YouTube Plagiarism and Generation Loss
Keep making copies of copies of copies.
Hello, readers. Parker here. I hope you all had a good weekend.
One of my favorite YouTubers (Hbomberguy1, née Harris Brewis) released a new 3-hour and 51-minute video titled “Plagiarism and You(Tube).” In it, he shines a light on an epidemic of intellectual dishonesty and outright plagiarism in the online video essayist space.
We’re coming up on the end of the year, and I’d absolutely love it if I could hit the 40,000 free subscriber benchmark before then. If you haven’t yet subscribed, please consider signing up for free. Thanks!
Cycling between examples of people creating entire videos out of unattributed quotes from journalists to the scourge of online content mills to the direct adaptation of someone else’s work without the addition of any additional reporting, Brewis reflects on how YouTube’s somewhat perverse incentive structures reward those creators.
It’s a great video. I loved it and hope it leads to a renewed interest in quality over quantity on YouTube2.
Setting aside the core point of the video — that it is dishonest to effectively lift other people’s work and present it as your own — it touched on something that I’ve been thinking about quite a bit over the past several months: generation loss.
If you’re unfamiliar with the concept, generation loss3 is the idea that you lose a little bit of quality from the original with each subsequent copy of something. If you photocopy a document, the copy will be of marginally lower quality than the original; if you photocopy the copy, it’ll be even further from the original. With each passing generation, you lose more of the original. You know, copies of copies of copies.
In one of the examples from the video, Brewis points out that there are minor factual mistakes from the original. The rock at the center of “Man in Cave” was listed at the wrong weight in the YouTube video that plagiarized the original Mental Floss article; cave names were mixed up, photos were erroneously captioned, etc. If someone were to make another video that used the Internet Historian video as a source, it would have those same mistakes baked into it, plus whatever mistakes would follow.
A few weeks ago, a YouTube video popped up in my recommendations titled “BEST AI Video Generator || Convert Wikipedia to YouTube Videos With Canva And AI.” Out of curiosity, I watched it. Basically, it tells you to copy/paste a Wikipedia article into an AI like Google’s Bard, ask it to rewrite the article in fewer than 300 words, and then put that “script” into an AI avatar service.
Here’s what their finished product looked like:
It’s just… unsettling? But also, asking a robot to summarize something that is already a summary to put out videos you can profit from while not actually adding anything to the world is just making the internet more cluttered and less useful for everyone.
Just last week, University of Pennsylvania’s Ethan Mollick noticed that the top image result for singer Israel Kamakawiwo'ole (you’ve no doubt heard his cover of “Somewhere Over the Rainbow”) wasn’t actually a picture of Kamakawiwo'ole at all — it was AI-generated.
I’ve been fascinated by the technology behind generative AI, but this, along with threatening the jobs of artists and illustrators everywhere, has been depressing to watch. The internet was heralded as a source for the entirety of human knowledge. Now it’s being cluttered with AI approximations in place of actual photos of people and content-mill copies of others’ original reporting.covered this in June post, whereworried that “the glut of AI text will leave us with a web where the signal is ever harder to find in the noise.” I think we’re already seeing that play out.
Between AI and the type of copy/paste plagiarism plaguing YouTube and making up most of what people seem to think of as “citizen journalism” over on social media outlets where videos and reports get pulled from other platforms and labeled “BREAKING!” or “DEVELOPING!” or “JUST IN!” from people who didn’t actually break the news, the internet information ecosystem sure feels like a dire place at the moment.
What are your thoughts on this? I’d love to hear it. Please feel free to leave a comment below or send me an email.
That’s it for me today. Stay well, all.
This is something YouTube can make happen — if it wants to. Right now, the platform incentivizes the creation of massive amounts of content while punishing creators who, like Brewis, may only put out one or two videos per year. YouTube could tweak its algorithm to change this.
My first experience with this was when I was a kid just learning how to play guitar. I had a cheap microphone and would record myself playing something onto a two-deck cassette player. Then, I’d switch the recording to the first deck and record myself playing over that. Then I’d switch the new recording to the first deck and record over that. With each generation, the quality got weirder with more static. This was probably good because my guitar playing was not great.