Olympics 2026

Graphs

Graph by Event

Graph by Top Country Downloaders

Maps

Tables

Commentary, Questions

The Olympics 2026 sample used 684 input sources and sampled each for 15 weeks, from 2026-02-08 to 2026-05-25.

USA vs. International Downloaders

At five percent of all downloaders, the USA is not a dominant force in this swarm distribution.

This may mean that the USA audience is elsewhere; watching live on terrestrial broadcast stations, on paid cable subscriptions expressedly for the Olympics, watching on YouTube, or otherwise adequetly serviced for media using exising platforms.

See the graph above for swarm composition by country. Russia, China, Korea, Turkey all are bigger parts of the global swarm than the USA.

USA vs. International by Content Creation

Of these input sources, the originating creator can be identified as:

1. 277 BBC (UK), 40%
2. XXX Global (Canada), ~30%
3. 45 ARD (Germany), 6.5%
4. 24 Russian, 3.5%
5. 21 Eurosport, 3%

The absence of content from USA and Japanese sources may indicate that broadcasters in these regions took legal action to take down unauthorized content.

Action item 1 ATDK

IOC licenses the media feed for the Olympics to each national body? Can you detail the licensing of media for the Olympics? Is it a set licene fee per country or is it negotiated per-country? What countries pay? Some color please.

Action item 2 ATDK

Metadata analysis of BTIHA sample set, looking for host languages, subtitles, resolution, creation time, originating source. Frame analysis for watermarks. Can IOC loops be detected with computer vision, ML, AI techniques? TBD. This is a suitable project for University of California (or other) Data Discovery. Do athelete-specific features and intros only appear on USA platforms? Do other countries have region-specific content creation traditions?

Does media of sporting events “decay” faster then film/tv media?
Is there a sharper decrease in audience as the results of the sporting event become known?

The research team’s initial assumption was that sporting events on the internet were “quick” events, events that in real life are at a particular Monday night, or game day, match day, event day. And then faded quickly.

But this does not seem to be the case.

For context, here is the Olympics sample put in with all the other Alpha60 sampling completed in the years 2025 and 2026.

What do these graphs show? The Olympics sampled at the top of the media heap, along with such streaming stars as Fallout, Stranger Things, etc.

If the Olympics curve in the graph above flattened, then that would mean that it became gradually less popular over time compared to the rest of the media objects. But this did not happen: sports, drama, comedy, animation look to have similar patterns over time.

If the audience for sports content is a combintation of both event day and long-tail days in practice, why don’t any streaming platforms support this kind of viewing?

Just curious. Could USA platforms gain international market share by creating long-tail platforms that provide a canonical URL that is durable over time?

AKA, use pirate archival techniques to organize content and use something like

https://ioc.org/2026

as the canonical root for streaming media in perpetuity?

The biggest win for this approach would seem to be in the field of collegiate atheletics. Given the social media and AI harm pressures on students, leaving harm-reduction in the hands of streaming platforms or network administrators seems foolish. Also given the trade portal and increasing commercialization and regulation of student atheletes, a concept for a solution that provided per-school cannonical clips of sporting events, by school athelete (knowing that 4 years in one place will be considerably more rare going forward), and ways to verify of authentic conntent as becoming more and more important as AI deepfakery evolves in sophistication in the future.

By Gender: Open vs. Men vs. Women

Each of the Olympics 2026 input sources was hand-tagged to fit into one of three categories gender: men, women, and open.

1. open, 378 (55%)
2. men, 179 (26%)
3. women 127 (19%)

These results can be graphed (for the largest 5 objects in the slice) as Men in blue, Women in green, and Other as gray for the global data set as:

Action item 1 BTDK

A follow-up question that remains is how these media objects differ by country. Does the USA watch the same women’s sports as Sweden or Japan? Eileen Gu in CHN and Alyssa Liu in USA? Or does Curling really take it all for women?

Olympics as a baseline for comparisons with other media

The Olympics dataset is useful. And more useful over time, with repeated samples every two years providing data about how the internet infrasture is changing year over year. Because of this, the Olympics should be sampled at every opportunity as a big media event that is global in nature. This archival swarm data can be used as a reference swarm to see how the composition of the interent is changing, and how different parts of iternet infracture are emerging and becoming more important.

Here is the reference table for the 2026 Olympics, itemized by internet infrastructure: fiber, wireless, satellite, and two categories of network privacy: hosted (hosting on another machine), and service (using VPN, proxy, or other privaccy-forward services.)

Using the Olympics 2026 sample as the origin data point for comparisons with streaming media objects like Stranger Things 5.1 and the Epstein Leaks hosted by the Internet Archive and Distributed Denial of Secrets, some immediate observations:

• Lack of participation in Africa, Brazil, India.
• China based Olympics swarm smaller than Epstein leaks, which is a pattern also seen in other leak data sets (UKR-RUS Cyberwar).
• Mobile big in Seoul x Japan x Vancouver and Europe.
• Privacy methods being used in USA/Europe mostly, biggest difference is with leaks dataset, again.