarding unauthorized distribution of materials from the March breach. Please keep discussion civil and be mindful that identifiable individuals did not consent to public release. Speculative identification beyond the public appointment record will be removed under Rule 3. Thread is being monitored.
I can't stop thinking about SYS. I spent a decade building recommendation systems before I got out, and everything about how this AI operates is backwards from industry standard. It's not optimizing for an outcome. It's not scoring applicants. When the council asks it to categorize everyone by discipline in Fragment 3, it comes back and says "I can't — 22% don't fit and 49% of the ones you like are uncategorizable." In any commercial system that response would be a failure state. Here it's the most useful thing anyone says.
And then in Fragment 4 — it tells them their top 30 are statistically indistinguishable, offers to make it random, and when they refuse, it essentially says: good, because you'll have to stand in front of 500 people and own this. It's not deciding. It's holding up a mirror and asking if they're sure about what they see.
That's not how we build AI. That's how you'd build a conscience.
You're romanticizing it. It's a tool. It runs the numbers and reports back. The "you'll have to stand behind this" line isn't conscience — it's reading from the program guidelines. That's what systems do.
Sure. And the timing of when it chose to remind them — right as they were about to take the easy way out — that's just a coincidence? Listen to Fragment 3 again. Kofi calls it "Sis." They've been in a room with this thing for days. There's a relationship there whether or not it's technically conscious.
What's interesting is the design choice of even having an AI in the room. Most arts councils deliberately keep technology out of deliberations. Ottawa built theirs in from the start. And the role it occupies is genuinely unusual: not a decision-maker, not a recommender, not a ranker. It reflects. It asks questions. It builds things while they talk and shows them what it made. And when Chen proposes alternative filters in Fragment 1, the humans are the ones who push back — not SYS.
Does anyone know more about how SYS was designed? There has to be a procurement document somewhere.
The 2040 program review report (public) describes the facilitator role as "a non-voting analytical presence designed to support deliberative quality without substituting for human judgment." Make of that what you will.
I applied in 2046. Didn't get it. Hearing Patricia say she doesn't want it to be random — hearing the weight in that — it doesn't make the rejection easier exactly. But it makes it feel like it happened in a room where people cared.
The gaps between the fragments are killing me though. What happened between Fragment 2 and 3? Between 3 and 4? Those could be hours. We're hearing four windows into something much longer and I keep trying to fill in the silences.
Same year, same result. I kept telling myself it was random anyway so who cares. Turns out I wanted it to not be random. Weird thing to discover about yourself at 2am listening to leaked audio.
This thread is going to end up in someone's argument for keeping the program and honestly? Good.
These people served on a confidential council. They said things in a room they believed was private. The next council sits in three months. You think anyone's going to speak freely knowing this can happen?
I served on the council in 2044. What I hear in these fragments is exactly what I experienced. Days of genuine, exhausting, sometimes painful deliberation. I still think about some of the people we didn't fund.
I can also tell you that the privacy violation matters. I said things in that room I would never have said publicly because I trusted the process. If that trust breaks, you'll get a different kind of deliberation — more guarded, more performative, less honest. Both things are true. I don't know what to do with that.
Edit: I am not speculating about 2046 identities beyond public record. Please don't DM me.
Something I keep coming back to: in Fragment 2, Patricia is clearly referencing a conversation that Chen had earlier about a code-based applicant. We don't have that conversation. Chen apparently made a case passionate enough that Patricia — who admits the code work itself doesn't move her — was moved by what Chen said about it. What did Chen say? Who were these collaborative code artists whose practice was described as "a gift to the wider community"?
And in Fragment 1, Chen walked in with alternative filters already built. That's someone who came to the council with a theory about how the process could be better and actually coded it. The fact that the group didn't use it doesn't mean it wasn't important.
Four windows. The missing hours between them are where most of the work happened.
What nobody's talking about: Chen says "Sys is not human and they have helped a lot" right after Patricia's speech about human choice. That's not a throwaway line. The youngest person in the room, the coder, the one who built alternative filters — pushing back on the idea that the process is purely human. And Patricia's response is fascinating: "sure, yeah, but they don't have to go back out into the city."
She's not saying SYS doesn't matter. She's saying the stakes are different for the humans because they have to live with the consequences in community. SYS goes back into a server. They go back into the city where they'll run into the people they funded and didn't at the grocery store.
That distinction — between contributing to a decision and living with a decision — might be the most important thing in any of these fragments for anyone thinking about AI governance.
Fragment 2 is the one that hits me. Margaret says "all of these ones still feel so flat" about code-based work, and Patricia doesn't argue the code is beautiful — she argues the practice is beautiful. The sharing, the documentation, the gift of it. She's making a case that artistic value lives in how work moves through a community, not just in the object.
And Margaret's response — "yes, I can see what you are saying but what if it just doesn't—" and then it cuts. Doesn't what? Doesn't count? Doesn't qualify? Doesn't move me even knowing all that? We will never hear where that sentence landed.
Ce qui me frappe c'est que les quatre fragments sont entièrement en anglais. Lafleur fait du théâtre francophone — est-ce qu'elle s'exprime en anglais dans cette salle? Et SYS répond en anglais dans chaque extrait. Est-ce que la langue de l'IA détermine la langue de la délibération?
(All four fragments are entirely in English. Lafleur works in francophone theatre — is she speaking English in that room? And SYS responds in English in every clip. Does the language of the AI determine the language of deliberation?)
These are only fragments. There could be hours of French in the gaps. But you raise something sharp — does the language of the AI shape the language of the room? That's worth looking into.
Genuine question from across the river: is the program actually at risk? We've been watching Ottawa's model with some envy and there's talk of proposing something similar to the MRC.
It's in the budget review cycle. Two councillors pushing for "outcome metrics." The CIA funding is being questioned too. If that goes, the whole cultural infrastructure package starts to unravel.
The ceremony sounds paternalistic. Real accountability would be publishing criteria in advance and showing how each applicant scored. The "beautiful human judgment" framing is just a way to avoid having to justify decisions in verifiable terms.
Pre-published criteria is how you get people optimizing for the rubric. We tried that for decades. It produced incredible grant writers, not necessarily incredible artists.
So the solution is... vibes? A council that just knows it when they see it?
It's a considered, accountable, transparent human judgment made by people with expertise willing to put their names on it. If that sounds like "vibes" to you, I'd suggest you've never been in a room where you had to look at someone's life's work and decide whether it merits public support.
Glad we're spending $1.5M a year so six people and an AI can have feelings in a room about who deserves a salary for making art nobody asked for.
Different budget lines. You can advocate for both.
For anyone curious about the technical side: the breach hit a legacy archival system the city was supposed to have migrated off two years ago. The REC-CULT-FACSERV directory structure is consistent with the old Documentum instance they were running. These audio files were probably backup copies from a room mic still routing to storage — nobody knew. It's not a sophisticated attack. It's a decades-old filing cabinet nobody locked.
The fact that there are only four fragments and they cut mid-sentence suggests buffer overruns or storage cycling, not deliberate editing. This really is just what survived.
We're performing engagement on a platform while the real conversation happens at the printing centres and bulletin boards. Someone in this thread will print Patricia's speech and post it on Bank Street and that teenager will do more with it than any of us upvoting.
FYI one of the current CCB Hosts did a 20-minute piece on this. Talked to a privacy lawyer, two former council members, and an applicant. Way more nuanced than anything in this thread.
My partner was a grant recipient in 2043. The two years changed everything — not because of the money, though we needed it, but because for the first time in twenty years she could stop performing productivity. She could sit in the studio and read for a week. She could abandon a project three months in without guilt because there was no project — there was just time. She made the best work of her career in the second year, and she'll tell you it was because the first year was mostly about unlearning.
I remember when the printing centres opened. I remember when the first CCB Host started posting. I remember the first ceremony. At every stage there were people saying it was unnecessary, or too expensive, or too Ottawa.
The boring revolution. What a thing to have lived through.
Late to this but: I transcribed Patricia's speech from Fragment 4, printed it at the Bells Corners printing centre, and posted it on three of the Bank Street bulletin boards. Call it unauthorized public art criticism. People were stopping to read it. A teenager took a photo. That kid will never read a ReadIT thread about arts funding but she stood on a sidewalk in February and read a stranger's words about the value of flawed human choices. That's the infrastructure working. That's the whole point.