Industries / Technology / Social media
Creator economy
The creator economy is the work of independent people who make videos, podcasts, newsletters and posts for an audience, and of the platforms and tools that carry that work and pay for it through shared ad money, fans' subscriptions and tips, and brand deals.
2005
On 23 April 2005 Jawed Karim, one of YouTube's three founders, uploaded the site's first video, "Me at the zoo," a short clip of himself at the San Diego Zoo. NPR reports that the founders made it themselves because they were struggling to attract users.
37
Milestones achieved across 6 domains.
11
To watch for: announced, not yet confirmed outside.
13
Open deliverables. Nobody has shipped them.
In one minute
Where it started
YouTube's founders posted its first video in April 2005, and within two years the site was sharing ad money with its most popular amateur uploaders, among them the makers of the scripted video diary lonelygirl15. Fans began paying creators directly through Patreon from 2013, and newsletter readers paid writers through Substack from 2017. Short-video apps came next: Vine closed in 2016 after its stars asked to be paid, and TikTok absorbed Musical.ly in 2018 and set up a fund for creators in 2020.
Where it stands
Goldman Sachs Research counted about 50 million creators worldwide in 2023 and estimated that only about 4% earn more than $100,000 a year; from survey data, it put brand deals at about 70% of creators' income. YouTube said in January 2026 that it had paid more than $100 billion to creators, artists and media companies over the previous four years, a total that includes record labels and studios. Governments now write rules for the work: France passed a law on paid influence in 2023, Illinois and other US states make parents set aside part of a child vlogger's earnings, and China has restricted tipping on live streams since 2022.
What is open
Much of what platforms promise creators has not been checked from outside: whether copies of a video lose reach to the original, how often matching systems wrongly claim public-domain music on apps other than YouTube, how well face matching finds AI-made copies of a creator, and how the rules on ad-friendly videos are applied. Most of these checks can be built from test accounts, consenting volunteers and made-up clips. So can open test sets for tools that crop video for phones or spot hidden ads, a calculator for child-vlogging trust laws, and a disclosure step built into editing software.
Domains
AchievedWatch forOpen
Domain
Progress
Last achieved
Watch for
Open
Making content
Making content is the work of planning, filming, writing and posting videos, short clips, posts and shows, usually by one person or a small team.
5 · 1 · 2
Last achievedChild vloggers get a share2024
Watch forAI-made videos in new feeds
2
openEditing tools
Editing tools are the programs and apps creators use to cut footage, add titles, captions, music and dubbed audio, and shape a video for each platform.
5 · 2 · 2
Last achievedAn editor switched off2025
Watch forRaw footage to first draft
2
openPlatforms and pay
Platforms and pay covers how video and social apps share their ad money with the people who post, and the rules that decide who qualifies and how much each view is worth.
6 · 2 · 2
Last achievedShort videos share ad money2023
Watch forOriginals over their copies
2
openSubscriptions and tips
Subscriptions and tips are payments that fans send straight to a creator, by the month, by the post or during a live stream, with the platform and the payment system each taking a share.
7 · 2 · 2
Last achievedApple's cut of fan pledges2024
Watch forTen billion dollars from fans
2
openBrand deals
Brand deals are the payments, gifts and free products that companies give creators to feature what they sell, and the rules on telling viewers about them.
7 · 2 · 2
Last achievedOne in five disclosed2024
Watch forSponsorships up by half
2
openRights and credit
Rights and credit is who owns a creator's work, who may reuse it, and who is named and paid when it is copied, matched by software or used to train AI.
7 · 2 · 3
Last achievedVideo captions in AI training2024
Watch forFace matching for creators
3
open
Advances to build with
1 research note
Open in creator economy
3 of 13 open deliverables
Open · Making content
Child vlogging pay calculator
A free calculator, with a test suite, that applies each US child-vlogging law to a family channel's own records. For each paid video the user enters its length, the seconds each child appears or is talked about, its gross earnings and its views. The calculator says whether the law's threshold is met, how much belongs in each child's trust, and which section of the law each number comes from. It covers Illinois, Minnesota, California, Montana and Utah.
Built with what exists
the five statutes as published by each legislature; a spreadsheet or a short Python program; a second person who reads the statutes separately and writes their own expected answers.
Open · Editing tools
An open reframing test
A public test set and scoring script for tools that crop wide video into vertical clips. Record 40 one-minute clips with consenting volunteers in common creator setups: one speaker, a two-person interview, a speaker with slides, a cooking table, and a game capture with a face camera. Mark by hand, second by second, who is speaking and which parts of the frame carry information. Run each clip through Adobe Premiere Pro's Auto Reframe, CapCut, OpusClip and Google's open-source AutoFlip, then score the share of seconds in which the speaker's face and the marked areas stay inside the crop.
Built with what exists
MediaPipe and its AutoFlip pipeline; MediaPipe's face detector to find faces in each output frame; ffmpeg to pull frames; Python for the scores; free or trial versions of the commercial tools.
Open · Platforms and pay
Answer key for ad rules
A public set of 500 short test items written for the purpose (a title, a thumbnail description, a script and, where useful, a short clip made from them), each rated by three trained raters under YouTube's advertiser-friendly guidelines as full ads, limited ads or no ads, with the clause that decided it. Use it to score AI models on the same task and publish where the raters themselves disagree. As an extension, upload the clips as private videos to a monetized test channel and record YouTube's own decision for each.
Built with what exists
YouTube's published guidelines; a spreadsheet with a rating form; three raters; open-weight language models such as Mistral Large 4 (covered in the Note "A trillion open parameters") and hosted models; Fleiss' kappa, a standard measure of agreement among raters, from the statsmodels Python library; YouTube Studio for the private-upload extension.
Where it sits
Broader
