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

So much of academia is about connections and reputation laundering

statmodeling.stat.columbia.edu

Post reply on HN
Print view View original

121–130 of 223 posts

Re: So much of academia is about connections and reputation laundering

#121

Post by naringas »

euix wrote: post #96
sytelus wrote: post #92

This article doesn't do good job describing the gory details. In AI/ML research, this is so unpleasantly prominent that its nauseating. How does workshop speakers at top conferences gets selected? How does keynote speakers get selected? Whoes workshop proposals goes through? Who gets to be area chair? Why are there 18 people listed as authors when almost 100% of work was done by one student? I call it favor economy .…

Putting people on the author list who made no contributions to the paper is known as academic fraud. I am surprised this is allowed.

it's only forbidden if you get caught?

Re: So much of academia is about connections and reputation laundering

#122

Post by gowld »

starpilot wrote: post #31

A moving average would make sense, but cubic fit? WTF? In epidemiology, is there a theoretical basis for death rates to follow any type of polynomial trend with time? If you measure say, drag force vs. airspeed at finite intervals, sure, use a second-order fit. Drag varies with the square of speed, theory predicts that relation. But there isn't such for death rates, is there?

The article explains that cubic models are backed by the theory that Excel built-in functions exist for a reason.

Re: So much of academia is about connections and reputation laundering

#123

Post by gowld »

CrazyStat wrote: post #115
scythe wrote: post #36

I'm just going to comment on the graph, not the tweets. The graph shows three predictions from IHME. One, the blue line, is a model from 3/27. It matches the data okay . The second, the yellow line, is a model from 4/5. The agreement of this line is much worse , and the fact that the model did worse with more data is not promising. The third, the teal line, is a model from 5/4. The data (black) ends at 5/4. So the ag…

> By "cubic fit" I infer that they mean some kind of low-pass signal filter. It's literally just ax^3 + bx^2 + cx + d, optimize a, b, c, and d to minimize some loss function (probably L2).

And since cubics can't be nearly flat over two disjoint intervals separated by an extrenum, they added an unlabelled nonsensical bell curve continuation to the to the end where a cubic would predict the case rate plummeting to negative infinity

Re: So much of academia is about connections and reputation laundering

#124

Post by roenxi »

rayiner wrote: post #110
roenxi wrote: post #69

Earlier quoted context omitted.

Steve Jobs didn't own Apple; he'd apparently sold his entire stake around the time he was ousted in 1985 [0]. Your evidence doesn't quite say what you think it does; you are talking about day-to-day management, I'm talking about ownership structure and how day-to-day management gets appointed/fired. But no specific evidence, just first principles reasoning. It is hard to see why it would do badly. [0] https://www.bus…

> But no specific evidence, just first principles reasoning. It is hard to see why it would do badly. There are concrete examples of worker-dominated organizations: schools, public transit entities, etc., where powerful worker unions dominate policy. They are almost universally unsuccessful, as worker interests take precedence over delivering a product to the consumer.

But none of those things are worker owned either; they are generally owned by the public. The success or failure of a school or public transport entity has nearly no impact on the success or failure of the workers (short of catastrophic mismanagement, anyway).

I suppose we have trailed things like worker ownership in early stage startups with high equity compensation. That might be evidence that it works well. Letting workers capture most of the value they create would be at least as interesting experiment for me than Universal Basic Income; but I think UBI has much more coverage as a political idea.

Re: So much of academia is about connections and reputation laundering

#125

Post by dnautics »

CrazyStat wrote: post #114
dnautics wrote: post #65

Earlier quoted context omitted.

on the other hand, for reals calculus (namely taylor polynomial estimation) also says, "why not a cubic polynomial?"

Because Taylor approximations can lose accuracy very quickly as you move away from the point they are centered around. There are available epidemiological models that are actually grounded in a scientific understanding of the problem. "Why not a cubic polynomial?" is a stupid question.

Yes, and until you do the correct analysis and estimate just how bad the estimate is going to be (off by 10%? Off by 50%? An order of magnitude?) You can't just dismiss it out of hand. So how far off will it be?

Re: So much of academia is about connections and reputation laundering

#126

Post by anjc »

>What’s striking is that the professor and A/Chairman doesn’t know that he doesn’t know. I’m struck by his ignorance of his ignorance, his willingness to think that he knows what he’s talking about when he doesn’t.

Does this make sense to anybody? The author is "struck" that somebody is "willing" to talk about something they believe they understand? The only annoying thing about academia is that it seems to attract smug, selfish people like this.

Re: So much of academia is about connections and reputation laundering

#127

Post by earthboundkid »

oneplane wrote: post #71
KittenInABox wrote: post #64

Earlier quoted context omitted.

> I think one advantage of computers is that what you produce at the end has to work - and if the code doesn't work or you don't know what you're doing, you're quickly exposed as a fraud. Is this actually true? You can bullshit coding...

Take the old guestbook example: if you can't add entries to the guestbook the code apparently doesn't work. No way around that.

Subcontract it to someone who writes it with security holes.

Re: So much of academia is about connections and reputation laundering

#128

Post by throwlaplace »

Myrmornis wrote: post #26
chrisseaton wrote: post #4

Very difficult to understand what's going here, through the snark, personal abuse, name calling, and political bias. Everyone's trying to communicate through increasingly scathing Tweets and declaring each time that the previous Tweet is yet another new low point. Not a useful way for anyone to get a point across. Everyone should be embarrassed.

No, I don't think that's right. That was my initial impression also, but what's going on is that people are ridiculing the academic advisor who is working at the Council of Economic Advisors at the Whitehouse because his agency (a) released a tweet about Covid with childishly absurd mistakes in it and then (b) he vigorously defended it in a follow-up tweet. The problem they are ridiculing is mainly that an observed t…

>Secondarily, the fact that the polynomial is described as cubic, when it appears to be quadratic.

wut. a quadratic only has one inflection point. that plotted curve clearly has 2.

Re: So much of academia is about connections and reputation laundering

#129

Post by sytelus »

euix wrote: post #96
sytelus wrote: post #92

This article doesn't do good job describing the gory details. In AI/ML research, this is so unpleasantly prominent that its nauseating. How does workshop speakers at top conferences gets selected? How does keynote speakers get selected? Whoes workshop proposals goes through? Who gets to be area chair? Why are there 18 people listed as authors when almost 100% of work was done by one student? I call it favor economy .…

Putting people on the author list who made no contributions to the paper is known as academic fraud. I am surprised this is allowed.

1 student and 4+ "advisors" aren't uncommon these days. They would show up in meetings, put their attendance and claim their spots. So one can argue its not really "fraud".

Re: So much of academia is about connections and reputation laundering

#130

Post by User23 »

TheOtherHobbes wrote: post #51
User23 wrote: post #9

Earlier quoted context omitted.

It’s difficult for me to take any discussion of imposter syndrome seriously that doesn’t consider the possibility the subject feels like an imposter because he is. Having been personally acquainted with several such imposters and having never seen such a discussion I’m left to conclude the entire concept is deeply unserious. Edit: sibling is making much the same point. The concept is only useful if actual imposters c…

You can't have impostor syndrome if you don't care about your ability to do your job. Likewise if you believe - probably correctly - that your failures don't matter because you can bullshit and bluster your way through them. And if that doesn't work, a quick word with your sponsors will sort out your problems. Impostor syndrome is the opposite - caring about quality, feeling you fall short (because quality is hard),…

There are at least three dimensions here: Feels like imposter, is imposter, and cares. I'm not going to enumerate all 8 values since that's trivial, but realize my point is that conflating the first two dimensions is what makes the exercise intellectually empty. There are people who feel like imposters (or aren't), are imposters (or aren't), and still don't care about their ability to do their job.

I'm still quite skeptical, but even this simple conversation is far deeper than any discussion I've seen on the subject. I would never have bothered to think about the caring dimension if you hadn't mentioned it.

Post reply on HN