User:Spinster/Thoughts about WikiFlix (and dynamic multimedia portals)

WikiFlix is an experimental film portal on Wikimedia Commons: a dedicated portal focused on browsing and watching full-length public domain (and CC-licensed) films, a la (yep) Netflix. This page collects my personal thoughts about it.

To me WikiFlix is absolutely a hobby project, with which I experiment ‘for the common good’. I don’t feel strong ownership of it and will be happy if others jump in, help out, or make it their own. I do have some strong opinions on what I think are good directions, which I will express below.

• Yes, we can do this on Wikimedia projects!
• Yes, it can use a better name!
• Yes, this model can be applied to many other topics!

A short history of WikiFlix

• I (Sandra Fauconnier, User:Spinster) have started WikiFlix at the Wiki Techstorm in Amsterdam on 22-23 November 2019. I have first built it with my own limited knowledge of SPARQL queries and Listeria lists, as a mini-portal on Wikimedia Commons, part of my own user space, with a set of Wikidata-driven gallery pages showcasing cinema history. This has remained a hobby/side project for me and I haven’t given it much attention since.
• I did showcase WikiFlix in WikiSpore, a Wikimedia wiki built for experimentation with new formats.
• A short demo of WikiFlix during the 2020 WikiSpore Day drew the attention of at least one enthusiastic film fan. I have not really followed up on this, sorry!
• In June 2022, a group of Wikimedians proposed a version of WikiFlix as a separate Wikimedia sister project. This proposal was declined. As I was busy with other things at that time, I didn’t really engage in this discussion, but I was really happy to see this interest out of the blue! Personally, I fully agree that it shouldn’t be a separate (new) Wikimedia project. IMO we should have (or invest in) all the infrastructure we need inside/via existing Wikimedia projects already (especially Wikimedia Commons) to build and maintain resources like this, as a community.

Why did I start this project?

I’m actually not a particularly huge fan of old movies! (I don’t hate them either! I like early experimental abstract films and old anime, for instance.)

I am a fan of cultural heritage, and of video and multimedia materials in general. It struck me that we have so many video and audio files on Wikimedia Commons, including probably all highlights of public domain films and music. But they are badly described, inattractive, and usually pretty undiscoverable. Wikimedia Commons is a treasure trove of undiscovered high-quality multimedia, and I want to make it better contextualized and discoverable.

Helping Wikimedians imagine what’s possible: topical browsing and discovery ‘portals’

Presentation during WikiSpore Day 2020 with some not yet very developed thoughts about better discovery and browsing of audiovisual content on Wikimedia projects.

In the Wikimedia community, we are great at creating textual encyclopedic content (Wikipedia), but less good at making other types of content discoverable, even though that other content fully fits the mission of the Wikimedia movement (educational value, notability). A bunch of individual Wikipedia articles and a messy category tree of Commons categories about film history is not the same as a comprehensive portal of freely licensed and public domain film history focused on allowing users to watch all the films, browse, study and compare them. I want to help build the latter. On a more meta level: I want to show the potential of Commons beyond its current function as ‘a database / repository of free files’. I want to help showcase what structured data (and Wikidata) can really do for the Wikimedia projects, if we think big, and if we think out of the box. Structured data can help us build things like:

• Dedicated, topical portals focused on audiovisual / multimedia content
• Searching, browsing, navigating, discovery within a specific topic
• Features like generous interfaces, faceted search, dynamic lists, galleries, word clouds…

Other topics in which such portals could make sense include (but are not limited to):

• Arts and culture in general; a free alternative to Google Arts and Culture. It exists (a bit) as external hobby projects: Crotos, Open Art Browser. But I think we should be able to totally do this on Wikimedia platforms.
  • Cultural sub-topics like
    • History of textiles
    • History of musical instruments
    • Architectural history
    • Public art
• Music. A Spotify of public domain and freely licensed music
• History of trains, ships, airplanes, bike models…

The name is ... eeeh ...

Yeah, the name WikiFlix is not ideal. It's catchy but maybe also problematic re: trademarking etc. I'm not particularly attached to it. Better suggestions are welcome.

How I’d like to build (structure) WikiFlix (and similar portals)

For me, the most interesting aspect of WikiFlix is: building a sustainable, user-managed portal with great access points (browsing / filtering / lists / facets) to discover film history. I would prefer to see this built as part of the Wikimedia ecosystem, because the Wikimedia community and projects possess all the necessary elements (data, media, platforms) to make this happen.

While building this as an external platform or website seems like an attractive idea (more freedom! more technical possibilities!) I think the long-term most sustainable idea is to make it fully part of the Wikimedia movement, community, platforms, and technical ecosystem.

In general, I think the following building blocks make most sense:

• Use Wikidata as the backbone, with basic metadata about all films, and sitelinks to include pointers to / excerpts from Wikipedia articles. Using SPARQL queries and APIs to display content according to certain rules. Also using structured data on Wikimedia Commons.
  • Wikidata's WikiProject Movies is a logical partner here.
• As Wikidata is used as the backbone, this project should ideally be multilingual by design! Information on dynamic pages should fully switch to a different language as the user changes their language settings.
• Video files are mainly stored on Wikimedia Commons. But I can also totally imagine that there can be some form of (clearly delineated) federated linking to / usage of non-free files on e.g. Internet Archive or other external repositories (UbuWeb? others?)

WikiFlix entry wireframe, drawn by an interested Wikimedian (thank you!)

I’m personally a bit less interested in the ‘nitty gritty’ of individual film pages, or typical ‘film viewing’ features like playlists, wishlists, recommendations, etc. They are very important for a film portal! - I am personally just more interested in the ‘navigation and discovery’ aspects. Another Wikimedian has started thinking about these individual film pages, see a sketch on the right.

Pitfalls and bottlenecks

So far, I've encountered the following issues as major bottlenecks to make (specifically) WikiFlix and (more generically) topic portals better:

• Data modeling on Wikidata is inconsistent; and community consensus is often missing (individual Wikidata contributors will model data differently). Data is often not really modeled with broad re-use in mind. Films, for instance, have very widely varying instance of (P31) statements, making pages like this look very messy and (frankly) unprofessional.
  • It can be circumvented by building quite intricate SPARQL queries which will return more 'clean' data, but that's not for the faint of heart.
• Listeria lists, which I currently use, are really just a prototype. Listeria is amazing, but is a volunteer-built initiative with various shortcomings:
  • Limitations on size of lists that will be built (as I understand, also because of limitations of Wikidata's SPARQL endpoint)
  • Limited markup and limited ability to create custom data (can be circumvented with extra row templates, but that's not for the faint of heart)
  • Not multilingual (i.e. upon switching language, users will not suddenly see information switch to the language of their choice)

Random brainstormy notes

Entry points (for WikiFlix)

Which ‘entry points' are interesting for the general public?

• Filming locations near me
• Film directors born near me
• Most important films for my country
• Today, nn years ago
• The first … in … (genre…)
• The most awarded … in … (genre…)
• The most ‘important’ / ‘famous’ … in … genre etc
  • Based on # of Wikipedia articles
• Curated showcases of films that are interesting according to (cool person)
• Very famous actors
• Specific topics/subjects close to people’s hearts, eg LGBTQIA+, colonial history

Which ‘entry points’ are interesting for professionals and people interested in old films in general?

• Alphabetical listings of actors, directors
• Complete listings by year, by country, by genre…
• Faceted search and the possibility to re-sort search results and list pages according to diverse criteria (date, location, length)

Technology ideas

• It would be very nice to experiment with Snowman, a static website generator for SPARQL backends. (This would however then create an external website, while I'd prefer to keep doing stuff on Wikimedia projects.)
• Folks at WikiSpore pointed me to Extension:Cargo (not deeply investigated yet)