<?xml version="1.0" encoding="utf-8"?>
<rss version="2.0"><channel><title>Björn Hedin – research notes</title><link>https://hedin.it/research/</link>
<description>Work in progress from KTH: data physicalization, energy, climate, and how people make sense of numbers they can touch.</description><language>en</language>
  <item><title>A globe of where the species are</title><link>https://hedin.it/research/biodiversity-globe/</link>
    <guid>https://hedin.it/research/biodiversity-globe/</guid>
    <pubDate>Mon, 16 Nov 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>dataviz</category><category>biodiversity</category><category>3dprint</category>
    <description>An interactive globe where the land rises with the number of mammal, bird and amphibian species living there. It can also show which mammals would live in a place without humans, and it exports as an 8 cm two-colour 3D print.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/biodiversity-globe.png&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/biodiversity-globe/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/research/img/biodiversity-globe.png&quot; alt=&quot;A globe where land rises with the number of species; the Amazon and the Andes stand highest&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;A globe where land rises with the number of species; the Amazon and the Andes stand highest&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;Most of my globes show climate. This one shows life: on the &lt;a href=&quot;https://hedin.it/biodiversity-globe/&quot;&gt;biodiversity globe&lt;/a&gt; the land rises in proportion to the number of species of mammals, birds and amphibians that live in each place, regardless of national borders.&lt;/p&gt;
&lt;h3&gt;What you see&lt;/h3&gt;
&lt;p&gt;The tallest point is in the western Amazon, near Iquitos in Peru, with about &lt;strong&gt;920 species&lt;/strong&gt; of the three groups together. Stockholm has about &lt;strong&gt;190&lt;/strong&gt;: roughly 35 mammals, 145 birds and 5 amphibians. The central Sahara has under 40.&lt;/p&gt;
&lt;p&gt;The groups peak in different places. Birds are richest on the border between Uganda and DR Congo, with about &lt;strong&gt;650 species&lt;/strong&gt; in one spot. Amphibians peak in the western Amazon at about 130.&lt;/p&gt;
&lt;p&gt;The counts come from range maps on a 10 km grid. I smooth them over 50 km (or 100 km) on land and cut the surface along the coastlines, so the continents stand as plateaus with a vertical edge to the sea.&lt;/p&gt;
&lt;h3&gt;Mammals with and without humans&lt;/h3&gt;
&lt;p&gt;A second mode uses the PHYLACINE database, which covers &lt;strong&gt;5,741 mammal species&lt;/strong&gt; and estimates where each would live today if humans had not affected it. That includes species that died out during the last 130,000 years or so, such as the woolly mammoth. You can switch between today, without humans, and the difference. Click a place to get the list of mammals that live there, and of those that should but do not.&lt;/p&gt;
&lt;h3&gt;Print it&lt;/h3&gt;
&lt;p&gt;The page exports the globe for 3D printing: 8 cm in diameter, in two hemispheres that print flat side down, as a single-colour STL or a two-colour 3MF. With the default setting the richest place rises 8 mm above the sea.&lt;/p&gt;
&lt;h3&gt;Sources and permission&lt;/h3&gt;
&lt;p&gt;Species richness: &lt;a href=&quot;https://biodiversitymapping.org&quot;&gt;BiodiversityMapping.org&lt;/a&gt; (Jenkins, Pimm &amp;amp; Joppa 2013, &lt;em&gt;PNAS&lt;/em&gt; 110: E2602–E2610), based on range maps from the IUCN and BirdLife International. Their data may not be republished without written permission. They gave it for this page, and I am grateful. Mammals then and now: &lt;a href=&quot;https://github.com/MegaPast2Future/PHYLACINE_1.2&quot;&gt;PHYLACINE 1.2.1&lt;/a&gt; (Faurby et al. 2018, &lt;em&gt;Ecology&lt;/em&gt; 99: 2626), which is CC0.&lt;/p&gt;
&lt;p&gt;The page is in English, Swedish and Japanese.&lt;/p&gt;</description></item>
  <item><title>How much electricity will Norrbotten need?</title><link>https://hedin.it/research/electricity-for-norrbotten/</link>
    <guid>https://hedin.it/research/electricity-for-norrbotten/</guid>
    <pubDate>Mon, 16 Nov 2026 09:00:00 +0000</pubDate>
    <category>energy</category><category>dataviz</category><category>dataphys</category><category>3dprint</category>
    <description>The Swedish Energy Agency&#x27;s scenarios for electricity demand per county, 2025 to 2050, as 3D bars next to today&#x27;s production at the same scale. In the highest scenario you can stack the production of ten counties before you reach what Norrbotten alone would need.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/energy-scenarios.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/energiscenarier/?lang=en&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/research/img/energy-scenarios.jpg&quot; alt=&quot;Sweden&#x27;s counties as 3D bars of electricity demand in 2050, with today&#x27;s production stacked beside Norrbotten&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Sweden&#x27;s counties as 3D bars of electricity demand in 2050, with today&#x27;s production stacked beside Norrbotten&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;The Swedish Energy Agency publishes scenarios for how much electricity each of Sweden&#x27;s 21 counties will need up to 2050. They are tables. &lt;a href=&quot;https://hedin.it/energiscenarier/?lang=en&quot;&gt;Energy scenarios per county&lt;/a&gt; turns them into 3D bars on a map, next to a second map with today&#x27;s electricity production per county, drawn at the same scale.&lt;/p&gt;
&lt;h3&gt;The numbers&lt;/h3&gt;
&lt;p&gt;Sweden&#x27;s demand in 2025 is about &lt;strong&gt;128 TWh&lt;/strong&gt;. For 2050 the three scenarios give &lt;strong&gt;226&lt;/strong&gt;, &lt;strong&gt;258&lt;/strong&gt; and &lt;strong&gt;332 TWh&lt;/strong&gt;. Production today (around 2023) is about &lt;strong&gt;165 TWh&lt;/strong&gt;: 70 from hydropower, 47 from nuclear, 34 from wind and 14 from combined heat and power.&lt;/p&gt;
&lt;p&gt;Most of the growth is industry in the north. Norrbotten needs about 9 TWh today. In 2050 the scenarios give 45, 66 or, in the scenario called International Growth, &lt;strong&gt;120 TWh&lt;/strong&gt;. That is close to what the whole country uses now.&lt;/p&gt;
&lt;h3&gt;Stack it&lt;/h3&gt;
&lt;p&gt;A number like 120 TWh is hard to feel, so the page has a button: &lt;em&gt;Is today&#x27;s electricity enough for Norrbotten?&lt;/em&gt; Each click lifts one county&#x27;s current production and puts it on a stack beside Norrbotten, starting in the north and moving south. A glass column shows the demand you are trying to reach.&lt;/p&gt;
&lt;p&gt;In the International Growth scenario for 2050 you use all the hydropower and wind of northern Sweden and keep going. The glass is full when you add Uppsala county with the Forsmark nuclear plant, the tenth county on the stack.&lt;/p&gt;
&lt;h3&gt;Print it&lt;/h3&gt;
&lt;p&gt;The model exports for 3D printing, as one STL per sector or power source and a coloured 3MF. The default scale is the one we used for our printed models: Sweden is 230 mm from north to south and 1 TWh is 2 mm of height. At that scale Norrbotten&#x27;s bar in the highest 2050 scenario is 239 mm, taller than the country is long.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Demand: &lt;a href=&quot;https://www.energimyndigheten.se/energisystem-och-analys/samhallsbyggnad-och-energiplanering/framtida-elbehov-i-ditt-lan/&quot;&gt;Swedish Energy Agency&lt;/a&gt;, regional scenarios. Production: Swedish Energy Agency and Statistics Sweden, around 2023. The scenarios are not forecasts.&lt;/em&gt;&lt;/p&gt;</description></item>
  <item><title>A printable globe from one prompt</title><link>https://hedin.it/research/a-globe-from-one-prompt/</link>
    <guid>https://hedin.it/research/a-globe-from-one-prompt/</guid>
    <pubDate>Fri, 13 Nov 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>climate</category><category>teaching</category><category>3dprint</category>
    <description>At a teacher training day at Vetenskapens hus I showed how a free AI chat account turns a climate dataset into a 3D-printable globe in about four minutes. What that means for teaching with physical data.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/globe-one-prompt.jpg&quot;&gt;&lt;/p&gt;&lt;p&gt;On 5 November I gave a short talk at a teacher training day on climate teaching at Vetenskapens hus in Stockholm, next to the 3D-printed climate globes we opened there in October. The talk, in Swedish, is online as a &lt;a href=&quot;https://hedin.it/s/larardag&quot;&gt;clickable presentation&lt;/a&gt;.&lt;/p&gt;
&lt;h3&gt;Four minutes, one prompt&lt;/h3&gt;
&lt;p&gt;The centrepiece was a recorded demo. I asked Claude, on a free account, in a single prompt, for a 3D globe showing &lt;strong&gt;CO₂ emissions per person for every country&lt;/strong&gt;, with an export for 3D printing. It took &lt;strong&gt;about four minutes&lt;/strong&gt;. The first version was mirrored. One follow-up question fixed it. Here is the whole session, with the waiting sped up (in Swedish, with subtitles):&lt;/p&gt;
&lt;figure&gt;&lt;iframe src=&quot;https://www.youtube-nocookie.com/embed/_IKSdNW0Hhw&quot; style=&quot;width:100%;aspect-ratio:16/11.6;border:0&quot; loading=&quot;lazy&quot; allow=&quot;encrypted-media; picture-in-picture; fullscreen&quot; allowfullscreen title=&quot;YouTube video&quot;&gt;&lt;/iframe&gt;&lt;/figure&gt;
&lt;h3&gt;What changes&lt;/h3&gt;
&lt;p&gt;A few years ago, getting from a dataset to a printable object took a skilled person days. Now the geometry is nearly free, and the hard part moves elsewhere: choosing what to show, what scale to use, and what the object should be compared with. That is a skill teachers and students can practise, and it is where I think the teaching should go.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Presentation: hedin.it/s/larardag. Video: &lt;a href=&quot;https://youtu.be/_IKSdNW0Hhw&quot;&gt;YouTube&lt;/a&gt;.&lt;/em&gt;&lt;/p&gt;</description></item>
  <item><title>Solna&#x27;s emissions as pillars of gas</title><link>https://hedin.it/research/solna-emissions-in-3d/</link>
    <guid>https://hedin.it/research/solna-emissions-in-3d/</guid>
    <pubDate>Fri, 13 Nov 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>climate</category><category>energy</category><category>3dprint</category>
    <description>The 3D map of Solna now shows greenhouse gas emissions: a 1×1 km grid from SMHI and, for every road link, one year of traffic exhaust drawn as a pillar of pure CO₂ gas at true scale.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/solna3d-co2-pillars.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/solna3d/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/research/img/solna3d-co2-pillars.jpg&quot; alt=&quot;Solna in 3D with one year of road-traffic CO₂ drawn as gas pillars on each road; the E4 stands out&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Solna in 3D with one year of road-traffic CO₂ drawn as gas pillars on each road; the E4 stands out&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;&lt;a href=&quot;https://hedin.it/solna3d/&quot;&gt;Solna 3D&lt;/a&gt; started as a printable map of terrain and buildings, and then got &lt;a href=&quot;/research/solna-in-3d/&quot;&gt;energy classes and new detailed plans&lt;/a&gt;. This week it got emissions.&lt;/p&gt;
&lt;h3&gt;Where the emissions are&lt;/h3&gt;
&lt;p&gt;The first new layer is SMHI&#x27;s national emissions database, which publishes greenhouse gases on a 1×1 km grid. Each square becomes a 3D column, and you can switch between years from 1990 to 2024 and between sectors.&lt;/p&gt;
&lt;p&gt;The numbers behind it: Solna&#x27;s territorial emissions were about &lt;strong&gt;205,000 tonnes CO₂e in 1990&lt;/strong&gt; and &lt;strong&gt;99,000 tonnes in 2024&lt;/strong&gt;, roughly half. Transport is now about &lt;strong&gt;three quarters&lt;/strong&gt; of the total (74,000 tonnes), and passenger cars alone account for 51,000.&lt;/p&gt;
&lt;h3&gt;One year of exhaust, as gas&lt;/h3&gt;
&lt;p&gt;A square kilometre is still abstract, so the second layer breaks road traffic down to each road link. I took the road network from OpenStreetMap and traffic counts from Trafikverket&#x27;s road database (NVDB). Where a link has been measured, it uses the measured daily traffic. Elsewhere it uses a typical value for that kind of road. Emission factors are HBEFA-level averages for a mixed fleet, and the whole model is then &lt;strong&gt;calibrated so the sum matches SMHI&#x27;s total for road traffic in Solna&lt;/strong&gt; (about 73,000 tonnes in 2024). The total is official. How it is split between the links is a model.&lt;/p&gt;
&lt;p&gt;Measured links are only 373 of 1,833, but they carry about three quarters of the emissions, because they are the big roads.&lt;/p&gt;
&lt;p&gt;Then the trick: each link&#x27;s emissions for one year are drawn as &lt;strong&gt;a pillar of pure CO₂ gas&lt;/strong&gt; (1.87 kg/m³) standing on the road, as wide as the road and at the same scale as the buildings next to it. On the busiest stretch of the E4 (Uppsalavägen, about 68,000 vehicles a day) that pillar is around &lt;strong&gt;70 metres high&lt;/strong&gt;. The E4 through Solna accounts for about 45% of all modelled road emissions in the municipality.&lt;/p&gt;
&lt;h3&gt;And on the printer&lt;/h3&gt;
&lt;p&gt;The print export now has a four-part option, with &lt;strong&gt;ground, buildings, water and CO₂ pillars&lt;/strong&gt; as separate parts for a four-colour print, sized by default for a 256 mm print bed. The gas pillars become something you can hold next to the houses they pass.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Emissions: SMHI, Nationella emissionsdatabasen (RUS). Traffic: Trafikverket, NVDB. Roads and buildings: © OpenStreetMap contributors (ODbL).&lt;/em&gt;&lt;/p&gt;</description></item>
  <item><title>Housing costs in four colours</title><link>https://hedin.it/research/housing-costs-in-four-colours/</link>
    <guid>https://hedin.it/research/housing-costs-in-four-colours/</guid>
    <pubDate>Fri, 06 Nov 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>energy</category><category>3dprint</category>
    <description>Thirty years of the Nils Holgersson reports as a printable map of Sweden: district heating, electricity, water and waste stacked as four coloured layers, at a fixed 1 mm per 10 kr/m².&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/nils-holgersson-four-colours.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/nils-holgersson/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/research/img/nils-holgersson-four-colours.jpg&quot; alt=&quot;Print model of Sweden: each municipality a column of four stacked cost layers, district heating, electricity, water and waste, with legend and QR code on the base plate&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Print model of Sweden: each municipality a column of four stacked cost layers, district heating, electricity, water and waste, with legend and QR code on the base plate&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;Every year since 1996 the Nils Holgersson reports have compared what it costs to run the same typical apartment block (1,000 m², 15 flats) in every Swedish municipality: district heating, electricity, water and sewage, and waste. I have turned all thirty years into an &lt;a href=&quot;https://hedin.it/nils-holgersson/&quot;&gt;interactive database&lt;/a&gt; with a 3D map, time series and rankings. This week the 3D map learned to print in colour.&lt;/p&gt;
&lt;h3&gt;One layer per cost&lt;/h3&gt;
&lt;p&gt;Each municipality becomes a column on a map of Sweden, and the column is built from four stacked layers, one per filament: district heating in blue at the bottom, then electricity in red, water in white and waste in yellow on top. The base plate is white with blue text. A legend east of Gävle shows the four layers with their names, each printed in its own colour.&lt;/p&gt;
&lt;h3&gt;A fixed scale&lt;/h3&gt;
&lt;p&gt;Height follows a fixed scale, &lt;strong&gt;1 mm = 10 kr per m² and year&lt;/strong&gt;, so two models printed separately can stand next to each other and still be compared. A district-heating-only model and a model of the total share the same height scale.&lt;/p&gt;
&lt;p&gt;In the new 2026 report, the total ranges from &lt;strong&gt;388 kr/m² in Piteå&lt;/strong&gt; to &lt;strong&gt;687 kr/m² in Gnesta&lt;/strong&gt;, with a median of about 529. On the print that is a column of about 39 mm next to one of about 69 mm. District heating on its own ranges from about 142 kr/m² (Luleå) to 323 kr/m² (Älvdalen).&lt;/p&gt;
&lt;p&gt;Over the whole period, the total cost for the typical block has risen about &lt;strong&gt;168% since 1996&lt;/strong&gt;, while consumer prices rose about &lt;strong&gt;63%&lt;/strong&gt;.&lt;/p&gt;
&lt;h3&gt;Small details&lt;/h3&gt;
&lt;p&gt;Each model carries one QR code linking to the matching view in the online database. I tested how small it could get: error-correction level L gives the smallest code (21×21 modules, under 1 mm each), and a raised height of just 0.2 mm on top of a 0.4 mm recess was enough to read it. Higher relief only made it harder to print.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Data: Nils Holgersson reports 1996–2026 (nilsholgersson.nu). Consumer prices: Statistics Sweden (KPI).&lt;/em&gt;&lt;/p&gt;</description></item>
  <item><title>Solna in 3D: terrain, buildings, energy classes and plans</title><link>https://hedin.it/research/solna-in-3d/</link>
    <guid>https://hedin.it/research/solna-in-3d/</guid>
    <pubDate>Fri, 16 Oct 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>energy</category><category>3dprint</category>
    <description>An interactive 3D map of Solna that you can also print: terrain, every building, the energy class of apartment blocks and premises, and the volumes that new detailed plans allow.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/solna3d-energy.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/solna3d/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/research/img/solna3d-energy.jpg&quot; alt=&quot;Southern Solna in 3D, apartment blocks and premises coloured by energy class&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Southern Solna in 3D, apartment blocks and premises coloured by energy class&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;&lt;a href=&quot;https://hedin.it/solna3d/&quot;&gt;Solna 3D&lt;/a&gt; is an interactive 3D map of Solna, just north of Stockholm, built on open data: terrain from elevation data, and about 3,500 buildings with footprints and heights from OpenStreetMap. You can pick the whole municipality, a district, or draw your own rectangle, and export it for 3D printing.&lt;/p&gt;
&lt;h3&gt;Energy classes, building by building&lt;/h3&gt;
&lt;p&gt;This week a new layer went public: the &lt;strong&gt;energy class from each building&#x27;s energy declaration&lt;/strong&gt;, from Boverket&#x27;s register. Boverket gave permission to show it at building level for apartment blocks and commercial premises, with Boverket as the source. Detached houses are left out entirely.&lt;/p&gt;
&lt;p&gt;That covers &lt;strong&gt;1,369 declared addresses&lt;/strong&gt; in Solna: 1,109 apartment blocks and 260 premises. The most common class is &lt;strong&gt;E (543 addresses, about 40%)&lt;/strong&gt;. Only &lt;strong&gt;113 (about 8%) reach A or B&lt;/strong&gt;, and 24 are in class G. Seeing it in 3D makes the pattern spatial: you can see which neighbourhoods are new and efficient and which older blocks are waiting for renovation.&lt;/p&gt;
&lt;h3&gt;What the new plans allow&lt;/h3&gt;
&lt;p&gt;A second new layer shows the future. It reads Solna&#x27;s digital detailed development plans from Lantmäteriet&#x27;s national platform (14 plans so far) and draws the building volumes they permit as clear red shapes next to today&#x27;s buildings. A small job checks for new plans every month.&lt;/p&gt;
&lt;h3&gt;Printed in colour, straight from the browser&lt;/h3&gt;
&lt;p&gt;The export used to give a zip of STL files that you then had to colour one by one in the slicer. Now it can also write a &lt;strong&gt;3MF project&lt;/strong&gt; where buildings, land, water and streets are already separate, pre-coloured parts, so a multi-material printer like a Bambu can print it more or less directly.&lt;/p&gt;
&lt;p&gt;The point is the same as in my other data-physicalisation work: a printed piece of your own neighbourhood is something people pick up, turn around and argue about, in a way they rarely do with a map on a screen.&lt;/p&gt;
&lt;p&gt;&lt;em&gt;Energy data: Boverket, energy declaration register. Buildings and boundaries: © OpenStreetMap contributors (ODbL).&lt;/em&gt;&lt;/p&gt;</description></item>
  <item><title>A rain map that updates itself</title><link>https://hedin.it/research/rain-map/</link>
    <guid>https://hedin.it/research/rain-map/</guid>
    <pubDate>Fri, 09 Oct 2026 09:00:00 +0000</pubDate>
    <category>climate</category><category>dataviz</category><category>dataphys</category>
    <description>Monthly precipitation as 3D surfaces on the curved Earth, side by side with the 1961–1990 normal. The Swedish map now refreshes itself every month as soon as new ERA5 data is out.&lt;p&gt;&lt;img src=&quot;https://hedin.it/rainmap/preview.png&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/rainmap/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/rainmap/preview.png&quot; alt=&quot;Three synced 3D views of Sweden: the 1961–1990 normal, the latest month, and the difference&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Three synced 3D views of Sweden: the 1961–1990 normal, the latest month, and the difference&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;&quot;Was that a wet month?&quot; is a surprisingly hard question to answer by feel. The &lt;a href=&quot;https://hedin.it/rainmap/&quot;&gt;rain map&lt;/a&gt; answers it by putting three views of a country next to each other: the &lt;strong&gt;normal&lt;/strong&gt; for that month (1961–1990), the &lt;strong&gt;actual&lt;/strong&gt; month, and the &lt;strong&gt;difference&lt;/strong&gt;. All three are drawn as 3D surfaces on the Earth&#x27;s real curvature, one bar per 0.25° grid cell from the ERA5 reanalysis.&lt;/p&gt;
&lt;p&gt;In the difference view, zero is a water surface. Places that were drier than normal sink below it, wetter places stick up. You can drag to rotate all panels at once, play through the months since 2016, or put two countries side by side on a shared height scale.&lt;/p&gt;
&lt;h3&gt;It keeps itself current&lt;/h3&gt;
&lt;p&gt;Sweden is now updated automatically. Every morning a small script checks whether the Copernicus Climate Data Store has published a new month (usually around the 5th–6th). When it has, the map is rebuilt, published and checked. September 2026 went live on 7 October without anyone touching it.&lt;/p&gt;
&lt;p&gt;So what did September look like? Averaged over Sweden, about &lt;strong&gt;74 mm&lt;/strong&gt; of rain against a normal of about 77 mm, so roughly &lt;strong&gt;95% of normal&lt;/strong&gt;: an ordinary month. The year as a whole was anything but ordinary. April got about half its normal rain and June about one and a half times.&lt;/p&gt;
&lt;p&gt;Temperature is in the same tool. September 2026 was about &lt;strong&gt;1.9 °C warmer&lt;/strong&gt; than the 1961–1990 September normal. That was only the fifth warmest September in the series. In fact, &lt;strong&gt;all eleven Septembers from 2016 to 2026&lt;/strong&gt; were warmer than that normal.&lt;/p&gt;
&lt;h3&gt;Rain you can hold&lt;/h3&gt;
&lt;p&gt;Like most of my tools, the map can be exported for 3D printing: either the month in four colour layers (0–50%, 50–100%, 100–150% and over 150% of normal), or the difference as a &quot;bathtub&quot; whose rim is the normal level, with dry areas sunk below it and wet areas rising above.&lt;/p&gt;</description></item>
  <item><title>What worked at ForskarFredag: half-globes beat paragraphs</title><link>https://hedin.it/research/what-worked-at-forskarfredag/</link>
    <guid>https://hedin.it/research/what-worked-at-forskarfredag/</guid>
    <pubDate>Fri, 02 Oct 2026 09:00:00 +0000</pubDate>
    <category>climate</category><category>dataphys</category><category>education</category>
    <description>Notes from showing 3D-printed climate models to the public at ForskarFredag 2026. The Arctic sea ice half-globes needed no explanation, the layered Sweden models needed a legend, and our printed info cards had far too much text.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/forskarfredag-table.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;img src=&quot;https://hedin.it/research/img/forskarfredag-table.jpg&quot; alt=&quot;Our table at ForskarFredag: Sweden and Europe temperature models, the global temperature ridgeline and Arctic sea ice for 15 September 1980 and 2025&quot; loading=&quot;lazy&quot;&gt;&lt;figcaption&gt;Our table at ForskarFredag: Sweden and Europe temperature models, the global temperature ridgeline and Arctic sea ice for 15 September 1980 and 2025&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;On 25 September we brought a table of 3D-printed climate models to ForskarFredag, the Swedish researchers&#x27; night. It was a good test of what printed data actually does in front of a walk-up audience: people who have a minute or two, no instructions, and no reason to be polite to a chart.&lt;/p&gt;
&lt;p&gt;What we brought:&lt;/p&gt;
&lt;ul&gt;&lt;li&gt;&lt;strong&gt;Sweden models&lt;/strong&gt; of temperature change, one layer per period (the same set as earlier years, this time without the boxes that used to hide them).&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Three Europe models&lt;/strong&gt;: mean change, the warmest June-day anomalies and the coldest December-night anomalies.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Sweden as a clock&lt;/strong&gt;: one month at each hour position.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Half-globes of Arctic sea ice&lt;/strong&gt; for September 1980 and September 2025, from the &lt;a href=&quot;https://hedin.it/arcticicethickness/&quot;&gt;sea ice thickness&lt;/a&gt; tool.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;A ridgeline stack&lt;/strong&gt; of daily global temperature anomalies since 1940, from &lt;a href=&quot;https://hedin.it/klimatskivor/&quot;&gt;Klimatskivor&lt;/a&gt;.&lt;/li&gt;&lt;/ul&gt;
&lt;h3&gt;What worked&lt;/h3&gt;
&lt;p&gt;&lt;strong&gt;The sea ice half-globes needed no explanation at all.&lt;/strong&gt; Two Arctics side by side, same scale, and people simply got it. In the PIOMAS model the Arctic held about 16,000 km³ of ice in September 1980 and about 4,000 km³ in September 2025, roughly a quarter. You don&#x27;t need to say the number; the two objects say it.&lt;/p&gt;
&lt;p&gt;I think the reason is that the encoding is literal. Ice is shown as ice, on a globe, where the Arctic is. There is nothing to decode.&lt;/p&gt;
&lt;h3&gt;What didn&#x27;t, quite&lt;/h3&gt;
&lt;p&gt;&lt;strong&gt;The layered Sweden models worked, but raised the question &quot;what do the layers mean?&quot;&lt;/strong&gt; Anything that uses height or layers for something other than height needs its legend &lt;em&gt;on the object&lt;/em&gt;, or at least on a small card next to it. This is also an argument for the QR code on the underside that the newer models carry.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Our info cards had too much text.&lt;/strong&gt; Fine for a classroom, wrong for an event where people stop for ninety seconds. Next time: one sentence and one number per model.&lt;/p&gt;
&lt;h3&gt;Next&lt;/h3&gt;
&lt;ul&gt;&lt;li&gt;A family of Nordic models at the same scale, showing topography, precipitation and temperature, so visitors can look for the connections themselves.&lt;/li&gt;&lt;li&gt;Questions printed next to the models (&quot;which month has warmed most?&quot;), and maybe a puzzle format where solving a task gives you a code to a lock, so that the challenge is tied to reading the data rather than to guessing.&lt;/li&gt;&lt;li&gt;At least two geography teachers were interested, so social-science teachers may be a good audience to talk to.&lt;/li&gt;&lt;/ul&gt;</description></item>
  <item><title>A year of global temperature as a printed disc</title><link>https://hedin.it/research/climate-discs/</link>
    <guid>https://hedin.it/research/climate-discs/</guid>
    <pubDate>Fri, 25 Sep 2026 09:00:00 +0000</pubDate>
    <category>climate</category><category>dataphys</category><category>3dprint</category>
    <description>Klimatskivor turns every year since 1940 into a printable disc showing how its days were distributed against pre-industrial temperature. It&#x27;s a ridgeline chart you can hold, with a QR code on the bottom that links to the exact model.&lt;p&gt;&lt;img src=&quot;https://hedin.it/klimatskivor/preview.png&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/klimatskivor/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/klimatskivor/preview.png&quot; alt=&quot;Stacked discs, one per year, showing the distribution of daily global temperature anomalies&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Stacked discs, one per year, showing the distribution of daily global temperature anomalies&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;&lt;a href=&quot;https://hedin.it/klimatskivor/&quot;&gt;Klimatskivor&lt;/a&gt; (&quot;climate discs&quot;) makes printable 3D models from daily climate series. Until this week each year was a flat disc with a wavy top edge: the days of the year left to right, temperature as height.&lt;/p&gt;
&lt;p&gt;The new default view asks a different question: &lt;em&gt;how were the year&#x27;s days distributed?&lt;/em&gt; Temperature now runs left to right, and height shows how many days of the year landed there. Stack the years front to back and you get the familiar ridgeline chart of daily anomalies (Copernicus publishes one), except that this one is made of plastic. Seen straight on at eye level, the model &lt;em&gt;is&lt;/em&gt; that chart.&lt;/p&gt;
&lt;h3&gt;What the discs show&lt;/h3&gt;
&lt;p&gt;The reference is now pre-industrial temperature, not 1991–2020. ERA5 starts in 1940, so the pre-industrial level is taken as 1991–2020 plus Copernicus&#x27; offset of 0.88 °C. The scale has a heavy mark at 1.5 °C. Counting days of daily global air temperature (ERA5) at or above 1.5 °C:&lt;/p&gt;
&lt;p&gt;| Year | Days ≥ 1.5 °C | Mean | |---|---|---| | 2000 | 0 | 0.63 °C | | 2015 | 11 | 1.14 °C | | 2020 | 33 | 1.31 °C | | 2023 | 186 | 1.48 °C | | 2024 | 282 of 366 | 1.60 °C | | 2025 | 154 | 1.47 °C |&lt;/p&gt;
&lt;p&gt;On the model, those years stand to the right of the mark, and the older discs cluster well to the left of it. The 0.88 °C offset is meant for annual means, so applied to single days it is an estimate.&lt;/p&gt;
&lt;h3&gt;Design notes&lt;/h3&gt;
&lt;ul&gt;&lt;li&gt;&lt;strong&gt;Lower target height&lt;/strong&gt; (4 cm instead of 10). A ridgeline works because one peak hides a couple of years behind it, not the whole stack.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Tails are clipped&lt;/strong&gt; where density falls below half a percent, so thin spikes don&#x27;t snap off.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;A QR code on the bottom&lt;/strong&gt; links to the exact version you are holding: the series, the reference and the view, with a short readable text beside it. When a model gets passed around a classroom, someone can always find out what it shows.&lt;/li&gt;&lt;/ul&gt;
&lt;p&gt;Try it at &lt;a href=&quot;https://hedin.it/klimatskivor/&quot;&gt;hedin.it/klimatskivor&lt;/a&gt;. The older time view is still available at &lt;code&gt;?vy=tid&lt;/code&gt;.&lt;/p&gt;</description></item>
  <item><title>Starting a research notebook</title><link>https://hedin.it/research/a-research-notebook/</link>
    <guid>https://hedin.it/research/a-research-notebook/</guid>
    <pubDate>Tue, 22 Sep 2026 09:00:00 +0000</pubDate>
    <category>meta</category>
    <description>Why I&#x27;m starting to post short notes about work in progress, and what you can expect to find here.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/blog-preview.png&quot;&gt;&lt;/p&gt;&lt;p&gt;Most of what I build ends up as a page on &lt;a href=&quot;https://hedin.it/&quot;&gt;hedin.it&lt;/a&gt; and then quietly sits there. The thinking behind it (why a globe and not a map, why the ice is printed as a puck, what went wrong on the first three prints) mostly stays in my notes.&lt;/p&gt;
&lt;p&gt;This notebook is my attempt to change that. Roughly once a week I&#x27;ll post short notes on what I&#x27;m working on at KTH:&lt;/p&gt;
&lt;ul&gt;&lt;li&gt;&lt;strong&gt;Data physicalization&lt;/strong&gt;: turning climate, energy and economic data into objects you can hold, and figuring out when that actually helps people reason about the numbers.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Feel the Energy (FETE)&lt;/strong&gt;: our research project on tactile representations of household energy use. The project is heading into its final phase, and its earlier blog at &lt;a href=&quot;https://hedin.it/fete/&quot;&gt;hedin.it/fete&lt;/a&gt; now continues under the tag &lt;a href=&quot;../topic/fete/&quot;&gt;Feel the Energy&lt;/a&gt;.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Climate and energy visualization&lt;/strong&gt;: interactive web tools, most of them with an export to 3D printing.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Engineering education&lt;/strong&gt;: teaching sustainability to future engineers and teachers.&lt;/li&gt;&lt;/ul&gt;
&lt;p&gt;Every post is tagged, and every tag has its own RSS feed, so you can follow just the parts you care about. Short versions go out on Bluesky as &lt;a href=&quot;https://bsky.app/profile/hedin.it&quot;&gt;@hedin.it&lt;/a&gt;.&lt;/p&gt;
&lt;p&gt;Much of the building happens together with AI coding assistants these days. That&#x27;s a research topic in itself, and I&#x27;ll write about it now and then too.&lt;/p&gt;</description></item>
  <item><title>Arctic sea ice, any two years since 1979</title><link>https://hedin.it/research/arctic-sea-ice-any-two-years/</link>
    <guid>https://hedin.it/research/arctic-sea-ice-any-two-years/</guid>
    <pubDate>Tue, 22 Sep 2026 09:00:00 +0000</pubDate>
    <category>climate</category><category>dataviz</category><category>dataphys</category>
    <description>Two globes side by side showing daily Arctic sea ice thickness from PIOMAS. Pick any two years from 1979 to 2025, play them day by day, and print the ice as a 3D puck.&lt;p&gt;&lt;img src=&quot;https://hedin.it/arcticicethickness/preview.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/arcticicethickness/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/arcticicethickness/preview.jpg&quot; alt=&quot;Two globes comparing Arctic sea ice thickness in 1980 and 2025&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Two globes comparing Arctic sea ice thickness in 1980 and 2025&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;Sea ice &lt;em&gt;extent&lt;/em&gt; gets the headlines, but thickness is where the real loss shows up. An Arctic covered by thin first-year ice can look similar to one covered by thick multi-year ice in a satellite photo, yet it holds far less ice.&lt;/p&gt;
&lt;p&gt;The new version of &lt;a href=&quot;https://hedin.it/arcticicethickness/&quot;&gt;Arctic sea ice thickness&lt;/a&gt; shows daily ice thickness from the PIOMAS reanalysis on two globes side by side. Until now it compared two fixed years. Now you can pick &lt;strong&gt;any two years from 1979 to 2025&lt;/strong&gt; and play them day by day. Each globe also shows the ice &lt;strong&gt;volume&lt;/strong&gt; for that day, and the right globe shows how much it differs from the left. The comparison lives in the URL, so a specific pair of years can be shared as a link, for example &lt;a href=&quot;https://hedin.it/arcticicethickness/?a=1980&amp;amp;b=2025&quot;&gt;1980 vs 2025&lt;/a&gt;. On 15 January 1980 the Arctic held about 26,400 km³ of sea ice. On the same day in 2025 it held 15,200 km³, 42% less.&lt;/p&gt;
&lt;h3&gt;Making it fit in a browser&lt;/h3&gt;
&lt;p&gt;PIOMAS raw files are 17–21 MB per year, 850 MB in total. By storing thickness in 4 cm steps (at most 2 cm off, which is 0.06 mm in a print), keeping only cells that have ice, and storing each day as a difference from the day before, a year shrinks to just over 1 MB. The volume computed from the compact format agrees with the official PIOMAS daily volume to within half a percent.&lt;/p&gt;
&lt;h3&gt;Ice you can hold&lt;/h3&gt;
&lt;p&gt;Each globe can now be exported as a &lt;strong&gt;3D-printable puck&lt;/strong&gt; in four parts (ocean, ice, coastlines and a date label) so it can be printed in several colours. Two pucks from different decades next to each other on a table make a very different argument than two lines on a chart. Why that is, and when it isn&#x27;t true, is a big part of what I study in data physicalization.&lt;/p&gt;
&lt;p&gt;The page is available in English, Swedish and Japanese.&lt;/p&gt;</description></item>
  <item><title>Relief globes, now open source</title><link>https://hedin.it/research/relief-globes/</link>
    <guid>https://hedin.it/research/relief-globes/</guid>
    <pubDate>Mon, 14 Sep 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>education</category><category>3dprint</category>
    <description>An interactive globe where land rises and the sea floor sinks, exportable as two- or three-colour 3D prints. The code is now on GitHub under an MIT license.&lt;p&gt;&lt;img src=&quot;https://hedin.it/relief-globes/preview.png&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/relief-globes/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/relief-globes/preview.png&quot; alt=&quot;Relief globe with exaggerated mountains and ocean depths&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;Relief globe with exaggerated mountains and ocean depths&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;&lt;a href=&quot;https://hedin.it/relief-globes/&quot;&gt;Relief globes&lt;/a&gt; started as a simple question from teaching: what does the Earth look like if you exaggerate the height enough to actually &lt;em&gt;see&lt;/em&gt; the mid-ocean ridges and the trenches?&lt;/p&gt;
&lt;p&gt;The page draws the ETOPO elevation data on a globe. You can show full relief, flatten the sea, or flatten the land, and choose the vertical exaggeration (the default is now ×250). The globe exports as &lt;strong&gt;two STL files, one for ocean and one for land&lt;/strong&gt;, so it prints in two colours on an ordinary multi-material printer. Since last week, country borders can be shown too, and printed as a thin third part.&lt;/p&gt;
&lt;p&gt;The code is now &lt;a href=&quot;https://github.com/datamogulen/reliefglober&quot;&gt;public on GitHub&lt;/a&gt; under the MIT license. Use it, change it, print it in class. I just ask that you credit the source.&lt;/p&gt;
&lt;p&gt;It runs on phones as well. The camera now adapts to the space next to or above the control panel, which turned out to matter more than anything else for mobile use.&lt;/p&gt;</description></item>
  <item><title>Printing climate globes in two halves</title><link>https://hedin.it/research/globes-in-two-halves/</link>
    <guid>https://hedin.it/research/globes-in-two-halves/</guid>
    <pubDate>Thu, 03 Sep 2026 09:00:00 +0000</pubDate>
    <category>dataphys</category><category>climate</category><category>3dprint</category><category>fete</category>
    <description>Colleagues asked for large printed climate globes. Making that work meant splitting the globe at the equator, adding a peg hole, and capping the extremes.&lt;p&gt;&lt;img src=&quot;https://hedin.it/climate-globes/preview.jpg&quot;&gt;&lt;/p&gt;&lt;figure&gt;&lt;a class=&quot;imglink&quot; href=&quot;https://hedin.it/climate-globes/&quot; title=&quot;Open the project&quot;&gt;&lt;img src=&quot;https://hedin.it/climate-globes/preview.jpg&quot; alt=&quot;A climate globe where relief shows change between two 30-year periods&quot; loading=&quot;lazy&quot;&gt;&lt;/a&gt;&lt;figcaption&gt;A climate globe where relief shows change between two 30-year periods&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;On the &lt;a href=&quot;https://hedin.it/climate-globes/&quot;&gt;climate globes&lt;/a&gt;, the relief shows the &lt;em&gt;difference&lt;/em&gt; between two 30-year climate periods (1995–2024 versus 1961–1990, from ERA5): warming or wetter becomes mountains, cooling or drier becomes valleys. Up and down use the same scale, so a millimetre means the same thing in both directions.&lt;/p&gt;
&lt;p&gt;Colleagues wanted four of them in print, larger than before: change in precipitation (which grows a striking &quot;cockscomb&quot; along the tropical rain belt), energy use per capita, 30-year temperature change, and CO₂ per capita. A large sphere doesn&#x27;t fit on an ordinary printer, so this round of work was mostly about manufacturing:&lt;/p&gt;
&lt;ul&gt;&lt;li&gt;&lt;strong&gt;Split at the equator.&lt;/strong&gt; The export now writes two hemispheres at print size, each with a flat face that sits on the build plate.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Peg hole.&lt;/strong&gt; The southern half can get a hole of any diameter (2 cm by default), so the globe can sit on a stand. A bug here made the STL silently non-watertight, which is the kind of error you only discover after eight hours of printing.&lt;/li&gt;&lt;li&gt;&lt;strong&gt;Caps and top layers.&lt;/strong&gt; A few extreme values (tiny, rich countries per km², for example) would turn into needles that snap off. A height cap flattens them, and whatever the cap cuts off gets its own colour layer so the information isn&#x27;t hidden.&lt;/li&gt;&lt;/ul&gt;
&lt;p&gt;The same export options now work on the &lt;a href=&quot;https://hedin.it/owid-explorer/&quot;&gt;OWID explorer&lt;/a&gt;, which puts over a thousand Our World in Data series on a globe.&lt;/p&gt;
&lt;p&gt;There&#x27;s also a new stripped-down demo mode for phones and exhibitions: just the globe and a play button, with the whole setup in the link.&lt;/p&gt;</description></item>
  <item><title>A coin meter for electricity</title><link>https://hedin.it/research/elpolett/</link>
    <guid>https://hedin.it/research/elpolett/</guid>
    <pubDate>Wed, 02 Sep 2026 09:00:00 +0000</pubDate>
    <category>fete</category><category>energy</category><category>tangible</category>
    <description>Elpolett borrows the design of 1950s prepaid gas meters. A column of lights shows how much electricity money is left, and one light goes out for every krona you use.&lt;p&gt;&lt;img src=&quot;https://hedin.it/research/img/elpolett-wide.jpg&quot;&gt;&lt;/p&gt;&lt;p&gt;In the 1950s many Swedish homes had prepaid gas meters. You put in a coin, and when the money ran out, so did the gas. Nobody needed a chart to understand their consumption: you could &lt;em&gt;feel&lt;/em&gt; it in your pocket.&lt;/p&gt;
&lt;p&gt;&lt;strong&gt;Elpolett&lt;/strong&gt; (&quot;electricity token&quot;) is an experiment in bringing that feeling back for electricity, as part of the &lt;a href=&quot;../topic/fete/&quot;&gt;Feel the Energy&lt;/a&gt; project. A column of lights shows how much of a small balance is left. For every krona of electricity used, one light goes out from the top. At zero, a smart plug cuts the power, and a button tops it up one krona at a time.&lt;/p&gt;
&lt;p&gt;Getting the cost right is harder than it sounds. Swedish household electricity has five parts: the hourly spot price, the retailer&#x27;s markup, the grid transfer fee (with peak-hour rules), energy tax and VAT. The spot price alone, which is what most apps show, is often less than half the total. Elpolett prices every minute of consumption with the full model, using live readings from an energy meter.&lt;/p&gt;
&lt;p&gt;Right now it&#x27;s a web prototype that also runs as an app on the phone, which lets us test the concept for real at home. The goal is the same logic on a Raspberry Pi with real LEDs, a buzzer and a physical button. A companion widget shows the day&#x27;s spot price at quarter-hour resolution without draining the phone&#x27;s battery.&lt;/p&gt;
&lt;figure&gt;&lt;img src=&quot;https://hedin.it/research/img/elpolett-phone.jpg&quot; alt=&quot;Elpolett on a phone home screen: the app icon and the widget with the column of lights, the remaining balance (12.81 kr), the current full price per kWh and the power right now&quot; loading=&quot;lazy&quot;&gt;&lt;figcaption&gt;Elpolett on a phone home screen: the app icon and the widget with the column of lights, the remaining balance (12.81 kr), the current full price per kWh and the power right now&lt;/figcaption&gt;&lt;/figure&gt;
&lt;p&gt;The design question we&#x27;re interested in: does a &lt;em&gt;budget that runs out&lt;/em&gt; change behaviour differently from a &lt;em&gt;meter that counts up&lt;/em&gt;?&lt;/p&gt;</description></item>
</channel></rss>
