74 comments

[ 2.6 ms ] story [ 150 ms ] thread
Why would the owners of an online dictionary feel compelled to thwart cheating on tests? It's their job to provide accurate information, not justice to cheats.
It's a volunteer project. What they decide to do with their free time is probably their own business. If they want to temporarily alter some definitions to validate a hypothesis then more power to them
What they do is their own business, of course, but how would you feel if the folks running npm temporarily altered some of the javascript packages they served?
If I was using that dictionary I would stop using it. Way to loose any credibility irregardless of what you are trying to do.

Also, story seems made up.

Indeed, something smells fishy.

First, the words "Everywhere, [to] spot" seem too simple to need to be researched - are these words so uncommon in Romanian that students need to look them up?

Second, they impacted ALL users to their site, merely because they saw an increase in use that coincided with the test? Did they even check their website logs to see where the source IPs were geo-located? They laugh off the impact caused to a blogger's readers with their replacement of "treachery"'s meaning.

> There were only nine searches for those two words in the hour before the exam started, but the number soon began to soar. In the final hour of the test there were 989 requests for their meaning

The article suggests you could likely count the number of innocent users affected on fingers & toes?

The article suggests many readers of a popular author's blog were impacted.

> The DEX editors assumed it was and changed its definition, but later discovered that it had been used by a well-known Romanian author in a blog post that morning, something they call an "unfortunate coincidence".

No, these aren't common words at all. The definitions the site is giving here are their more commonly-used analogues. In the same way that "domicile" in English means "home", but it doesn't mean you don't know what a home is because you have to go to the dictionary for "domicile".
It almost seems as if you've stopped using all dictionaries.
Many of us here have English as a second or third language.
Story is 100% true. You can check with Google Translate any recent article in the romanian online press regarding dexonline and exams.
Maybe next they'll change the definition of "irregardless".
(comment deleted)
Does anyone else find this somewhat wrong?

They are literally changing the definition of words (albeit momentarily) to try to identify cheating. This is likely a private institution administering the test and a government dictionary. Even if both were governmental (dictionary and test), it still seems silly to do this, as opposed to trying to catch the cheaters.

Isn't trying to identify the cheaters a good way to catch them?
Assuming I understand this correctly, the only way to catch them is that they misuse one of the three most commonly used words....

That doesn't seem like it will provide enough evidence to actually "catch" a cheater.

These aren't commonly-used words, these are very obscure words the students are being asked to define. If a bunch of students all settle on the same completely-wrong definition, it's a fair guess they were cheating.
With an astoundingly high false positive rate, assuming it's a multiple choice test.
The guys running the dictionary didn't even know what was on the test, judging by the fact that they had to guess them by looking at traffic logs. So it's highly unlikely they settled on a fake definition that just happened to match an alternative choice on a multiple choice test.
Why would the editors of an online dictionary want to catch cheaters to begin with?
I don't think they wanted to "catch" the cheaters. This was not a real examination to begin with, it was just a simulation in order for the pupils to get a feel for how things will happen. The real testing will happen on the 31st of March.

It was more of an experiment/eye opener from the team of volunteers behind the online dictionary project.

Identifying them by misinforming them seems counterproductive.

Imagine if a math site temporarily said "2 + 2 = 5" in order to catch cheaters on a math test. You've caught some cheaters, but then some of them will continue thinking that 2 + 2 really is 5.

There must be better ways to do this.

This is a fairly well-known technique for spotting copycats (and I'm sure I've seen examples on HN for both cartographers and Trivial Pursuit).

But yeah, it is somewhat morally hazy in this instance, I agree.

Actually, the test is administered by the public education system and the dictionary is a volunteer job. I see no ethical issues with what they did, the ethics issues lie with the ones that tried to cheat.
> the dictionary is a volunteer job

So it would only be unethical if they were being paid? I'm not sure what the distinction is.

I merely pointed out that they are volunteers in order to underline that there is no connection between the state educational system and the private team of volunteers which provide the dictionary service.
This was a mock exam held by the government.

The dictionary website is private, but very well known.

As it's a mock exam, nobody will get into trouble for it. But, if nothing else, it should make the students think twice before trying to cheat on the next exam.

This was a mock exam held by the government.

The dictionary website is private, but very well known.

As it's a mock exam, nobody will get into trouble for it. But, if nothing else, it should make the students think twice before trying to cheat on the next exam.

A bit of context.

1. As others pointed out, this is not a full-blown exam but a simulation, in preparation for the exam that shall take place in June.

2. Cheating is not unheard of in Romania, in some places is common practice, sometimes with the aid of teachers. As the Ministry of Education is not striving too much to fix it, this is just a private initiative to spot it and hopefully put raise awareness / public pressure towards this phenomenon. Noteworthy, the uncommon volume of search on the terms that were present in the exam started even before the official start of the exam.

I was at first also thinking that they could've just hid the words and put up a notice for the regular users, but after thinking more about it, I think they're approach was optimal.

Sidenote, the guy[1] that started the project is an ex-googler[2], I trust he (& team) gave this some thought before choosing this approach.

[1] https://catalin.francu.com/

[2] https://catalin.francu.com/resume.txt

Sigh, I understand this was done with good intentions, but a dictionary is one of the last things whose integrity should be compromised. They call the searches for "treachery" an "unfortunate coincidence", but it's really a betrayal of people searching for authoritative definitions. When you're deliberately falsifying the definition of words, you've failed as a dictionary. Instead, the schools giving the test should have sought other solutions: make proctors be more vigilant, collect student cellphones, or even reach out to dictionary websites beforehand so they could omit, rather than modify, the definitions of any words on the test day.
Or proxy the dictionary entry in the ISP DNS at the test site. No need to corrupt the dictionary for everybody.
The test is taken in schools national-wide. You should basically ban the entire country of Romania from accessing the dictionary.
That's actually censorship.
If kids are looking up answers on cell phones, I don't know how well that will work.
Just to clear some points: there is no administrative connection between the volunteers which provide the dictionary service and the testing. For this kind of national test you are not supposed to be able to bring in dictionaries, communication devices or any other helper material, it's just you and your brain. Classrooms where the exam is taken are outfitted with cameras, so in the event that there are suspicions of cheating, the footage may help. Of course, all of this does not mean that nobody will be able to cheat.

Finally, I think it's useful to point out that this was just the team of volunteers behind the online dictionary who modified the definition of 3 very uncommon words for a few hours.

> The third word that appeared abnormally popular, "treachery", wasn't actually on the test. The DEX editors assumed it was and changed its definition, but later discovered that it had been used by a well-known Romanian author in a blog post that morning, something they call an "unfortunate coincidence".

I'm all for punishing cheaters, but this measure has just too much collateral damage. Legitimate dictionary users are being mislead. It's a bad idea, and it would shatter my trust in any dictionary.

"Legitimate dictionary users are being mislead"

They replaced the words with similar sounding words that have no link with the original words. Anybody could easily spot the change.

This seems somewhat Ludditic. Technology makes the format of the test obsolete, why counteract that?
(comment deleted)
Because there has to be a point where you start using your own brain and memory. Just like a muscle you need to train and use them in order to keep them efficient. In a normal conversation you don't have time to access a word in a dictionary. How many words are you willing to memorize? How much offline knowledge is required for you to "survive"? If we offload all cognitive efforts to machines, what is left except meat and bones?

Where exactly is that point? I don't know.

> In a normal conversation you don't have time to access a word in a dictionary.

If you haven't the heard the word up until that points, odds are pretty good that you might not have needed it.

Coupled with the question "how many words are you willing to memorize" I was trying to offer a sense of scale to the absurd.

If looking up words in dictionaries is sufficient, how many words should you memorize offline? If everyone agrees learning words is outdated, what will they use to speak? Orwell's newspeak?

Most communication, as opposed to art for instance, works both ways. You listen and read, but also speak and write. Dictionaries help you with the first part, but you'll need a little more effort with the second.

Forklifts make my muscles obsolete, why bother having gym class?
Your argument assumes that the tests are useful as opposed to somewhat arbitrary.
I would say that my argument assumes that having a larger vocabulary is useful, and that these tests are one way of encouraging that.
What kind of exam asks for the meaning of the word "everywhere"? Doesn't everyone know that when you have an age where you can participate in an exam?

It says it's for Romanian students and is in the Romanian language, so it doesn't look like it's a foreign language to them but their native language.

In English, a lot of standardized exams look for knowledge of lesser-known synonyms or variants of words; it's quite possible we're getting an article here which translates the words to a more common English variant for easier understanding.

For example, a similar article translated to Romanian might claim English-language students were tested on knowing a word for "rural" (simplified translation) but the exam actually used "bucolic" (a word which I know has been part of the GRE vocabulary in the past).

The actual words are archaic, lesser known synonyms.
> What kind of exam asks for the meaning of the word "everywhere"?

It's probably closer to something like "ubiquitous", which would be a reasonable entry in a vocabulary exam.

Shows the superiority of offline dictionaries.

Edit: And not-needlessly-online software in general.

I agree, offline dictionaries have a certain esquivalience about them.
I feel like the Hacker News hive mind is short-circuiting here, the "anti-censorship" impulse is overriding rationality. This isn't some Orwellian conspiracy to compromise language, it was a three-hour redefinition of a small handful of obscure words.

Obviously schools already try to curtail cheating, it's not useful to just say "be more vigilant instead". This is a clever way to identify cheats from an unexpected angle; and obviously if they're cheating on one section of the test, they're likely to be cheating on others.

Cheating is such a fluid concept anyway, and it's inconsistently applied. Some schools will suspend students for using calculators on a math test; some force students to bring calculators to their tests. Some give out the questions before a test, while others expel students for distributing test questions. Some require students to collaborate or use online sources; some forbid it. It's arbitrary.

Punishment for cheating also presumes the validity of 20th century-style academic testing, which is debatable.

What is the purpose? To ensure I'm qualified for a job? A job where I am free to look up words in a dictionary whenever I want?

I'd also add this: The internet is developing quickly into a literal extension of the human mind. I don't think it will be all that long before we're connected much more intimately to the internet than we are right now with just our eyeballs and fingertips. And that means we need to reevaluate what it means to learn information versus to find information.

> Cheating is such a fluid concept anyway, and it's inconsistently applied. Some schools will suspend students for using calculators on a math test; some force students to bring calculators to their tests. Some give out the questions before a test, while others expel students for distributing test questions. Some require students to collaborate or use online sources; some forbid it. It's arbitrary.

These are students sneaking cell phones into an exam, and going out of their way to surreptitiously look up answers with them. There's no moral ambiguity here.

> Punishment for cheating also presumes the validity of 20th century-style academic testing.

No it doesn't.

> What is the purpose? To ensure I'm qualified for a job? A job where I am free to look up words in a dictionary whenever I want?

These are primary school-children. There's nothing wrong with asking them to learn basic skills like literacy or numeracy (or, say, basic integrity) even if they're disconnected from "jobs".

> I don't think it will be all that long before we're connected much more intimately to the internet than we are right now with just our eyeballs and fingertips. And that means we need to reevaluate what it means to learn information versus to find information.

Sure, and there will be tests for that set of skills where accessing the internet during the exam won't be considered cheating. But in this situation it is.

No, in 8th grade the teachers (and parents) are not that concerned that you you're not qualified for a job. They just need to make sure that you are able and willing to study by assessing your last (and first) 8 years of school.

At minimum they want you to be able to articulate a phrase and mostly to understand one and, who knows, maybe to have you try to make sure that you'll be able get critical thinking and hopefully not be a little manipulated, cheating prick that will vote an incompetent as a president just to prove a point to an unknown "enemy".

I hated these exams as a kid, but thinking about it now, they do have a point.

The purpose of the exam is to show that you have a vocabulary, and you can understand and answer some questions about a text you're seeing for the first time.

This is important for a lot of jobs, and it isn't something you can google. Sure, you can google individual words, that is less useful if you're the one writing.

> Cheating is such a fluid concept anyway, and it's inconsistently applied. Some schools will suspend students for using calculators on a math test; some force students to bring calculators to their tests. Some give out the questions before a test, while others expel students for distributing test questions. Some require students to collaborate or use online sources; some forbid it. It's arbitrary.

Just because the rules are different for different exams doesn't make cheating fluid.

Cheating is such a fluid concept anyway, and it's inconsistently applied. The Tour de France will kick you out for riding a motorbike; MotoGP forces you to ride a motorbike.

The rules are there to provide some semblance of a level playing field and to ensure that the relevant skills are being assessed. An elementary school arithmetic test is a pointless exercise if calculators are allowed; an undergraduate engineering test would be a farce if calculators weren't allowed.

> This isn't some Orwellian conspiracy to compromise language, it was a three-hour redefinition of a small handful of obscure words.

That backfired immediately, when one of the words being searched wasn't actually part of the alleged cheating.

They should just hire a psychometrician. They have PhDs specifically for this sort of analysis.
The dictionnary in question: www.dexonline.ro
It would be far less harmful if the exam asked the students to define a fictitious word which appeared only in the dictionary. Only cheaters would get it "right".

https://en.wikipedia.org/wiki/Fictitious_entry

The exam and the dictionary are completely unrelated. The exam is devised and organized by the state, while the dictionary is an online service run by a group of volunteers.
> The exam and the dictionary are completely unrelated.

No, this article is specifically about them working together.

Nope, it's about the dictionary admins noticing cheaters during last year's exam and deciding to do something about it -- without any coordination with the exam organizers.
(comment deleted)
Some context:

Cheating at exams and plagiarism is rampant in Romania. Multiple gov't officials, MPs and even a former prime minister nicknamed copy/paste were accused of plagiarism in the last few years. They tried to subordinate comitees that analyze PhD thesis just to get away with it and even succeeded in doing so.

National exams use cameras to twart cheaters.

Dexonline.ro is a volunteer effort and assumes no responsability for the correctness of its content.

Administrators want to catch people cheating at exams. So what do they do?

They cheat the public by spreading misinformation and lying.

Quote at end of article:

"The third word that appeared abnormally popular, "treachery", wasn't actually on the test. The DEX editors assumed it was and changed its definition, but later discovered that it had been used by a well-known Romanian author in a blog post that morning, something they call an "unfortunate coincidence"."

They fucked over a bunch of unrelated innocent people with their fake news/fake definition and simply call it "unfortunate".

Talk about the irony and double standards here. What a pathetic bunch of sorry human beings.

I'm fascinated by this HN discussion! Personally I consider it an excellent hack, but it's good to learn why people think otherwise. (And it's great that the article even describes one of unfortunate side effects.)

I can imagine a MITM product for schools to help here, perhaps not though if everyone is using LTE instead of the school's network.