Skip to content

How The Internet Is Rigged

2024-08-27 · letter vlog · John · 7:15 · watch on YouTube

Topic: Internet & YouTube · mode: News analysis · tags: timely hook

Replying to Hank's video on AI training, John argues that the user-generated internet has always been rigged against its users: early YouTube creators got nothing while investors got rich, platforms still undervalue creators, and Google's training of language models on their work endangers YouTube's revenue-sharing bargain; he names ways forward and says the future is not inevitable.

What is said

The parts of the video and what is said in each (★ — how much it matters to the video).

00:00:00 — Reply to Hank's AI-training video; the human-learning analogy rejected; thesis that the user-generated internet is rigged

  • ★★ John tells Hank he loved his video about Google and other large companies stealing from independent creators to train generative AI. — 00:00:00 · Hank Green, Google
  • ★★ John notes the counterargument: some say this isn't stealing, since language models ingest web information as people do, just as he might turn reading about tuberculosis into an original book. — 00:00:10 · tuberculosis
  • ★★★ John calls that a nonsense analogy: computers doing what corporations instruct is fundamentally different from humans doing intellectual labor, though he concedes the point is arguable. — 00:00:29

  • ★★★ John's thesis: the user-generated internet has always been rigged against the users who generate it. — 00:00:44

    “the user-generated internet has always been rigged against the users who are generating that internet”

  • ★★ John says he wants to explore the history of this path and why he thinks its future might not be quite as dystopian as many fear. (hedged) — 00:00:50

00:00:58 — Early YouTube: selling out, two kinds of investors, creators who got nothing

  • ★★ John has been on YouTube since January 2, 2007, long before videos could be meaningfully monetized. — 00:00:58 · YouTube
  • ★★ John recalls that in 2006 one of the most-subscribed early YouTube celebrities asked her audience for direct support to buy an editing computer, which many considered selling out. — 00:01:06 · YouTube
  • ★★ Fearing being seen as a sellout, John kept his books out of the background of his videos; the feeling was that nobody should make money from something so weird and non-commercial. — 00:01:27
  • ★★★ John says early YouTube had two kinds of investors: those who put in financial capital and those who put in intellectual property as videos. — 00:01:38 · YouTube
  • ★★ John says money investors in early YouTube became phenomenally wealthy: every $100 invested, rolled into Google stock, is now worth more than $5,000. — 00:01:56 · YouTube, Google
  • ★★★ John says users who invested intellectual property got no return: lonelygirl15's creators and cast, who helped put YouTube on the map, did not become wealthy. — 00:01:56 · YouTube
  • ★★ Had he and Hank stopped making videos in December 2007 as planned, John says, they would have netted nothing from their YouTube project. — 00:02:15 · Hank Green, Brotherhood 2.0, YouTube
  • ★★★ John says we value financial over intellectual capital, yet without lonelygirl15 and Brookers becoming popular in 2005 and 2006, YouTube's investors would never have had their return. — 00:02:26 · YouTube
  • ★★★ John disputes that investors took risks creators didn't: early adopters risk time, talent and treasure on new platforms, and when platforms like Dailymotion or DailyBooth fail, creators get nothing. — 00:02:40

00:02:55 — The culture shift on selling out; platforms still undervalue creators; TikTok's and YouTube's revenue shares

  • ★★ John says creators classically got little even when a platform succeeded, but that has changed somewhat: building an audience on a platform you bet on now generally pays. — 00:02:55
  • ★★ In 2011 John built a large audience on Tumblr, which, as Tumblr grew, sold many copies of The Fault in Our Stars. — 00:03:06 · Tumblr, The Fault in Our Stars
  • ★★ John says Hank bet on TikTok in 2018 and benefited financially when it blew up. — 00:03:14 · Hank Green, TikTok
  • ★★ John says this is partly because platforms got very slightly better at sharing revenue, but mostly because the culture around selling out changed. — 00:03:20
  • ★★ In 2007 John wouldn't even hold up Looking for Alaska on camera; by 2012 he reblogged every piece of Fault in Our Stars fan art with a link to buy the book. — 00:03:24 · Looking for Alaska (novel), The Fault in Our Stars
  • ★★ John says the culture shift on selling out has let many more people make a living, which he considers good news. — 00:03:40
  • ★★★ John says user-generated content remains tremendously undervalued by the platforms that distribute it. — 00:03:45
  • ★★★ John says about 3% of TikTok's revenue goes to independent creators and about 30% to record labels, because labels have far more collective bargaining power. — 00:03:53 · TikTok
  • ★★ Agreeing with Hank's Friday video, John says YouTube differs from TikTok, Reddit, Facebook and Instagram, which take content, monetize it and give users only an audience and karma. — 00:04:06 · Hank Green, YouTube, TikTok, Reddit, Facebook, Instagram
  • ★★★ John says YouTube became a massive business for Google because it decided long ago to share around 55% of revenue with independent creators and treat them much like large corporations. — 00:04:22 · YouTube, Google
  • ★★★ John says that decision let hundreds of thousands of creators make a living, made YouTube the most desirable platform, and produced a breadth and depth of content nobody can match. — 00:04:37 · YouTube

00:04:52 — Google's AI training puts the creator relationship at risk; concentration of power; ways forward

  • ★★ John credits this to a mutually beneficial relationship between independent creators and Google staff with the capital and expertise in ad sales and servers. — 00:04:52 · Google
  • ★★★ John says that relationship is now at risk because Google has, as Hank pointed out, in a very roundabout way basically admitted to stealing the content of creators to train language models without permission. — 00:05:05 · Google, Hank Green
  • ★★★ John says these language models will earn Google billions of dollars while the users whose content made them possible get nothing. — 00:05:21 · Google
  • ★★★ John says this is obviously bad for people who create intellectual value, whether making YouTube videos or writing comments. — 00:05:30 · YouTube
  • ★★★ John says it also harms the user-generated internet and the social order by concentrating capital and power in fewer hands, which history shows tends to work out terribly. — 00:05:40 · the social order, economic inequality
  • ★★ As a first way forward, John says Google, where employees have a lot of power, could see the light, stop taking creators' content and start paying for it. — 00:05:59 · Google
  • ★★ As a second way forward, John says creators could unionize, as Hollywood creators did in the 1920s. — 00:06:06
  • ★★★ John calls it insane that online we are governed not by governments but by corporate terms of service. — 00:06:10

00:06:19 — Eighteen years online: the internet is rigged against creators, but the future is not inevitable

  • ★★★ As a third way forward, John says governments could regulate terms of service so Reddit can't sell users' intellectual property without permission and Google can't use every hosted video for billion-dollar products. — 00:06:19 · Reddit, Google
  • ★ John says he has been online a long time and the channel will soon turn 18. — 00:06:28 · Vlogbrothers
  • ★★★ John says he has long watched people who create value online fail to realize it, because the internet is rigged against independent creators who lack collective bargaining power. — 00:06:33
  • ★★★ John says we feel we must accept the world as it is, but the world will change, each of us has a little say in how, culturally and structurally, and the future is not inevitable. — 00:06:49

Threads in this video

Where the stories that moved here went next

Mentioned

Central: Google · large language model · user-generated content · YouTube

Also: People: Brooke Brodack

Projects: Brotherhood 2.0

Works: lonelygirl15 · Looking for Alaska (novel) · The Fault in Our Stars

Organizations: DailyBooth · Dailymotion · Facebook · Instagram · Reddit · TikTok · Tumblr

Terms: the social order

Concepts: collective bargaining · crowdfunding · economic inequality · generative artificial intelligence · intellectual property · music industry · regulation · sponsorship · terms of service · trade union · tuberculosis

Places: Hollywood

Also centred on large language model: Just Trying to Figure Out My Job

Also centred on user-generated content: Rocky Horror + Glee = ?

Replies to

  • continues · Is Google Training AI on YouTube Videos? — “I really loved your video about how Google and other large companies are stealing from independent creators to train generative artificial intelligence.”

← Previous: Welcome to the Learn-a-Thon · Next → The Two Coolest Things we Know About Sharks