---
title: "friendsof.wiki: a quick changelog"
date: 2026-08-20
description: "Some quick notes on a couple of weeks of small-bore hacking."
tags: [wikipedia, wikidata, open-knowledge, commons, glam]
draft: false
---

I've updated [friendsof.wiki](https://friendsof.wiki). Quick changelog since [my last blog post](https://lu.is/2026/08/wikipedia-and-friends/):

![The 1903 Wright Flyer on display: a wood biplane with fabric-covered wings, a figure lying face down on the lower wing, and interpretive panels and historic photographs along the surrounding gallery walls](/images/wright-flyer-hero.jpg)

*The [1903 Wright Flyer](https://airandspace.si.edu/collection-objects/wright-flyer/nasm_A19610048000) on display at the [National Air and Space Museum](https://airandspace.si.edu/), [CC0](https://creativecommons.org/publicdomain/zero/1.0/). One of 19 openly licensed images on the museum's own digital record of the aircraft.*

- **Added an FAQ.** Might (or might not!) be a little more clear what I'm trying to do here. TLDR: mostly experiment and learn about how open data works (and doesn't) in the wild in 2026.

- **Single objects, held at a single museum, merge the museum data into infoboxes.** As a different approach from the round of experiments, which hoovered up as much data from as many different places as possible, I wanted to see what it would look like if we trust and prioritize the institution that holds a specific object. For [The Night Watch](https://friendsof.wiki/wiki/The_Night_Watch), [the Spinola Hours](https://friendsof.wiki/wiki/Spinola_Hours), or [the Hours of Jeanne d'Evreux](https://friendsof.wiki/wiki/Hours_of_Jeanne_d%27Evreux), the page now leads with the museum's own catalog record and every other partner sits the page out. I don't think that is a good long-term position (many other partners have information about Rembrandt, for example!) but it's been a good thought experiment for me, and I learned a lot about how infoboxes can/can't interact with Wikidata that I hope to be able to put into practice soon.

- **Fixed how I show copyright information.** Getting copyright correct is important to Wikipedians, for good reason; I tried to make it easier to see what we know from "pure" metadata, including where the upstream source and Wikidata conflict. I already knew this was going to be a point of conflict, so not necessarily much learning from this.

- **I added glTF support.** Wikidata doesn't have a ton of glTF data points, basically all from the Smithsonian. I added support for species articles, like [the common seadragon](https://friendsof.wiki/wiki/Common_seadragon). Learning: there is some nifty stuff out there! would be nice to get this onto the main Wikipedia.

- **I added DigitalNZ as a data source.** Based on a discussion with Stuart Yeates, I added DigitalNZ as a source. It helps some New Zealand articles, but is high noise on others (like Apollo 11). Shows that if we want to go this route at scale, curation looks different -- but is still very important.

- **I probed IIIF more.** IIIF is still pretty limited, because many of the IIIF partner institutions have challenges--eg one had a broken SSL cert; another had broken licensing data (in an otherwise exemplary data set). But where it is present, it can be rich. This is a shame; it'd be great to support this better on-wiki, but it may be hard to justify.

- **Closed approximately 200 SonarSource issues, including 24 complexity issues.** I genuinely enjoy using SonarSource (disclaimer: my employer) with LLM-generated code. It's clear these tools can over"think" code. So when Sonar spots a complexity issue and the bot fixes it, I feel a lot better about the long-term maintainability of the codebase.

I'm probably going to let this sit for a bit; I think most of the next steps are to do with learnings of how to start bringing this back into Wikidata -- and from there onto the real wiki -- in an impactful way.