Guest article by Simon Very
I have taught science at the post-compulsory level for ten years. Much of the work I assessed was coursework-based and internally evaluated. While these assessments did not rely on standardized tests, they did depend on uniform criteria to judge student submissions; the coursework briefs themselves had to meet prescribed standards. That requirement to comply with fixed criteria created persistent challenges around the originality of learners’ work, and concerns about plagiarism were common.
For many years I have used plagiarism detection software—Turnitin, for example—to discourage and detect copied work. More recently, I discovered that students have started using paraphrasing tools that can evade plagiarism-detection systems. The output from these tools often reads oddly, but the oddity is not always enough to convince markers that the work is inauthentic; similarly, some genuinely student-produced assignments can also read poorly.
The growing availability of paraphrasing software raises the real possibility of an arms race between educators deploying text-analysis tools and learners using increasingly sophisticated obfuscation. That dynamic is worrying because it shifts the focus away from actual learning toward tactical evasion and detection.
I have already experienced a related problem in my classes, albeit on a smaller scale. Students frequently submitted ready-made answers that were easy to find with a search engine. Often they simply copied text that appeared verbatim in assignment briefs and used those exact phrases as search terms to locate model answers online.
At first, I refused to accept ready-made answers as valid work. That position was complicated by the fact that using secondary sources was allowed so long as students cited them correctly. Most students quickly learned to reference material properly, and so they complied with the letter of the rules while still submitting work that lacked original analysis or synthesis.
Students had discovered that the external regulations governing the qualifications they were taking did not specify how much original content was required in coursework. They also learned that exam centers could not impose overly rigid local rules without causing standardization issues. As a result, many assignments became compilations of referenced quotes stitched together with a few transitional sentences. Different students’ submissions often differed only by those small connecting lines.
To address this, I introduced a policy that granted the first student to submit a given secondary source the right to use it, while subsequent identical submissions would be treated as plagiarism of that first student. Initially this reduced direct copying, but it was short-lived. Students found that numerous variant answers existed online; by paraphrasing or selecting slightly different sources, entire groups could submit non-identical but essentially identical responses.
This shift did not indicate that students were analytically recognizing shared structures in generic answers. Instead, they used instrumental methods—trial, error, and iterative feedback—to eliminate options until they arrived at acceptable versions. In other words, many learners employed collaborative, goal-oriented strategies to bypass originality requirements without necessarily developing deeper content understanding.
I began to appreciate that the ability to find general answers online serves well for standard questions—those that align closely and explicitly with curriculum criteria. However, search engines and readily available resources are much less effective when questions are non-standard, context-specific, or phrased idiosyncratically. If a question has the same underlying intent as a typical assessment prompt but appears in an unusual context or uses different terminology, ready-made answers are hard to find.
This limitation is partly due to how search algorithms respond to phrasing and context rather than purely to semantic intent. By deliberately designing questions with unusual terminological combinations or specific situational cues, I found it possible to reduce students’ reliance on canned online answers. Such questions force students to think about meaning rather than simply copy or lightly edit existing material.
That said, some students learned to work around these measures by using advanced search techniques or by rewording the question in ways that made it easier to locate answers. Even so, many learners struggled to accept that search engines can fail to produce useful results for nominally straightforward queries, and that effective use of search requires deliberate, reflective thinking and digital literacy.
Students’ reluctance to accept these realities highlights a broader curricular gap: the need to teach digital literacy as a core competency. Digital literacy should include critical search skills, the ability to evaluate sources, and an understanding of ethical academic practice. Teaching these skills would help learners distinguish between legitimate research and tactical shortcuts, and would encourage genuine engagement with subject matter rather than strategic compliance with assessment rules.
In my classes, I have come to value assessment designs that reward original reasoning, contextual application, and problem solving—approaches that are harder to outsource to paraphrasing tools or to locate verbatim online. While technological tools for detection and obfuscation will continue to evolve, thoughtful question design, clear expectations about originality, and explicit teaching of digital literacy can together shift the balance back toward authentic learning.
Author:
I am Simon Very, and I blog at https://metalearningsite.wordpress.com/