How Plagiarism Changed Over the Past 20 Years

Last week, I wrote about how Plagiarism Today is celebrating its 20th anniversary this month. It’s been a long road, and a lot has changed in that time.

The 2005 internet was, in many ways, a completely different place than it is today. Social networking was in its infancy (Facebook didn’t open to the public until 2006), YouTube had just launched, and blogs and RSS feeds dominated the web.

Today, we have a much more siloed internet. A handful of sites and services make up the lion’s share of the web for most people. For many, YouTube, Facebook, Reddit, etc., are the internet. This concentrates a massive amount of power and content in a handful of companies.

That has had massive changes to how we communicate, including how we teach and learn. Those changes, predictably, have also changed the ways we commit, detect and address plagiarism.

Some of those changes have been very obvious, others are more subtle. But there’s never been a more interesting time for authorship and plagiarism issues.

Party Like It’s 2005

In 2005, we were in the tail end of what I now call the dark ages of plagiarism. It’s a time when the internet was ubiquitous and accessible, but the tools for detecting and stopping plagiarism were not widely available.

The 700 plagiarists that I caught before launching Plagiarism Today were all detected using Google and other search engines. Tools like Turnitin or Copyscape were available, but were far from widespread.

This was a time of copy-and-paste plagiarism. There was little fear of getting caught, and people would often copy and paste without much thought. As I learned firsthand, this plagiarism was frequently for no reason at all. People were posting on forums and in other spaces where there was no obligation or reward (beyond community praise).

The epitome of this issue was the rise in RSS scraping. Since blogs were the dominant kind of site, RSS scraping made it easy for others to copy and paste the entirety of another’s site and then keep their version up to date.

But things were already changing. As detection tools improved and became more available, people worked to find ways around them. One popular method for a time was article spinning, which allowed a user to take one article and create thousands of “spins” of it using synonyms to swap out key words.

However, article spinning (and similar tactics) would come to an abrupt end in 2011. A Google algorithm update largely destroyed the benefits of content farming, severely hurting sites engaged in the practice.

But it wasn’t just Google that was improving its technology. Plagiarism detectors like Turnitin and SafeAssign were both improving and becoming more commonplace. The battle against plagiarism was rapidly turning into a game of cat and mouse.

Cat Meets Mouse

As technology to detect copying became more prevalent, so did the attempts to defeat it. This happened both in and out of the classroom.

Some of the approaches were technical, such as swapping letters out for similar characters or translating works from one language to another, but much of the conversation focused on rewriting or editing copied text. Students and authors alike would ask questions like “How much do I have to change for it not to be plagiarism?” or “How much plagiarism is acceptable?”

This, of course, is not how you paraphrase or write original text. Still, the focus on plagiarism detection tools, which only detect copied text, meant that many were “writing” with the goal of avoiding detection, not creating original work.

However, it was also this time when the internet started to become more consolidated. This was a mixed bag when it came to plagiarism. On one hand, searching for and finding plagiarism became more difficult. Many of these sites are walled gardens that don’t show up on traditional searches.

On the other hand, Facebook, Twitter (now X) and other such sites established norms for sharing and attributing work. This created new habits and social norms for sharing content and, for the first time, sharing and attributing was easier than just stealing.

Attribution and citation were not just topics for the classroom. People were discussing the ethics of sharing vs. copying and finding the original creator was part of the discovery that came with social media.

But, inevitably, things began to shift again. As social media grew, so did the competition for attention. More and more focus was paid to the algorithms the various platforms use, and those algorithms often favored plagiarized or pirated content.

Facebook was (and still is) very prone to this. It strongly favors self-hosted video content. The intent was to reward creators for uploading content to Facebook rather than simply linking to YouTube. However, it also created a perverse incentive for people to pirate content from other sites to get more views. It’s an issue they are just now beginning to address.

Plagiarism was being driven not just by fear or laziness, but by algorithms that rewarded it. For example, on Facebook, the simple act of including a link to a source can cut engagement by half. The algorithms not only reward plagiarism, but they punish those who choose to cite.

Of course, this wouldn’t be the biggest impact that algorithms would have.

The AI Boom

On November 30, 2022, OpenAI announced the public launch of ChatGPT. Though generative AI had been available before that moment, this was when it became proadly available and kicked off the AI race we’re in the middle of now.

AI raises a myriad of questions around authorship, copyright and scholarship. Generative AI has made copy and paste plagiarism all but obsolete. Why steal directly from humans when it’s faster to use a robot? Not only is it easier, but you avoid the legal and some of the ethical risks that come with plagiarizing humans.

This has created its own technology race where AI detectors try to spot AI writing and AI humanizers aim to make such work more difficult to spot.

To make matters more complicated, there’s a gradient to AI usage. Not all uses of AI are plagiarism. Some are no different than what we currently allow human editors and proofreaders to do.

This has made it important to talk not just about whether AI is good or bad, but when and how it could or should be used. When does AI go from a tool used in creation of a work to being the author of that work? That’s not an easy question and there are many answers to it.

To be clear, human plagiarism does still happen. But it’s changing how often it happens, when it happens and why it happens.

All in all, we’re still very new in this post-AI era. We don’t know the impacts that it will have and those may only be visible many years down the road. All that we can say for certain is that it will drastically change both the classroom and the internet at large.

Bottom Line

The story of plagiarism in the last 20 years largely mirrors the story of the internet and technology more broadly.

In the nascent internet, we dealt with copy and paste plagiarism that often have little purpose. As algorithms began to dominate our internet experience, plagiarism became more about gaming those. Now, with AI, we are entering a new phase, one that needles at longstanding questions about authorship and originality.

However, one thing that I have noticed over the years is simple. Almost no one believes plagiarism is a good thing. If you ask that question to 1,000 random people across the planet, almost no one will stand up to call plagiarism a good thing.

What does change is the definition of what is and is not plagiarism. We see this at different age levels, across different cultures and, more recently with shifts in technology. The boundaries of plagiarism and the requirements of citation vary from place to place and person to person.

There is no one standard of plagiarism and likely never will be. While that tension can be frustrating, it also is a big part of why this field has been so interesting over the past two decades.

In the end, we’ve seen major changes is how, why and when people plagiarize. That evolution is not going to stop and, with AI, it may be moving faster than ever.

So, if nothing else, I have my work cut out for me over the next few decades.