24 April 2035

Well then—my first day as data-shepherd for the Malmö Truth Commission.

Apparently keeping a journal of thoughts and experiences related to the project is obligatory, alongside the more ethnographic stuff that goes with our actual data-gathering. This is meant to support the material we collect; I guess the models will use it to adjust for any bias we add to the data in the process of collecting it? I haven’t kept a journal since uni… all that ethnographic training, gone to waste! But they say it’s like riding a bike: you never forget how, but you do get rusty and unbalanced. A very Malmö metaphor…

Mostly induction stuff today. Easy for me, I’ve been working for the kommun for years: it’s the usual organisational boilerplate. I expected more from the talk by the model management team in the afternoon, to be honest—but other than “we don’t call it AI any more”, most of what they had to say was about tools and methods: getting good audio from the recorders, getting the consent forms cleared through BlinkID. And filling in our journals, of course! I guess they’ll tell us more about how they’ll work with the material as the project progresses.

It’s important work, and I’m glad I’m part of it. Things have become so messy in recent years: predictive models, misinformation, political division. The UN’s decision to set up a sort of curated repository of certified data was very welcome, but hell knows how long that’s going to take. Proud that Sweden’s taking the lead and trying to build out local versions first, with the intent of feeding the UN repository when it’s ready. In the meantime, the government reckons that the big firms will pay for access to verifiable data on the population… and that money will pay for the process of collecting it, plus prop up the sagging welfare state a bit.

There’s a lot of skepticism about it, globally and locally, but I think it’s better to do something than not. And honestly, it’s good to have avoided the latest round of redundancies at the kommun... and to get a position that involves getting out of the office on the regular! Spring has kicked in early again, and we’re not allowed to open the windows in case it messes with the air-con…

15 May 2035

I’ve been a bit slack with this journal, but apparently I’m not the only one: the model management team reminded us all to keep on top of it during the monthly meeting this morning. It’s partly lack of practice, but I’ve been lacking motivation. People have been coming in to the interview spaces around the city, but nowhere near the numbers that the management team planned for. Plus the demographic spread isn’t good.

There seems to be some disagreement in the management team. A couple of them said this morning that we should start trying to collect data in situ, actually going to the people rather than expecting them to come to us. There was a theoretical justification that kinda went over my head, but the practical case is that people are busy, and tend to have negative associations with being asked to come to the kommun offices. (I know how they feel!)

Other members of the management team didn’t look at all happy about this, though. I don’t see why; seems pretty logical. Plus it’ll save money if we don’t need to keep renting space for interviews that don’t happen.

18 May 2035

Looks like the in situ lobby won the argument. I volunteered to work Rosengård, because I grew up there—but also, if I’m honest, because they didn’t have many volunteers, and there have already been lay-offs since the start of the project. No harm in keeping myself useful, right?

29 May 2035

The illusion that Rosengård would be easier for me is disappearing fast. I’ve spent a few weeks now going door to door on the big estates, but most people are telling me they’re unable to commit to a time. A few are just outright refusing to be interviewed at all.

(Noticed a new slogan among the usual tags while out researching: VEM KOMMER ATT TITTA PÅ TITTARNA? I’m sure it’s not the intention, but I feel a little watched, myself!)

Gatherers for other neighbourhoods are doing better, apparently—a woman working in Limhamn says she’s actually having to turn away people who want to do a second interview! I bumped into an old friend from my university days, R, and he reminded me of what ethnographers call the ‘snowball’ approach: work hard to win the trust of a few influential people, and they’ll lead you to more sources. Gotta be worth a try.

3 June 2035

Met up with R at the Taproom for a few beers. He hung on in academia after I left; after over a decade, he’s finally got a faculty position at Lund, though it’s only part time. We talked about the project a bit. He thinks it’s fundamentally flawed.

“If more data meant more truth, we’d be drowning in truth by now,” he said. “Really, it’s the exact opposite.” More data means more possibilities, he says, which means more decisions, more deliberations: the truth is a product of interpretation.

I told him I thought postmodernism had died years ago. He laughed and said yes, it did—but postmodernity is shambling along just fine without it. I didn’t really understand the distinction… but hey, it was a hot afternoon, and we were a few beers down. He talked a lot about some French guy called Latour, who said that there’s obviously a real, objective world, but our understanding of that world is in a process of constant collective negotiation. “Meaning isn’t found,” R told me. “It’s made.”

12 June 2035

Emergency meeting this morning; model management team is now half the size. L_ stood up and asked whether they’d quit or been sacked. That information was subject to non-disclosure, they replied. L_ told them she’d submit her notice as soon as she got home, and just walked out. She’s always been a fighter.

Part of me would like to quit as well, to be honest. It’s pretty thankless work, and there are lots of very ugly rumours about the project. It would help if they did a better job of talking about how it works, maybe? But I need the salary.

Anyway, I might have made a breakthrough: I managed to convince B_, an elder in Rosengård’s Serbian community, to sit down for a chat next week. Maybe he can get me moving?

15 June 2035

Meeting with B_ was disheartening. I tried to make the case for the project, explained that the whole point of the Truth Commission is to make sure that the facts on the ground are represented properly—not just in the kommun but in Sweden as a whole, and eventually the world, once the UN gets its act together. He asked me if I’d heard the rumours about the project, and I asked which ones. “Well, exactly,” he laughed.

“But I’m thinking particularly of the one that says that the big companies who will be buying access to the database are paying people to give the answers that the companies would prefer to receive.” “But that… that doesn’t make any sense,” I replied. “Why would they do that? What would it gain them?”“They almost certainly aren’t doing that. But whether it’s factually true doesn’t matter. Put yourself in the shoes of the people living here—the ones who didn’t manage to leave, like you did. Should we believe that trusting technology and giving over more data—literally informing on ourselves, our families and friends and neighbours—will make anything any better? That has never been true for these people; it’s not even plausible. But corporate meddling, the rigging of an already unfair system? We’ve seen that movie before—and whoever is spreading that rumour knows it.”

He wouldn’t be drawn on who he thought was behind it. He kindly offered to ask some contacts if they would speak to me, but told me not to get my hopes up.

Heading home, more graffiti: underneath the rails at Rosengård station, someone had sprayed SANNINGEN KOMMER BEFRIA DIG over Zlatan’s smile.

In a different colour, someone else has crossed out the first word and replaced it with FÖRSENINGEN.

19 June 2035

Beers with R again. Told him about what B_ had said, how irrational and emotional it seemed.

“Of course it’s emotional,” he replied. He had a metaphor I didn’t quite understand, but it went something like this: it’s a fact that two plus two equals four, but it’s the truth that Sven deserves three while Sahima only deserves one.

I told him this was exactly what the project was supposed to fix! I’ve spoken with the model managers quite a few times, and while they use technical words I don’t quite understand, it’s obvious they want to make things better. (I think I’ve met enough real assholes to know the difference.)

R agreed on that point, at least, but said that saying ‘better’ always begs the question, which only confused me more.

“A month ago, you told me the people in charge couldn’t agree on the best way to run the damned project! And these guys, they’re somehow going to have their machine determine the capital-T Truth for the whole city? Good luck with that...”

I said I didn’t think we had time to play philosophy when people’s lives were at stake. Poisoned data isn’t an abstract problem, and it’s making a bad political situation even worse.

“Is it, though?” he asked.

“Maybe we’re mistaking cause and effect—maybe the politics is making the data and the models worse. Look at it this way: I think there’s more logic in our local efforts than there is in the UN’s idea. You and I know very well that Malmö’s truth is not Stockholm’s truth. And neither of them are Staffanstorp’s truth, right? And Sweden’s truth is not Canada’s truth, or the UK’s truth, or South Africa’s truth...

“So the question is: who gets to decide which truth applies where? And that’s the question that underlies all political activity in human history, right? Wanting to produce a better answer to it is a noble goal, I fully agree—but it’s a philosophical question, whether we like it or not. And pretending it isn’t, or assuming that we can outsource political philosophy to machines that, as we know well, can only amplify the biases of their programmers and data-sets… I mean, come on, people were saying this back when we were still doing our first degrees! Emily Bender, Timnit Gebru? Stochastic parrots?

“Ah, whatever,” he laughed. “It’s my own bias, perhaps, as an academic. But I honestly think you’d be better off talking philosophy with the people of Rosengård than asking them for truths to feed to the machines. From what you’ve been telling me, it sounds like the people of Rosengård mostly feel the same way.”

20 June 2035

Sick day. (Hangover, if I’m honest.)

My conversations with B and with R_ have left me feeling unsettled about the project. I want to do something that makes things better, especially for the least fortunate in our society… but do I really know what ‘better’ looks like for them? I’ve booked a meeting with the management team tomorrow, so I can talk to them about this. I’m sure there’s some way that we can incorporate the philosophical issues and the concerns of the citizens into the project, and it will be better for it. More people will want to engage if they feel like their concerns are being heard!

Saw more new graffiti on the soundbarriers of Rosengård station: VAD ÄR EN BERÄTTELSE UTAN EN BERÄTTARE?

# excerpts end

# 22-06-35 / analysis commence

# 23-06-35 / determination: [terminate, full severance pay, NDA]

# 23-06-35 / determination authorised [*REDACTED,REDACTED]*

# 25-06-35 / NDA #2035MTC-930813-8263 signed, filed → ETH blockchain

# ENDS

CLASSIFIED: INTERNAL ONLY SOURCE SCENARIO: “LOCALISED VALIDATED TRUTHS”

In the decade preceding 2035, the world experienced a crisis of truth brought to a head by the expansion of data-sets and the increasing use of models and algorithms to interrogate them. Efforts to use AI systems outside the art and creative spaces became futile as the solutions and answers provided by the system had no resemblance to what was actually happening in the world. The UN was given a mandate to establish a global repository of “ground truth”--a data-set whose careful validation will ensure that AI systems produce valid answers.

As the wheels of the UN turned slowly, local initiatives started as a stop-gap. In 2035 in Malmö, the city administration sends out researchers to talk to people on the ground to understand what their reality looks like; these stories are enriched by images, maps and videos, before being processed by algorithms into something resembling a local wikipedia. This will in turn supply the global repository, in what is described as a “federated approach to truth”, to be funded by commercial entities paying for access to verifiable data. However, not all demographic cohorts are equally engaged by the Malmö project, and many people believe its foundational assumptions to be false.