Sidequest: Kolada, the Municipal Spreadsheet of Everything, and the Library File That Came From It
Cold open
Kolada is a free, open database of key figures for every one of Sweden's 290 kommuner and 21 regions, and the file you saw, kolada underscore bibliotek, is a small slice of it: Åre kommun's library numbers going back to 1995. Your guess was right. Kolada really is the Swedish government statistics project, more or less. The "more or less" is where the interesting part starts.
Why it came up today
You were on the citat work, the quotes for edition 44. The session was checking your quote catalogue against Litteraturbanken's scanned books and setting up a hunt across 23 old place names in Åre kommun. Earlier the same day, the parforsk and pargrafik side had been looking at reading and libraries in Åre kommun. Pargrafik's library data folder got three files. The first was kolada bibliotek: 5 library measures, Åre's full series and its rank against everyone else. Its own source notes call it "a one-off, superseded by the next file". The next file is kolada lasning, which has 29 measures: libraries, the citizen survey's questions about the library, Swedish grades in years 3, 6 and 9, and study circles. The third is KB's own raw library statistics. So the name went past you while the session was cleaning up after its own research.
What Kolada actually is
Here is the backstory. Every Swedish kommun gets measured by about 50 different authorities. SCB counts people and houses. Skolverket counts grades. MSB counts fire and rescue call-outs. Kungliga biblioteket counts library loans. Each publishes in its own format, on its own site and its own schedule, with its own code for "Åre". If you were a kommun politician who wanted to know whether your school was worse than the neighbour's, you would have needed a research assistant and a free week.
The roots go back to the early 1990s. A government report from 2005 describes a council set up in 2002 to build and run a municipal database on the internet, and by 2006 that database was permanent. It is run by RKA, the Council for the Promotion of Municipal Analyses. RKA is an odd little creature: a non-profit association whose two members are the Swedish state and SKR, the association of kommuner and regions. The name Kolada is usually explained as a contraction of "kommun- och landstingsdatabasen". That dates it nicely, because landsting were renamed regions in 2019 and the name stayed put. It is a bit like a company still called Telegraph Ltd.
The key point for your work: Kolada produces almost nothing itself. It is an aggregator. It takes figures from SCB, Skolverket, KB, MSB, Naturvårdsverket and others, puts them all on one ID system, usually divides them by population, and serves them through one free API that needs no key and no account. A measure has a dull ID like N09807, and it comes back as a value per kommun, per year, split by gender with a total. A 2018 count put it at about 5,000 key figures from about 50 sources. Pargrafik's notes now say roughly 5,800.
That is why it shows up everywhere in your estate. It is the cheapest way to answer "how does Åre kommun compare with Sweden" for almost any topic.
Where you have already met it
Quite a lot, as it turns out. The traffic page from September used it for kilometres driven per inhabitant: 807 against 625 for Sweden, which became the line about the kommun's cars circling the equator 2,571 times. It also supplied the 374 charging points and the rescue service call-outs to traffic accidents, 3.6 per 1,000 inhabitants against 1.4 nationally. The population page used it for median age and for the end of the building boom: 1.9 new homes per 1,000 inhabitants in 2025, against 26.8 in 2022.
Then there is the parked 5:3 page. Åre, Ale, Hjo, Kil and Eda are the only five three-letter kommuner, and they compete in rounds with their names sized by the numbers. For that one the session pulled every KPI Kolada holds for the five kommuner and ranked them by who wins most lopsidedly. Pargrafik's notes record the lesson: most of the top hits are municipal accounting noise, and the readable ones are rare. With 5,800 measures, plenty of things will look dramatic by accident.
And on the to-do list, Budget 2027, the new coalition's first, is planned as Kolada kronor per inhabitant per area.
The library file, and Sweden's most-visited library
This is the backstory you will enjoy. The library numbers in Kolada come from Kungliga biblioteket. KB has been legally responsible for Sweden's official library statistics since 2011, and it publishes one Excel file per year with about 260 fields for every library. Kolada takes a per-inhabitant subset of that.
In the first library file, Åre kommun ranks number 1 in Sweden for physical library visits per inhabitant in 2011, at 41.8 visits per resident. The runner-up had 20.6. On paper, Åre kommun is not just the country's most bookish kommun, it is twice as bookish as anyone else.
Then the series falls off a cliff. Kolada has nothing for 2015 to 2017, then about 6.5 visits per inhabitant in 2018, and 1.4 in 2022. KB's raw numbers show the same thing. About 350,000 visits a year in 2014 and 2015. Then exactly 75,000 in 2017, 2018 and 2019, the same round number three years running, which looks like an estimate. Loans only fell about 22 percent across the same years, while visits fell 87.
The session's reading, which its notes mark as unverified, is that the old door counter was counting something other than library visitors. A shared entrance or a school, probably, since Åre reports as one combined public and school library. So "most-visited library in Sweden" is most likely a counting artefact. The notes even say not to draw 2014 and 2018 as one line. There was a similar problem earlier: one KB field claimed four out of five books were children's books, and your own look in the library proved it wrong.
The first file had a smaller problem too. Its rankings count about 3,500 entries per year, and Sweden does not have 3,500 kommuner. The newer script keeps only real kommun codes and gets about 289. My reading is that the first pull mixed in rows that were not kommuner, which would be reason enough to retire it. Unverified, but the replacement exists, so it no longer matters.
The two rules that come with Kolada
First: Kolada is the middleman, not the source. Every pargrafik and parforsk note says it: name the real source before print. Your printed source lines already do this, for example "Kolada and MSB" for the rescue figures and "SCB and Kolada" for the population figures. The 5:3 page cannot print until each round's real source has been checked.
Second: the per-inhabitant division is where the surprises come from. Åre kommun has about 12,000 residents and a lot of visitors who are not residents. Anything visitors use, like libraries, charging points, roads or rescue call-outs, gets divided by the residents, so Åre kommun looks extreme. Sometimes that is the story. Sometimes it is just the formula.
On the master prompt
You said the master prompt needs updating for longer episodes with backstories. This one came out at the top of the current range, around 1,200 words. If you want this length as the norm, raise the ceiling, because at the moment the prompt tells me to stop around here.
If you remember one thing: Kolada is the state's and the kommuner's shared shelf of borrowed statistics, perfect for "Åre kommun against Sweden", but every number on it was made by someone else, and a per-inhabitant figure can turn a miscounted door into the most-visited library in Sweden.
Sources: