Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Please commit more blatant academic fraud

jacobbuckman.com

Post reply on HN
Print view View original

161–170 of 365 posts

Re: Please commit more blatant academic fraud

#161

Post by YeGoblynQueenne »

baybal2 wrote: post #52

The motivation behind it is very simple: eye-watering salaries in the field. Moneyed people are ready to pay these salaries because a lot of them have succumbed to hype equating "AI" (that phrase alone is complete bullshit) to something bordering on "magic," and bringing equally magical money making opportunities. The prospect for business owners to get rid of white collar workers in a manner not unlike how machines…

>> The motivation behind it is very simple: eye-watering salaries in the field.

For a very broad meaning of "in the field". Academic salaries are kind of meh. The eye-watering salaries you mean are paid by large technology companies and I don't think those hire you on the merit of your publications, only. For example, allegedly google and facebook et al hire drop-outs from PhD programmes in highly rated institutions etc. so people who haven't yet had the chance to change the face of AI with their research, if I may be so bold.

Re: Please commit more blatant academic fraud

#162

Post by splithalf »

Cyril_HN wrote: post #117

In undergrad, my professor explained to me in very candid terms (from the position of being a generous mentor) that the optimal path to success in the Arts and Social Sciences is: 1. Find a niche only a couple of people operate in. 2. Make friends with them at all costs and work in their area. 3. Review each others work, amassing enormous citations in highly respected (albeit sometimes niche) journals.

This model has been applied to every field; all those “family run businesses” like the mafia, the American news industry, Harvard.

Re: Please commit more blatant academic fraud

#163

Post by Jabbles »

Q57C3HYc7g wrote: post #155
lettergram wrote: post #89

I manage an applied research group in the machine learning space, particularly AutoML and synthetic data generation. That means is we read the latest literature and create applications. Sometimes we build new research, most of the time we patent something new. But the goal is to get research to production as fast and as impactful as possible. The first thing I do when sorting research (papers, blog posts, presentatio…

Maybe this is a dumb question, but what do you mean by "run corrections based on the number of simulations they run"?

For example, https://en.wikipedia.org/wiki/Look-elsewhere_effect and more generally https://en.wikipedia.org/wiki/Multiple_comparisons_problem

Re: Please commit more blatant academic fraud

#164

Post by goalieca »

lettergram wrote: post #89

I manage an applied research group in the machine learning space, particularly AutoML and synthetic data generation. That means is we read the latest literature and create applications. Sometimes we build new research, most of the time we patent something new. But the goal is to get research to production as fast and as impactful as possible. The first thing I do when sorting research (papers, blog posts, presentatio…

I despised how data sets and code were considered a gold-mine and researchers would often refuse to release them lest they give up a serious paper publishing advantage for their group. More often than not, i found code full of serious errors and completely lacking basic test cases and sanity. Sometimes the data sets were much larger than the subset chosen for publication.

Re: Please commit more blatant academic fraud

#165

Post by unishark »

bumby wrote: post #94
josalhor wrote: post #82

Earlier quoted context omitted.

Well, I wonder if this could force people to start publishing their code so others can replicate the results. This would help greatly with the kinds of issues that are being discussed: - It would be much harder to hide makeup results and cherry-picked seeds - Useless research would not be as easy to hide - The real-world impact of the research would be more valued. - Researchers would be forced to recheck their code…

Publishing the code may not be enough. Part of the problem is that replication of a study isn’t usually rewarded in the academic sphere because of the obsession with novelty within publications. Most researchers want to spend their finite time on something they find interesting and also may help their career. One way may be to create a movement to foster good replication studies that can be published and rewarded.

Novelty is in the job description though. I think the problem, especially in AI/ML is the obsession with "State of the Art" algorithms. This can be gamed by making claims with numerical "experiments". If an idea had to stand on its own, and be interesting even if it may not make a better pattern-recognition product, most of them probably wouldn't make it. Indeed that's essentially the requirements presented by reviewers and editors: "prove we should read about this otherwise-uninteresting method by giving numerical test results where it wins".

Re: Please commit more blatant academic fraud

#166

Post by YeGoblynQueenne »

alisonkisk wrote: post #112
baybal2 wrote: post #52

The motivation behind it is very simple: eye-watering salaries in the field. Moneyed people are ready to pay these salaries because a lot of them have succumbed to hype equating "AI" (that phrase alone is complete bullshit) to something bordering on "magic," and bringing equally magical money making opportunities. The prospect for business owners to get rid of white collar workers in a manner not unlike how machines…

The eye watering salaries come from passing a whiteboard interview, not publishing papers.

This!

Re: Please commit more blatant academic fraud

#167

Post by frickinLasers »

h2odragon wrote: post #135
Al-Khwarizmi wrote: post #126

Earlier quoted context omitted.

So what part of falling in love with people in the lab, people in the lab falling in love with you, and that affecting concentration and productivity, is so evil as to get someone fired? Sorry, honest question. Being from a different culture, I honestly fail to get this kind of outrage. For me it's just a description of humans being human...

> a description of humans being human... Precisely. There's some sort of "if you have authority you must be better than me" feeling and "better than me" admits no flaws or human variety at all , apparently. Some folks want perfect Gods to follow and keep failing to make them from people made of meat.

It is difficult to distinguish genuine romantic feelings between two people, from the case of a superior using their position to get their genitals wet and a subordinate capitulating for fear of losing their job.

Since feelings are only a biological impulse, and we humans frequently suppress our impulses in the form of self control, it's much easier to look for that oxytocin fix in a more appropriate arena.

The military has been doing this for ages, forbidding officers from fraternizing with enlisted. And plenty of civilians abusing positions of power have proven the wisdom of such a policy.

Re: Please commit more blatant academic fraud

#168

Post by immmmmm »

FabHK wrote: post #62
MattGaiser wrote: post #10

Earlier quoted context omitted.

With laws, there are full time police officers, investigators, and prosecutors working to catch people. And even then, most crimes go unsolved and unpunished. Even without the complicity, stopping fraud and bad behavior is a difficult problem. Part time reviewers and reputation seem woefully insufficient counters.

That would be interesting, if universities had full time academic integrity officers whose job was to detect academic fraud.

after having spent a third of my life in academia i don't see this happening: everyone want as many as possible articles, if possible in the most high impact journals. detecting fraudulent or irrelevant pieces is against that goal.

Re: Please commit more blatant academic fraud

#169

Post by YeGoblynQueenne »

sillysaurusx wrote: post #23
2sk21 wrote: post #6

Thank goodness I am retired and don't have to publish stuff in the current environment. Thirty years ago when machine learning was a sleepy backwater, life was a lot easier for us.

As a present-day researcher (or at least someone who aspires to be), I'm eagerly waiting for the hype to die down, and for ML to become a sleepy backwater again. You were lucky to play around with ML during those days. The neat thing is, the next AI winter, we'll still have massive hardware rigs. That's very different from 30 years ago. It's a lot easier to discover new things when you can just test every possibility…

>> The neat thing is, the next AI winter, we'll still have massive hardware rigs.

So the next big thing in AI will still come from brute-forcing solutions by sheer power of compute. To clarify, you say this is a good thing?

Re: Please commit more blatant academic fraud

#170

Post by skystarman »

berndi wrote: post #28
berndi wrote: post #11

Wow, the amount of depravity in some academic circles is astonishing. The article links [1] the case of a PhD student at the University of Florida who was forced to participate in such a publication collusion ring and was pressured to commit scientific fraud by fabricating results and submitting them to a conference [2], being threatened with physical harm should he decide to go public. This student saw no way out an…

How is it possible that professors with impeccable academic credentials get fired for jokes nearly instantly [1], yet this student's professor was allowed to carry on until he resigned two years later [2]? [1] https://en.wikipedia.org/wiki/Online_shaming#Tim_Hunt_contro... [2] https://eu.gainesville.com/story/news/education/campus/2021/...

Comparing these personnel decisions from not only different universities and entirely different incidents but also an entirely different country seems unwise to me.
Post reply on HN