Plagarism or inspiration?

Thank you once again, ekuseru. I will read the article on computational linguistics with great interest. (I have a background in one of those two halves, so perhaps I will pick up the other half, in such a context, a bit faster than a neophyte would.) I am curious about the 15-year gap surrounding the origins of TF-IDF. This sounds like a history lesson I will have to pursue. (I did the same thing with theoretical physics—without understanding the actual mathematical models, since I lack such training—when I read a popular book on string theory several years ago. You know you are getting into something when you can recite even the lesser-known names and dates.)

Regarding that eventual debrief, your comment reminds me of one of my most popular posts. Just check out my profile page and look for a post with the word mistakes in it. The points I listed all came from personal experience, as well as from working in the test prep industry and watching students struggle.

If you do not mind my asking, and for a bit of interview practice, I suppose, how does business school figure into your plans? I am guessing you have some sort of IT background. In any case, I have enjoyed our dialogue, and I find myself curious.

- Andrew

Kind of ashamed to say that I actually went through the stereotypical tech-bro phase, dropping out of college like all the other cool kids . Developed some proprietary methods in a niche field of machine learning (natural language generation), found a field to exploit said methods, raised VC money, raised even more money at a 10 mil USD valuation, lost it all overnight because I couldn't convince investors to take GDPR seriously. Crawled back to college to finish my undergrad.

Now I'm in another startup trying to launch in China, but there's a long period where nothing is moving because of the zero-covid policy over there. Looking at some European MBA programs (INSEAD, Oxford) to give me more international exposure outside of Asia.

I'll give the debrief a shot, but the local education system has left me with quite a few unhealthy and borderline degenerate study habits that I wouldn't wish upon anyone else so I'm afraid there's not much utility there.

Of course it's easy to explain why some prep companies do this, just as it's easy to explain why some people lie on a job application or cheat at a poker game. But just because we can explain it doesn't mean we should excuse it. Yes, it's hard to write Verbal questions, but when you say it is "virtually impossible" to prove an original question is "GMAT-like", that can't be true. GMAC does it every day, when they design new questions. And there are thousands of official CR questions we can look at to get an idea of what GMAT-like CR questions are like.

Because you seemed almost impressed with the quality of the replica question, the carpet cleaning one, quoted above, and because I'd never tried to write an imitation official question, I decided to set myself a challenge. I set a timer for five minutes and tried to write as many duplicates of the official airplane engine question in this thread as I could. This is what I did in just over five minutes:

Joe's Marina is planning to reduce its costs by varnishing its docks once a month, rather than every six months. With more frequent application of varnish, Joe will need to less frequently repair his docks, and repairs can close his marina for 3 months. Furthermore, with more frequent varnish, marina customers will use the marina more often and revenue will increase by 1.2 percent.

Tower Skyscrapers is planning to reduce its costs by washing its windows once a month, rather than every six months. With cleaner windows, Tower will less often need to do a thorough cleaning, which shuts down the skyscraper for one day. Furthermore, with more sunlight through the cleaner windows, the office workers working in Tower Skyscrapers will increase their productivity by 1.2 percent.

The Qate Gallery is planning to reduce its costs by repainting its walls once a month, rather than every six months. With fresher paint, Qate will less often need to replace its walls, which can shut down the Gallery for up to six months. Furthermore, with cleaner looking walls, Qate can anticipate selling 1.2 percent more paintings.

Stake Farms is planning to reduce its costs by tilling its fields once a week, rather than once a month, as most farms do. With more frequently tilled fields, Stake will no longer need to let a field go fallow, which takes a field out of operation one year out of every four. Furthermore, frequently tilled fields will increase the harvest by 1.2 percent.

These questions are terrible, but I was writing them faster than one is meant to solve them, about 75 seconds per question. It's trivially easy to do. It's really just something of a semiotic exercise; the form of the question is this:

Company X will reduce its costs by doing activity Y more often (even though Y costs money). By doing Y more often, something expensive will be kept in better condition and won't need to be replaced. Activity Y also offers one other minor benefit.

Here X and Y can be almost anything -- the activity Y could be weeding a garden, retraining workers, conducting fire code inspections, replacing solar panels on a space station, anything you might do at regular intervals that you might choose to do more often than before. It's the logical template that is the essence of the question. That's the hard part to come up with. It's not hard to realize you can replace "engine cleaning" with "carpet cleaning". So if anyone copies the essential logical form of an official question, that person is fundamentally copying the question, no matter what they replace "X" and "Y" with. That's why I said the carpet cleaning question is "clearly a copy" of the official airplane engine question.

It's like if a songwriter released a new album claiming it contained "original material", but the songs were all Tracy Chapman songs with a few wording changes, so "Fast Car" became "Fast Truck", but used the same vocal melody and major 7 guitar line. Changing a few words doesn't change the essence of the song (in my example the wording change presumably makes the song worse, possibly ruining the original for anyone hearing it, just as an imitation prep company question can ruin the official version). We wouldn't excuse this "songwriter" for plagiarizing Tracy Chapman songs by saying "writing good songs is hard", or "how else could this songwriter quickly get an album out" or "it's impossible to know what a good song is like so you have to copy an existing one", even though there are thousands of examples of successful songs. If a prep company finds it too hard to write good, useful, original questions, the company simply shouldn't be selling questions, just as someone who finds it too hard to finish medical school shouldn't be practicing medicine, or someone who finds it too hard to pass the bar shouldn't be practicing law, or someone who finds it too hard to write a song shouldn't be copying someone else's songs and claiming to have written them.

The similarities become evident after decomposing both the official and derivative question down to formal logic.

But say we change the template to:
Activity X causes scenario A and activity Y causes scenario B.
Scenario B is has an obvious benefit over A and another auxiliary benefit.

What would be the most useful to evaluate the plan to replace X with Y?

This template is likely to pass a sniff test since all named entities can be replaced but would it be a "good, useful, original" question?

My hunch proved correct, then. The details you provided above about your background sound perfect for generating an essay. My guess is that not too many people have, at such a young age, launched a business that hit such a valuation. Other people I have worked with in the past have applied to European schools with scores deemed a little too low for M7s, but competitive for Europe, ranging from 680 to the low 700s. I imagine a 760 will do nothing to hurt your admissions chances, and I have say, your written English is well articulated (the missing "i" in "plagiarism" notwithstanding, which I take to have been a typo). Provided you interview well, I see no reason why you would not be attending your school of choice by the term in which you would choose to matriculate.

The debrief would be a bonus to the community, but you do not owe anyone anything. (I recall having read one debrief from a high scorer who said he put in a marathon study session—I believe twelve hours or so—to finish the entire GMAT Advanced guide the day before he took the exam. Everyone approaches the task a bit differently.)

Good luck with your endeavors.

- Andrew

Sorry for missing from the action in this discussion. Not trying to say that this is a good idea, but one piece of value would be teaching students to spot patterns. Having just a single question would be insufficient. I think the definition of the word pattern probably involves having multiple of the same thing. That’s the value I could see. I know that the number of people in the past have accused one old school prep company that is not on GMAT Club for inventing the wrong question patterns that were kind of fitting with their teaching strategy and showing students how they could be successful. In reality they should’ve done it the opposite where they should’ve created a teaching strategy that would fit official question patterns. In any case, the result was disastrous with many people feeling confident before the test only to be destroyed in the verbal section on the Test day.

Posted from my mobile device

This is really the point though, and if any test takers are still reading this thread I think it's the most important point: there aren't "question patterns" in GMAT CR. ekuseru can point to a unique official question that this prep company copied, not to ten different official questions that all follow the same "pattern". So about this:



It's definitely true that some companies do this -- design questions to "prove" their strategies work. Prep companies that teach backsolving in Quant, for example, often design questions that can only easily be solved using backsolving, to prove how useful backsolving is, even though there aren't questions like that on the real GMAT. So of course if a company is designing questions that test things in a way the GMAT does not, that company is providing test takers with misleading questions, and a test taker studying those questions might adopt the wrong approach to the test. But the same thing is true of these companies that produce copies of official questions. When companies do that, and test takers see the same question patterns in the prep company material and then in the OG, test takers can think the GMAT does recycle question patterns. Then test day comes, and none of the CR questions on the test follow the "patterns" the test taker has studied, and the test taker finds they're in the exact situation you describe in what I quote above. Even in Quant, it's mostly a myth that GMAT questions follow "patterns", unless you define that word very loosely.

It might be helpful to be precise with what we mean by "question template".

If we're strictly referring to the logical template of a question, I would argue that having a derivative question is of little utility. Once someone is comfortable with a particular way of playing with ideas (e.g. if X⇒A and Y⇒A: is ∀X ∀Y (X=Y) ?), it becomes fairly trivial to crunch through questions with the same logical template.

However, if we're talking about "question templates" as in the way certain types of questions (e.g. weaken, assumption, boldface) are phrased, I can understand where bb is coming from. The PowerScore CR bible spends a lot of time on the identification of question types, something that a lot of students apparently appreciate. In that sense, getting exposure to the many ways CR questions can be phrased might be helpful?



I can attest to that. Contrary to what GMATWhi‎z preaches, trying to "pre-think" by applying known logical templates often results in me falling in love with wrong options and second-guessing myself. During my actual attempt, in which I only got medium-hard to hard questions, I found that CR questions were all unique. Purely off anecdotal evidence (source: trust me bro), I won't recommend wasting time on dupes.

I agree it's important to distinguish between what we might call 'question templates' or 'question patterns' on the one hand, and what we might call 'question types' and 'question topics' on the other. There are definitely types of questions in GMAT CR -- strengthen questions, inference questions, assumption questions, etc. But two random official assumption questions will barely resemble each other. There are also certain 'topics' tested on GMAT CR (in some questions, not in all of them), just as there are certain topics tested in Quant -- some CR questions are fundamentally about correlation and causation, say, and some are fundamentally about sample bias, etc. But again, if you take two official correlation+causation questions, they'll usually barely resemble each other.

Good unofficial Verbal questions should often be of the same general type as an official question (assumption, inference, etc) and should sometimes test the same kinds of topics (correlation vs causation, say), but once a prep company question starts mimicking the template or pattern of an official question, then it becomes a plagiarized question that isn't useful to test takers who will study the better original version. I'll grant there's a grey area, where someone might take an official question's pattern as a source of inspiration, change some of the logic of the problem, and come up with something new. Depending on how distant the new question is from the source, the new question might become original in that case. But that carpet cleaning question is nowhere near that grey area.

At the minimum, test prep companies should never plagiarise GMATPREP questions. To make a disguised version of a GMATPREP question is just unforgivable.

Disguised versions of OG questions are not that harmful (though they are still dishonest).

And of course, nobody in test prep should use GMATPREP questions as teaching material.


Posted from my mobile device

CR questions aside, I think we can all agree that plagiarizing an article for RC questions without giving credit is a pretty open and shut case?

Read this article, then read the following passage from GMAT Whiz:

In 2008, scientists working in Denisova Cave in Siberia’s Altai Mountains unearthed a strange pink bone, broader than a typical human’s. The bone DNA revealed that its owner belonged to an entirely new group of ancient hominins, distinct from Homo sapiens or Neanderthals. That group became known as the Denisovans. Researchers have since decoded the Denisovan genome. However, still no one can say what they looked like. Every known Denisovan fossil would fit in your palm and all of these remains came from the same cave.

However, now an international team of scientists has announced the identification of another Denisovan fossil, from a site 1,500 miles away. It’s the right half of a jawbone, found in 1980, some 10,700 feet above sea level in a cave in China’s Xiahe County, on the eastern edge of the Tibetan plateau. The Xiahe mandible, as it is now known, is not only the first Denisovan fossil to be found outside Denisova Cave, but also the very first Denisovan fossil to be found at all. It just took four decades for anyone to realize that. The mandible lay unstudied until 2010, when a climatologist and an archaeologist, began examining it in earnest. The world learned about the existence of the Denisovans at around that time, and though fossils had only been recovered from Siberia, it was clear that these hominins likely existed throughout much of East Asia.

The mandible itself is very thick and sturdy. It has no chin, which rules out modern humans. The teeth within it are exceptionally large, and different in shape and size from those of Neanderthals, Homo erectus, and other known hominins. The molecules in the specimen were especially telling. The team couldn’t detect any traces of ancient DNA, but it did find the next best thing—fragments of ancient collagen proteins, still lurking in one of the teeth. These fragments closely resemble the proteins of Denisovans, more so than those of Neanderthals, modern humans, or other great apes. It confirms that the Denisovans were perhaps widely distributed through East Asia. After all, people across East Asia and Melanesia have Denisovan DNA in their genes. This pattern—the product of ancient sexual encounters between Denisovans and humans—wouldn’t be possible if the Denisovans were just confined to a small Siberian cave. Instead, it seemed that they were already living in much of East Asia by the time ancient humans also spread through the region.

This is not an isolated incident, I've documented many such cases. Even a simple google search will bring up the plagiarized article.
bb if you require more evidence, I'll be more than happy to provide them.
I'll reach out to the Atlantic and other victims as well.

Will the Atlantic mind that somebody is using am extract from an article to teach comprehension?
I really don't know, but it's possible that the magazine has no objection.

Posted from my mobile device

Well, of the 6 RCs I've checked thus far, they've managed to plagiarized from not only the Atlantic, but also paid articles from eNotes and even press releases from NobelPrize.org.


This is not a victimless crime, let's find out.

Article from ScienceDaily that GMAT Whiz plagiarized and an excerpt from GMAT Whiz:
For centuries, transitive inference, a form of logical reasoning used to make inferences, was considered the domain of human deductive
powers. However, in recent years, vertebrate animals including monkeys, birds and fish have displayed the ability to use transitive inference. The only published study that assessed Transitive Inference in invertebrates found that honeybees weren't up to the task. It is believed that the small nervous system of honeybees imposes cognitive constraints that prevent them from conducting transitive inference. Paper wasps have a nervous system roughly the same size as that of honeybees, but they exhibit a type of complex social behavior not seen in honeybee colonies. An evolutionary biologist, Liza, who has studied the behavior of paper wasps for 20 years, wondered if paper wasps' social skills could enable them to succeed where honeybees had failed.

And this is from ScienceDaily's terms and conditions of use:
Reproduction/Use of Material

ScienceDaily.com has been uniquely designed for presentation of content in a format specific to its users and with an integrity that we wish to maintain. Any unauthorized copy, reproduction, distribution, publication, display, modification, or transmission of any part of this Service is strictly prohibited. Users may download material from this Service for personal use. Without limiting the generality of the foregoing, you may not distribute any part of this Service over any network, nor sell or offer it for sale.

See our Permissions section for more information on distribution. In addition, these files may not be used to construct any kind of database.

Who is the victim?

Another point, even the OG does not credit sources. All the RC passages are excerpted and adapted from somewhere. If you google SC sentences, you will often find the original.

Perhaps the GMAC takes permission from the original sources. For SC and CR, such permission is probably unnecessary. I don't know about RC.

Posted from my mobile device

I've found an blatant excerpt from a book, GMAT Whiz copy here:

Half a century ago most African Americans were trapped in poverty. In 1944, the majority lived in the South, on the land as laborers and sharecroppers. A mere 5 percent of the African Americans men nationally were engaged in nonmanual, white-collar work of any kind; the vast majority held ill-paid, insecure, manual jobs—jobs that few whites would take. Over 50 percent of African-American women were household servants who, driven by economic desperation, often worked 12-hour days for abysmally low wages. Segregation in the South and discrimination in the North allowed only a handful of African Americans to own businesses (funeral homes, beauty parlors, and the like) that served only the African American community.

However, beginning in the 1940s, demographic and economic changes, followed by a marked shift in white racial attitudes, started the African Americans on the road to greater equality. New Deal legislation, which set minimum wages and hours, and eliminated the incentive of employers in the South to hire low-wage African American workers, put a dampener on further industrial development in the region. In addition, the trend toward mechanized agriculture and a diminished demand for American cotton in the face of international competition combined to displace African Americans from the land. As a consequence, with the shortage of workers in manufacturing plants in the north following the outbreak of World War II, southern African Americans boarded trains and buses in search of jobs, a Great Migration that lasted through the mid-1960s. They found what they were looking for: wages so strikingly high that in 1953 the average income for an African American family in the North was almost twice that of those who remained in the South. Through much of the 1950s, overall, wages rose steadily and unemployment was low.

Thus, by 1960 only one out of seven African American men still labored on the land, and almost a quarter were in white-collar or skilled manual occupations. Another 24 percent had semiskilled factory jobs that meant membership in the stable working class, while the proportion of African American women working as servants had been cut in half. A decade later, the gains were even more striking. Apart from increase in average income levels, African American life expectancy went up dramatically, as did their home ownership and college enrollment rates.

I'm quite sure that eNotes, which paygates their articles, and the authors of the book I've mentioned won't be too happy about their work being appropriated.
Just to be safe, I've messaged GMAC to ask about their legalese.

Of course GMAC is licensing any material they take from another source. The whole test could be shut down if they didn't. And permission would absolutely be necessary for any CR or SC questions adapted from an existing source.

If the prep company concerned here is licensing the material they're using, there's nothing wrong with them using it. If they're not licensing the material, then since the copyrights were violated for financial gain, this may be a criminal offense, and it's certainly subject to civil penalty.



I don't mean to sound snarky, but I'm not sure if this is a question you thought about before posting it?

Genuine question! I asked out of curiosity.
Posted from my mobile device