I never saw Raval as much more than an internet personality.
As such, the fundamental problem here is the world today values and rewards glossy content. Gobs and gobs of glossy content. Repackaged content. Aggregated content. Outright stolen content.
If you can make it at scale, views, subscribes and subsequent ad impressions will make it worth it.
And the good news: we're not discerning viewers. Someone stole something? We'll wag our finger, furrow our brow and click play anyway. There exists only incentive for producing.
That isn't new for better or worse. Even with new defined broadly as "times that a person who is alive today lived through".
I am reminded amusingly about how moderation at scale pleasing everyone is impossible by complaining about how many pages lower mirrors were than original websites on Google.
Putting aside the dispute about the novelty and intensity how is a decrease in gatekeepers both natural (the expense of distribution) and artificial relevant to the cited problem?
They haven't made any positive difference and cannot. The incentives for "gloss" to sell are the same and the alternatives have their own evils.
Because the nature of increased competition is accelerationism. The fastest way to change a product is to introduce more viable competitors; metaphorically, it's a form of evolution.
They've sold an entertainment/advertisement product, one that required almost no change in formula for hundreds of years.
In journalism school you learn early on the history of newspapers in New York. The consequence was more competition lead to drastic price wars and a race to the bottom. By many accounts it culminated in yellow journalism.
That all found a natural equilibrium and things remained mostly unchanged until television. With that, competition for short, snippet journalism increased and newspapers moved to longer, more in depth journalism and larger shelved the very long form because the audience was gone.
Cable news had a similarly tectonic force on network news, which changed its drab, almost newspaper-y style to compete.
Each of these forces fundamentally changed the nature of these long lived formulas. Twitter shoved another dagger in newspaper journalism by forcing them to compete on social media, eliminating journalist standards for interaction with the public.
The gatekeepers react based on what audiences do. Audiences are not discerning. Each one of these events puts a dent in the integrity of its former self. It's all a race to the bottom for advertisement, something a lack of competition previously kept slow enough to seem at bay.
Really information theory and limited sensible combinations make that an inevitability with good enough record keeping. There are only so many possible arrangements of words to make say a sensible arrangement of George Washington's biography that are coherently organized and remotely factual.
Are you sure about this? There are only a few ways to state the year and place of birth, perhaps, but once you get beyond that, there are what might as well be infinitely many possible biographies of Washington. That being said, if all your new bio has to offer is a re-organization of the same well-established historical facts, it probably doesn't need to be written.
Very interesting... I never bought the idea of GPT-2 as a threat to humanity, since it's not able to produce meaningful original new content (it just produces thematically and grammatically correct filler), but if your primary goal is to plagiarize, then maybe the threat is real.
Problem statement: given context C and sentence X, find a sentence X' that communicates the same idea as sentence X but in different words. This is not exactly GPT-2, but it seems doable.
This kind of plagiarism would be difficult to detect. The plagiarizing party could say something like "I didn't copy or reproduce, I just used as input to a NN."
I'm not actually as worried about this sort of direct copy-paste plagiarism as I am about a related issue.
What I'm worried about is the appearance of authority when there is none, something that I guess Raval is also accused of. But what I'm looking at is a bit more subtle than pretending to have written a paper.
In several fields, you get someone popularizing certain topics and observations. Some guy will write a book about economics or psychology with some compelling storytelling, and many articles will be written reviewing the books. But actually nothing new has been added. It's simply a rehash of a few well worn ideas from an area. Importantly what can also happen is something that was considered marginal is brought up (Marxist/Austrian economics) as accepted mainstream. Or something with weak reproducibility (Marshmallow) is brought up as the accepted truth. Nothing that's a complete lie, but you get a picture of a field that is quite different to what the experts see it as.
The fact that they aren't making up complete lies is what makes them hard to critique. If someone comes up with a list of studies that seem to support some tenuous case, it becomes quite hard to rebut them without seeming quite pedantic. For instance you might need to dive into the methodology of some survey to understand the critique, but the soundbite culture of our times does not permit that.
Often the person writing these books is portrayed as someone very authoritative, and certainly that makes sense for the publisher. They get in the news, they give commentary on relevant news items, and the circus continues. They never really have to defend their positions, because they're always a sort of conduit for some other historical authority figure. But the total mix of what they're saying doesn't add up.
You make a fair point, but ins't that what any educator does. I've even heard multiple teachers say "never let the truth get in the way of a good lesson". It becomes impossible to begin to explain anything if you don't simplify thing.
The problem with Raval is he is a narcissist. Check out his youtube apology video where he reads out, for 4 minutes, the names of the developers he plagiarised. It's a grand gesture suitable on veterans day or after a terrorist attack. He gets pleasure from the attention alone, others get pleasure from actually doing something. Maybe we should be asking the ESA who decided to give him a podium.
You've just described pretty much all 'public intellectuals'. This is not new, and they are professionally hated for a reason.
The most egregious example of this is Malcolm Gladwell. Not one of his books is true in any sense of the word even at the time he wrote them. Yet that didn't stop him from becoming an expert in ... expertise.
This. You did a very good job of describing the current pattern of behavior of these pariahs who make a career from plagiarism. A lot of times, this is the basis of their entire "work". They plagiarize nonstop and unapologetically.
A lot of people who engage in these behaviors on the internet are narcissists, at minimum. At worst, they could be malignant narcissists (narcissists with sociopathic tendencies) or narcissists who are straight up sociopaths, too. What I mean is that it is not just straight up plagiarism: there is an unbelievably strange level of self-promotion from them on the internet, which is at minimum, a huge red flag. Their presence and "influence" on the internet is unavoidable, really. It is really tragic when they start going in to "research", which is a profoundly dumb move by them. At some point, they will get caught, and it will not be brushed off at all by other researchers and academics.
One should be alert to identifying these people on the internet. They are "smart" in the sense that they know how to play and manipulate people, all the time.
If Raval were smart he’d take a lesson from the Johnsons and Lehrers of the world and go quiet for a bit. Wait until the heat and the scandal die down and he can make a return. While a real apology would probably help, it may not even be necessary.
Raval will, almost certainly, be fine. If he isn’t it will be owed not to the plagiarism alone, but to the combination of all of the scandals brewing around him. One plagiarized paper may be survivable, but a history of weak citation, a heavily-criticized course and a feeling he misled his community may be enough to make him a pariah.
Still, we’ve seen others survive much worse and it’s clear we’re entering an era where the title “plagiarist” doesn’t disqualify you from any position at all. It’s frustrating as many times it appropriate to sever all ties with plagiarists, but it’s the reality we live in.
In the end, it would be good if Raval could at least thoroughly reexamine his approach to citation, in everything he does. Citation isn’t just about formats and standards, it’s a thought process and a way of creating. Clearly, Raval doesn’t have that right now.
If nothing else good comes of this, maybe he can take this as a learning opportunity and grow from it.
Let's be real: Programmers plagiarize all the time. Especially other people's code, but it can often be from peer-reviewed journal articles, other peer-reviewed literature, literature in general, or just something off the internet.
It is in everyone's best interest to comment their code and provide where their information came from. More likely than not, you are no super-genius, and you needed this information to finish your project. It is highly unlikely that anybody can do this on their own.
It is no big deal to cite where you are getting your information. It is important in case somebody needs to review your project.
Yup, even in private code I always add the occasional `/* taken from K&R 2nd ed. */`. Helps when reviewing anyway so you can understand in what context the code was written and how it applies to the current project.
This "future" of plagiarism is the past of plagiarism in China, as regards technology. Without a strong regulatory regime to stop it, it's just more efficient and profitable to plagiarize (but please, call it "research").
This view might be conflating IP theft and plagiarism -- plagiarism has always had a strong regulatory regime in the form of peer review. Regulation of IP theft is a different story.
Minor update: after the fallout months ago (around when this article was published), Siraj recently made a video apologizing for the plagiarism with a promise to offer more credit explicitly: https://www.youtube.com/watch?v=1zZZjaYl4AA
Not many people found it convincing, especially as he did not apologize for the paid course plagiarism which brought his plagiarism mainstream.
I think Raval’s case is an interesting one for many reasons. First, Raval bills himself, at least in part, as an educator. Second, the way that this story fits into the larger picture of his actions. Finally, much of the work he does is in areas such as blockchain and machine learning, areas that often don’t have a great deal of public trust.
Plagiarism is very, very easy to understand: It boils down to money. Students plagiarize their homework to pass exams / get good grades, and to get better jobs. Researchers plagiarize in order to get grants and / or recognition which will strengthen their reputation, and thus ability to work.
Programmers plagiarize to get work done, authors plagiarize to publish stuff, musicians plagiarize to release better music. workers plagiarize to get credit, and hopefully a promotion. And the list goes on.
In the end, people plagiarize because there are incentives to do so.
Siraj plagiarized because he's an "influencer", so to speak, in the Machine Learning beginners scene. Why did he do it? Probably to build reputation and credibility. I'm gonna go ahead and guess that he did it not to impress researchers or industry workers, but to look more productive and legit to his followers.
In the end, he probably doesn't give a sh!t about getting caught. If his main focus is on teacher beginners, then that's where the bank is.
It's like with stand-up, when someone shamelessly lifts a famous routine - other stand-up comedians will ostracize him/her, but that's a exceedingly small group of people. The audience will still laugh.
So, that's my take. As long as there's a clear incentive, and no punishment, then plagiarism will continue.
And unfortunately, you're more likely to hear "Who give a sh!t?" from casual bystanders. When discussing plagiarism, I've heard so, so many variants of just that.
"Who cares, everyone does it once in a while"
"Most would do the same in his/her position"
"That's not cheating, it's just solid problem solving"
36 comments
[ 3.1 ms ] story [ 85.3 ms ] threadAs such, the fundamental problem here is the world today values and rewards glossy content. Gobs and gobs of glossy content. Repackaged content. Aggregated content. Outright stolen content.
If you can make it at scale, views, subscribes and subsequent ad impressions will make it worth it.
And the good news: we're not discerning viewers. Someone stole something? We'll wag our finger, furrow our brow and click play anyway. There exists only incentive for producing.
I am reminded amusingly about how moderation at scale pleasing everyone is impossible by complaining about how many pages lower mirrors were than original websites on Google.
We used to find amusement and information from a single newspaper, a few ota television shows ...
Even a decade ago the plethora of streaming options we have now we're absent or in infancy. Monetizing your videos as an average Joe was not reality.
All that's changed. Influencers, content producers by the thousands are inventivized to create nonstop content to compete. That is new.
They haven't made any positive difference and cannot. The incentives for "gloss" to sell are the same and the alternatives have their own evils.
In journalism school you learn early on the history of newspapers in New York. The consequence was more competition lead to drastic price wars and a race to the bottom. By many accounts it culminated in yellow journalism.
That all found a natural equilibrium and things remained mostly unchanged until television. With that, competition for short, snippet journalism increased and newspapers moved to longer, more in depth journalism and larger shelved the very long form because the audience was gone.
Cable news had a similarly tectonic force on network news, which changed its drab, almost newspaper-y style to compete.
Each of these forces fundamentally changed the nature of these long lived formulas. Twitter shoved another dagger in newspaper journalism by forcing them to compete on social media, eliminating journalist standards for interaction with the public.
The gatekeepers react based on what audiences do. Audiences are not discerning. Each one of these events puts a dent in the integrity of its former self. It's all a race to the bottom for advertisement, something a lack of competition previously kept slow enough to seem at bay.
Problem statement: given context C and sentence X, find a sentence X' that communicates the same idea as sentence X but in different words. This is not exactly GPT-2, but it seems doable.
This kind of plagiarism would be difficult to detect. The plagiarizing party could say something like "I didn't copy or reproduce, I just used as input to a NN."
What I'm worried about is the appearance of authority when there is none, something that I guess Raval is also accused of. But what I'm looking at is a bit more subtle than pretending to have written a paper.
In several fields, you get someone popularizing certain topics and observations. Some guy will write a book about economics or psychology with some compelling storytelling, and many articles will be written reviewing the books. But actually nothing new has been added. It's simply a rehash of a few well worn ideas from an area. Importantly what can also happen is something that was considered marginal is brought up (Marxist/Austrian economics) as accepted mainstream. Or something with weak reproducibility (Marshmallow) is brought up as the accepted truth. Nothing that's a complete lie, but you get a picture of a field that is quite different to what the experts see it as.
The fact that they aren't making up complete lies is what makes them hard to critique. If someone comes up with a list of studies that seem to support some tenuous case, it becomes quite hard to rebut them without seeming quite pedantic. For instance you might need to dive into the methodology of some survey to understand the critique, but the soundbite culture of our times does not permit that.
Often the person writing these books is portrayed as someone very authoritative, and certainly that makes sense for the publisher. They get in the news, they give commentary on relevant news items, and the circus continues. They never really have to defend their positions, because they're always a sort of conduit for some other historical authority figure. But the total mix of what they're saying doesn't add up.
The problem with Raval is he is a narcissist. Check out his youtube apology video where he reads out, for 4 minutes, the names of the developers he plagiarised. It's a grand gesture suitable on veterans day or after a terrorist attack. He gets pleasure from the attention alone, others get pleasure from actually doing something. Maybe we should be asking the ESA who decided to give him a podium.
As it is said, Education is the art of telling progressively smaller lies.
Marketing and pop-book-writing is the reverse.
The most egregious example of this is Malcolm Gladwell. Not one of his books is true in any sense of the word even at the time he wrote them. Yet that didn't stop him from becoming an expert in ... expertise.
A lot of people who engage in these behaviors on the internet are narcissists, at minimum. At worst, they could be malignant narcissists (narcissists with sociopathic tendencies) or narcissists who are straight up sociopaths, too. What I mean is that it is not just straight up plagiarism: there is an unbelievably strange level of self-promotion from them on the internet, which is at minimum, a huge red flag. Their presence and "influence" on the internet is unavoidable, really. It is really tragic when they start going in to "research", which is a profoundly dumb move by them. At some point, they will get caught, and it will not be brushed off at all by other researchers and academics.
One should be alert to identifying these people on the internet. They are "smart" in the sense that they know how to play and manipulate people, all the time.
To identify potential narcissists online, I use this guide (although not scientifically verified): http://www.issendai.com/psychology/narcissism/narcissists-on...
Raval will, almost certainly, be fine. If he isn’t it will be owed not to the plagiarism alone, but to the combination of all of the scandals brewing around him. One plagiarized paper may be survivable, but a history of weak citation, a heavily-criticized course and a feeling he misled his community may be enough to make him a pariah.
Still, we’ve seen others survive much worse and it’s clear we’re entering an era where the title “plagiarist” doesn’t disqualify you from any position at all. It’s frustrating as many times it appropriate to sever all ties with plagiarists, but it’s the reality we live in.
In the end, it would be good if Raval could at least thoroughly reexamine his approach to citation, in everything he does. Citation isn’t just about formats and standards, it’s a thought process and a way of creating. Clearly, Raval doesn’t have that right now.
If nothing else good comes of this, maybe he can take this as a learning opportunity and grow from it.
Lead with `>` for quotes in case you were unaware.
It is in everyone's best interest to comment their code and provide where their information came from. More likely than not, you are no super-genius, and you needed this information to finish your project. It is highly unlikely that anybody can do this on their own.
It is no big deal to cite where you are getting your information. It is important in case somebody needs to review your project.
By the way, did I spot a Tom Lehrer reference?
Not many people found it convincing, especially as he did not apologize for the paid course plagiarism which brought his plagiarism mainstream.
Programmers plagiarize to get work done, authors plagiarize to publish stuff, musicians plagiarize to release better music. workers plagiarize to get credit, and hopefully a promotion. And the list goes on.
In the end, people plagiarize because there are incentives to do so.
Siraj plagiarized because he's an "influencer", so to speak, in the Machine Learning beginners scene. Why did he do it? Probably to build reputation and credibility. I'm gonna go ahead and guess that he did it not to impress researchers or industry workers, but to look more productive and legit to his followers.
In the end, he probably doesn't give a sh!t about getting caught. If his main focus is on teacher beginners, then that's where the bank is.
It's like with stand-up, when someone shamelessly lifts a famous routine - other stand-up comedians will ostracize him/her, but that's a exceedingly small group of people. The audience will still laugh.
So, that's my take. As long as there's a clear incentive, and no punishment, then plagiarism will continue.
And unfortunately, you're more likely to hear "Who give a sh!t?" from casual bystanders. When discussing plagiarism, I've heard so, so many variants of just that.
"Who cares, everyone does it once in a while"
"Most would do the same in his/her position"
"That's not cheating, it's just solid problem solving"
"It's been happening forever"
etc.
Currently it's washing it through machine translation.
But to the actual topic, the fun of lynching Siraj Raval...I don't get it.
Is the course a scam? No one has really said it is. Seems to be ok.
Not sure I know of many programmers who cite their stackoverflow which they must. He has cited as required, hasn't he?
Wooooo he plagarised on a non peer reviewed site. I assume it's quality is crap anyway.
Why are we talking about this? I don't get it.