I think people are hesitant to over-criticise academia. Maybe they can already hear the resulting anti-science sentiment, and fear losing hard fought ground if academia is called into question.
However, I think there are serious signs that we should be looking for radical changes to the metstructures of academia. They're not necessarily related, but they point to something being really wrong.
First is the replicability crisis. Have we actually been accumulating knowledge in fields where this is bad? Do we know more about human behaviour due to the human behaviour research done in the last generation? What's the point of all these "more research needed" conclusions. The process of academic publishing and peer reveiw directs this whole thing, and it doesn't seem to be all that well directed..
While we're on publishing, everyone in academia complains bitterly about the publish-or-perish problem and about grant/funding politics.
One of the big complaints about grants is that senior academics spend all their time on grants and administration. Success in many fields is more a function of being good on administration and politics, than being good at research.
While we're on administration... the number of administrators (not teachers or researchers) has skyrocketed over the last 2 generations.
Ultimately, I think that the "scientific method" in practice is in large part embodied the academic publishing system. ...the algorithm determining how science works.
> I think people are hesitant to over-criticise academia.
Really? Maybe inside academia itself that's true, but I think people outside are fairly eager to criticize academia and put down academics.
I feel like anytime a study with even a slightly intriguing/controversial abstract gets circulated, one of the following happens (in order of frequency):
1) Narrow criticism of something very specific (usually sample sizes) with very little understanding about why that criticism might be misinformed (again, usually relating to sample sizes and how statistical significance works) which they use to discredit the whole paper
2) Pointing to previous papers that may (but often don't) contradict what the paper in question is suggesting - and "suggesting" is a loose term here since people tend to draw all sorts of conclusions that the author often doesn't. This is then used to basically throw hands up and say "it's unknowable."
3) Just plain dismissal, especially when the paper's implication is something that the reader doesn't want to hear. This comes in all forms, from the pedigree-based "oh who cares what someone from Montana State says what do they know" to the more direct ones like "why the hell would I listen to Bernanke talk about business, he's an ivory tower economist"
In any case, I hesitate a little, for the above reasons.
Criticizing papers is..erm.. good and proper, I think. Sometimes the criticism will be crap, but a critical instinct isn't a bad thing here. Bias is, c'est la vie.
Critising academics... I feel we're far too comfortable dishing out personal criticism in the internet era. Leave that one there.
Criticising the system... All those "flags" that I raised are actually common and recurring criticisms of the system, within academia. But, they seem to fall short on how meta they go. Publish-or-perish & grant systems, the insider criticism sounds more like "my boss is a moron" talk, cheap shots.
What I'm hoping for is big ideas from academia, I guess. Ideas about how to structure the institutions of academia: publishing, grants, whatever else determines what research gets done, by whom. Something that promotes efficiency.^ Something that promotes completion/conclusion so that we end up with knowledge, not a collection of research findings. Negative results and data pooling. Seperation of data & interpretation. Replication. This is not an excel-vs-R problem. It's an "incentives-within-academia" issue.
^Remember, there're problems that these grant bureaucracies are clumsily trying to solve. This is a market, and a bad market can be a lot worse than a good one.
I think there are two main kinds of criticism here; the first comes from people who think a lot of published work is misleading at best and fraudulent at worst (see replicability crisis); the second comes from people who think there's something horribly wrong with the culture of academia, notwithstanding the quality of output. Criticism in the second category touches on topics like adjunct professors (most instructors of undergraduate courses are temp workers making near minimum wage with no benefits), the takeover of universities by a class of professional administrators (many of whom have unbelievably cushy jobs), the distribution of research grants to researchers with the most political pull rather than those doing the most interesting work, the pressure put on researchers to acquire grants (see Stephan Grimm), and the crushing debt burden most undergraduates take on to get a university degree (most of which is spent frivolously while instructors and researchers work themselves to death). I think what you've described sits more in the first category than the second, while the parent belongs more to the second.
Here are some related quotes on social problems in science cover a wide range of concerns (from an essay I wrote in 2011:
http://pdfernhout.net/to-james-randi-on-skepticism-about-mai... ) -- although perhaps they mostly all fit under the broad categories of fraud or culture as you suggest? Even if they all do fall into one or the other, perhaps one could use them -- long with your examples -- to begin to categorize the specific types of fraud and the types of dysfunctional cultural interactions and then begin to try to assess their frequency and impact?
From an article about a sociologist and anthropologist who studies science and technology, Bruno Latour: http://en.wikipedia.org/wiki/Bruno_Latour "In the laboratory, Latour and Woolgar observed that a typical experiment produces only inconclusive data that is attributed to failure of the apparatus or experimental method, and that a large part of scientific training involves learning how to make the subjective decision of what data to keep and what data to throw out. To an untrained outsider, Latour and Woolgar argued the entire process resembles not an unbiased search for truth and accuracy but a mechanism for ignoring data that contradicts scientific orthodoxy."
A quote from another academic, Brian Martin, involved with Science and Technology Studies: https://web.archive.org/web/20100221213343/http://www.suppre... "Textbooks present science as a noble search for truth, in which progress depends on questioning established ideas. But for many scientists, this is a cruel myth. They know from bitter experience that disagreeing with the dominant view is dangerous - especially when that view is backed by powerful interest groups. Call it suppression of intellectual dissent. The usual pattern is that someone does research or speaks out in a way that threatens a powerful interest group, typically a government, industry or professional body. As a result, representatives of that group attack the critic's ideas or the critic personally-by censoring writing, blocking publications, denying appointments or promotions, withdrawing research grants, taking legal actions, harassing, blacklisting, spreading rumors. (1)"
From David Goodstein, who was Vice Provost of Caltech: http://www.its.caltech.edu/~dg/crunch_art.html "Peer review is usually quite a good way to identify valid science. Of course, a referee will occasionally fail to appreciate a truly visionary or revolutionary idea, but by and large, peer review works pretty well so long as scientific validity is the only issue at stake. However, it is not at all suited to arbitrate an intense competition for research funds or for editorial space in prestigious journals. There are many reasons for this, not the least being the fact that the referees have an obvious conflict of interest, since they are themselves competitors for the same resources. This point seems to be another one of those relativistic anomalies, obvious to any outside observer, but invisible to those of us who are falling into the black hole. It would take impossibly high ethical standards for referees to avoid taking advantage of their privileged anonymity to advance their own interests, but as time goes on, more and more referees have their ethical standards eroded as a consequence of having themselves been victimized by unfair reviews when they were authors. Peer review is thus one among many examples of practices that were well suited t...
I think it is reasonable to question academia and research but I think a problem, often, is people want science and research to give us answers. It does not, it provides evidence, and once that evidence gets supported enough times it is viewed as closer and closer to reality. I would say it is asymptotic though, we never reach what is fact because there can always be a missing piece of evidence that shows a different result.
There is one statement you make that I do want to directly comment on. The point of more research is needed is simply, that more research is needed. It goes directly to your statement of "do we know more about human behavior due to the ..." We learn more with all research and the way research currently gets disseminated is through publication. More research is needed publications as well as null result research is extremely useful in the research community and often does not get the attention outside (or inside sadly) the scientific community that it deserves.
I'm not arguing at the philosophical level. My views are fairly mundane on that. I'm talking about the implementation of that philosophy (where I agree with you) in academia, as a system.
When I say "build knowledge", I'm speaking colloquially. For a contrived example, we know traits are inherited by animals via genetics. We're confident enough to treat this as knowledge and design other expirements that take this as an assumption. Are we building this kind of knowledge in social science, currently?
Does "more research needed" lead to more research being done, typically in a field? If not, why? Why are we moving on to a new question, when the last one is still unanswered.
Take a "willpower is a limited resource" hypothesis (replication issues aside). It does not seem impossible to answer this question. Nothing in science is "knowledge" in an absolute sense, but we can get to some level of confidence.
So.. do we want to know the answer to this question?
If not, why do the the first expirement? If yes, then what would it take to get to an answer? We can answer that partially in advance. Independant replication will be part of the answer. Sufficient sample size also. If we know in advance that these things won't be done, what is the point?
I'm an outsider and I realize I may be throwing unwarranted dirt, so if/where I am, apologies. I think that in some fields we systematically do not do the things necessary to get to useful knowledge, a sufficient accumulation of evidence to treat a result as more than anecdote.
I've been working full time on trying to help solve one of the problems in academia (access to research), and although I was superficially aware of most of the problems in academia, once you start to dive into one you will learn a lot about the others as well, as they're somewhat intertwined. And really, the scope and breadth of the problems is astonishing.
Personally, my hunch is that the root cause for most of them is a somewhat misguided focus on "excellence" combined with it being practically impossible to measure researcher performance with a reasonable amount of effort. Then again, the problems are so intertwined and have so many different actors, that finding the root cause is really hard, so who knows whether I'm correct.
I think measuring research performance is a dead end. People can do this, although they carry a ton of blindness and blindspots. Bureaucracies, systems and such can't. They end up measuring what they can measure, like grades, attendance, number of published papers, number of citations maybe... It just won't end well.
But, we can have systems that promote performance without measuring it. I think those are the kinds of ideas we should look at.
I agree with your first paragraph, but I don't see how your second paragraph would work. Do you have any concrete examples/ideas of how such a system could work? (Without requiring e.g. those on tenure committees to read and understand all the research of all applications - unless you believe that could work somehow.)
Nothing specific. I'm also not an academic, so anything I say will be a little contrived but... ideas generally... I think the first step is look outside the basic, low-level structure and look a level or two above it. I think the best ideas will likely touch on a few of the related problems and happen outside of individual departments. Either the publication or the funding structures seem like good places to start.
Publishing... This is obviously a centrepiece institution. It's central to replication/quality issues, to publish-or-perish career incentives and to funding allocation. So, I guess any radical ideas here will likely impact on a lot.
Can we modernize publication, make it more open, and update the peer review system? One step is to move away from seeing "peer reviewed" as a binary. Another might be increasing the role of peer review. Maybe papers should have critical "reviewer notes."
Maybe publications could take more of a 'director' role, influencing follow up, replications, being critical of practices...
..If a paper concludes that "more research is needed," does this mean anything? Will the research be done? What exactly needs to be done (eg, same study @ 5X the sample size). Is this likely to yield a result? Is that result important?
A big part of the replication problem (from my perspective, safely on my couch) seems to be that very little replication is done, because the incentives are all wrong.
If institutions were different (eg capitalism, for example) I could imagine the problem being reversed: replication and further work could be overinvested in, impoverishing exploratory science instead of the other way around. Replication is a lot more concrete (and money men love concrete).. you know what you need to do specifically, and what you might achieve by doing it. The problem in academia is that original work, even unreplicated an inconclusive is valued higher than subsequent work. We have to incentivize replication. In social sciences, we're sitting on a mountain of inconclusive results. Unless this work is done, we can't build on.
Funding... Measuring researcher output well enough to base funding on is not possible, imo. But, maybe we'd have better luck with portfolios. Ie, structure funding bodies as competing portfolio managers, looking to impress based on portfolio performance. This might give us a better mix of predictable returns (eg replicate these 175 studies) and long tail risky stuff.
These are interesting thoughts, so thank you for sharing. Especially the idea that reversed incentives (for replication) might indeed not be desirable as well was insightful to me.
I very much agree that it would be great if "peer reviewed" was not seen as a binary - because the research it represents is not binary. However, basing funding decisions on a lot of portfolio's full of non-binary research results is still, I'm afraid, too much work for e.g. tenure committees.
(The annoying thing here is that just because a system like what you describe is still far from perfect, people will stick with the system that's in place even though that's far worse...)
> I think people are hesitant to over-criticise academia.
is a message to people here. Of course, there's all the "anti-global warming, intelligent design flat earther republicans" who have always criticized academia, but that's not what you're talking about (I think).
I think you mean that people here - i.e. relatively educated, orthodox liberals for the most part - should really start to look hard at academia.
I'm someone with a traditional academic pedigree (double major in Econ + Math from MIT, most of my friends have PhDs) but I, too, am growing increasingly wary of much published research.
For example, there's this eye-opening blog post[0] by Andrew Gelman, a professor of statistics at Columbia who collaborates on a lot of psychology research, about the problem of bad statistics in research in psychology. There's this [1] famous paper by the Fed about problems of replication in what I studied, economics. Or how the NIMH "is re-orienting its research away from DSM categories".[2] And then there's the endless pop-science ideas that show up and get shot down like Power Pose, Air Rage, or Learning Styles. Or the constantly changing recommendations on nutrition.
I don't really take for granted anymore that academia is generally working towards greater understanding of the world. It's an imperfect system that has done some incredible things, but it definitely is dealing with some serious issues at the moment. I hope that initiatives like journals for null results, changes in the way grants are apportioned (like in this posted article), etc, can put it back on track.
The biggest problem with funding ideas - they fail more than proven conservative projects. When you select idea that fails, you can be accused of wasting funds, and not selected for next funding meeting.
It seems from an article that the funding agencies were pleased with the return they got from an "ideas" portfolio. Maybe more ideas failed, but the projects that succeeded did so in more spectacular ways — a bit like a VC portfolio.
The problem, it seems to me, is similar to VC: The funding agencies can spread their risk across a pool of ideas, but the investigators generally have to commit to one or a small number of projects for some time, so they risk spending several years without much to show for it if the idea fails.
Peer review is not about funding the best science, it is about funding the best connected and ensuring those that have climbed to the top keep their position even if they are out of ideas.
More seriously, the problem with approaches like this is the people who are chosen would be mad to actually work on the idea they put up. At most the funding is for 18 month and what happens if at the end of that time you have nothing?
If you want this idea to work the funding has to be for at least 10 years so that people can afford to take real risk and have enough time to build a track record if the idea doesn't pan out.
not sure this is entirely fair. While there's certainly some cronyism (see my earlier comment), study sections are mostly about funding the safest science - i.e., the project with the least chance of failure and the fewest flaws. But they don't have to be.
Many of the 'big idea'-focused mechanisms at NIH are 4 or 5 years - plenty of time to make real headway. These would work if we supported enough of them.
Parenthetically, high-risk science does not mean shooting the moon - a well-designed project, even if it's high risk, will still yield /something/ valuable both for the field, and for the investigator's career development. That's why reviewers tend to look closely at the alternatives/pitfalls section of grants - what will you do if the experiment blows up?
I am somewhat biased by having been through the Australian system where cronyism is alive and well.
If you have 4 or 5 years you really only have 2 years before you have to switch to low risk if things aren't working out to give yourself time to get enough papers out before you need to apply for the next grant.
What I did when I was an academic was spend 70% of my and my students/postdocs time on low risk activities and 30% on high risk. This keeps the risk down and still allow for some dreams.
- I started defending the value of good ideas 10 years ago in Quora. I debated with venture capitalist and entrepreneurs. They took me as a fool.
- I debated about the value of ideas in HN. They took me as a fool. "Ideas are worthless, Execution is king".
Now, I have a brain used to find good ideas quickly. My fuzzy process of analizing something complex is well advanced. And no venture capitalist, coder or entrepreneur will be able to compete with me cause I have years of advantage.
I've also trained emotional resilience and self-confidence cause I knew I was right even when the successful people didn't agree.
I don't even need to read the article to comment on this. Ideas are my area of expertise. I spent 7 years without working and letting my mind free, thinking about thinking.
Hacker News shepple: venture capitalists and entrepreneur sheeple: FUCK OFF!!!
Anyway, I know this sheeple will find another way to attract money for projects, cause they say just what others repeat and what people wants to hear. No what people need to hear.
Nature or another scientific magazine will publish an article in 2 years time about the power and value of being frank, blunt and honest. And how it helps the mind. Meanwhile, I will be taken again as a fool again.
Really: FUCK OFF!!!!!!!!!
I will do my fucking projects without any external money.
While I generally agree with your sentiment about building without external money (ie, bootstrapping), you may come across with your points better if you were a little more civilized and decent about it.
I happened to be evaluating a sentiment analysis service this morning and you'll be glad to know that your comment is regarded as being significantly more positive than the grandparent... :-)
Over the past decade I've sat on 10+ NIH study sections, and chaired a few. This commentary is spot on. Consider the response to a flaw in a grant. For a newer investigator or one no one has heard of, there will be much hemming and hawing about concern about ability to execute the project. This is code for, "who is this person in our sandbox anyway?" For one of us older guys (yes, unfortunately the pronoun is mostly right), the response is, "but I'm sure he can get this done, he must have thought about this." Even a modest scoring 'bonus' for new investigators doesn't correct this bias.
On the novelty front, reviewers are trained to focus on finding flaws or absence of detail or preliminary data. The presence of any of these requires that we score down - so it's just about impossible for something truly new to get funded. Until we do the project, we don't necessarily know what problems we'll encounter or what the data will look like.
The small number of NIH grants aimed at funding high-risk work by earlier-career folks (eg, the DP2 mechanism) are a great development but a tiny fraction of the portfolio.
The key thing to recognize is that any move to distribute funding towards younger folks or higher-risk work causes the old guard to scream bloody murder - "what do you mean I can't have 5 R01's at a time?". For study sections, I've observed first hand that old habits favoring established investigators and low-risk, incremental work are hard to break.
sure - they're the most common NIH grant mechanism - generally 4 or 5 years @ up to $500k/year (though they can be smaller, which NIH program officers always appreciate).
If you plot the number of NIH grants per investigator, there are a small but significant number of investigators out in the right tail with 4+ R01's. In general grants beget grants, such that bigger and better-funded labs are better positioned to get the next grant. Details re R01 mechanism are here [1].
A recent proposal that would cap the number of grants an investigator could lead at one time was shot down by the old guard [2]. NIH continues to struggle for alternatives, because unless the overall budget grows, which it will not in the foreseeable future, any strategy will involve taking from the rich(er) and giving to the poor(er).
Also for those who are curious, past and present NIH grants (e.g., grant names, numbers, summaries, award amounts, principal investigators) can be queried here:
By the way, I'm one of those people whose grants you turned down (well, I don't know if it was your study section). After several attempts at getting R01s with smart new ideas and receiving no feedback and no scores, and seeing the same proposals being funded a few years later for more experienced applicants, I reasoned that I could get more done in industry.
I left for Google, and got more done in 20% time (research-wise) than I ever did using 100% time in academia.
When I did sit on study sections, I worked hard to help NIH understsand why the older groups asking for closet clusters on a cloud grant weren't helping. But NIH is very slow to move- it took at least 7 years to get to the point where people could even apply for cloud credits instead of closet clusters.
You make a very good point. As PG says explains in [1], in a regime dominated by tail events, if most attempts are a moderate success (and no black swan successes), then the portfolio was probably too conservative.
On the other hand, due to limited resources, I get the feeling that most academics are operating in the risk-averse mode. See, for eg [2] -- it seems to be solid advice for succeeding in academia (I'd like to hear counterpoints to that from people who've had an extended/successful academic career), but it seems wholly focused on small predictable improvements.
I wonder whether this problem is a consequence of enforcing "accountability" in a regime dominated by fat tails.
Maybe we've picked up all the low-hanging fruit in research and we can only hope for slow incremental progress (which is what the management structure is set up to optimize), and we need to come to terms with that and stop wistful discussions about not having black swan successes.
That was my experience, and in retrospect it took my graduate advisor a decade to land an R01 based on the research I did while I was in school. She is a mid career tenured faculty at a decent school. Her husband is an endowed professor at a research one school and they practically call him on the phone to give him money (That is total hyperbole, but he lands funding at a much higher rate in spite of the fact that both have successful research programs). It seems really difficult to break into the sphere of 'successful PI' when so mnany other successful PIs win the majority of the funding. It was one of the reasons I left academia. Not the only reason, but it figured into it.
I have served on dozens of NSF computer science funding panels at the small (500k), medium (1.5m), and large (3m) levels. These review panels are not blind, and decisions are based on (a) what problems are explained in the grant, and (b) prior track record and proven expertise in the area.
In my own circumstance and what I know from my colleagues, CS researchers rarely write their best ideas in grants---not because they are afraid the ideas are too bold---but rather because those ideas are often not fully worked out, and nobody wants to just give those away to a review panel full of top-rank scientists who might make connections faster than you!
The problem with "blind review" is that project proposals can rarely be anonymous because just explaining the work and citing the relevant prior work leaks a lot about the possible author. So making review blind can often be an advantage to the higher profile researcher and a disadvantage to the capable but slightly less well known one.
That said, as my own experiment, during my next NSF review, I am going to tear away the first pages of all the proposals, and make my first pass without knowing the authors to see if it makes a difference.
And the thesis of almost all of venture capital, all of whom have made a lot of money on the insight and innovation of the past, but which has failed to created any new innovation of any significance.
This should be better said as, "Fund fresh insight, to find fresh insight." And while that sounds tautologically obvious, almost no one is doing it.
Upvoted the comment that says "I think this goes against YC philosophy". The YC guide explicitly calls out team experience and likelihood to succeed as a criterion.
I think there's a fetishization of complete blind objectivity in so many fields right now, which has at the root of it trying to iron out inequality. A laudable goal, but you shouldn't let that ideal completely upend the practical question to yourself of, where am I going to place my bets?
An unavoidable fact is that ideas alone are not enough to make some technology or venture successful. Sure, some one-off inventor may have a brilliant idea, but the idea is just like 1% of the problem.
If your model is to fund a team to develop an idea, the track record / personalities of the team are quite important.
Most people in this world are barely able to get their own lives under control. What's the likelihood that a random person with a great idea can take that to something commercially viable? There's a reason that who it is matters, and unfortunately, that still leads to people being selected who reflect the starting set of less-than-diverse people.
Most researchers applying for NIH grants aren't "random people"; all of them have gotten through PhD programs and potentially have done one or two postdocs, which already indicates that they have sufficient "team experience" and are already quite accomplished in their own right.
This article isn't saying that anyone should be able to get an NIH grant, but rather that grants should be awarded less due to politics and more based upon potential merit and the promise of an idea. I.e., This editorial is being published in Nature, not viXra.
The central issue is not one of team dynamics, or rooting out inequality on a societal level, but rather about rooting out inequality among equally capable candidates. Currently, many academic fields are dominated by cronyism and that is what the article is advocating against.
YC's philosophy is basically make noise lots of noise and the rest shall follow. They are never going to pay attention to something that cannot generate lots of buzz.
I would never expect something fresh or innovative to come out from what we would call "proper" people because these are the people less willing to change and question anything.
The problem is much deeper.
If I am having good time the bast policy to follow is the one involving fewer changes so I could continue having good time.
Change comes mostly from the ones not having good time the "loosers" you might call
> There's a reason that who it is matters, and unfortunately, that still leads to people being selected who reflect the starting set of less-than-diverse people.
I'd agree with arguments that this is true more because of network effects than whether others can do the job or not.
If you're only looking to repeat success, skilled folks are the best bet.
But it leaves others who want to learn behind. And a lot of the folks with skills got there because someone propped them up.
I'm just totally done with the idea our economy needs to be this fucking "always grow, always invent" scheme. So much of it is pointless shit that doesn't add to human discovery, rather it serves to change values in a bank account.
I'd rather we focus on getting the basics right; healthcare, food, shelter, for the masses. Make that our daily focus, and give people time otherwise to find their talent or cool idea.
But we're too addicted to universities telling us economics are like laws of physics and thus we must abide and live by them. Which is not much different than churches telling us the Bible is like laws of physics and the only model for living and we must abide.
Smart folks get lost in their discoveries and pretend they apply to all of us. And manipulative folks will take that and make it the law of the land for their benefit.
This might be a fine way to fund innovation (taking invention and turning it into capital ) but it's a terrible way to fund invention. By definition there is no idea when you set out to invent.
innovation produces billions of dollars of wealth . Invention produces trillions. Are we sure we have our priorities straight as a species?
I think this applies to YC-type incubators and VC investment as well. Often times pedigree seems to outweigh fundamentals regarding execution in VC funding
One other thing worth mentioning is the obsession of funders with hypothesis-driven research. They do this out of defensiveness to justify spending, but in the long term, it reinforces a daisy-chain of stale ideas and a selection bias that distorts data to adhere to popular models.
47 comments
[ 3.4 ms ] story [ 102 ms ] threadHowever, I think there are serious signs that we should be looking for radical changes to the metstructures of academia. They're not necessarily related, but they point to something being really wrong.
First is the replicability crisis. Have we actually been accumulating knowledge in fields where this is bad? Do we know more about human behaviour due to the human behaviour research done in the last generation? What's the point of all these "more research needed" conclusions. The process of academic publishing and peer reveiw directs this whole thing, and it doesn't seem to be all that well directed..
While we're on publishing, everyone in academia complains bitterly about the publish-or-perish problem and about grant/funding politics.
One of the big complaints about grants is that senior academics spend all their time on grants and administration. Success in many fields is more a function of being good on administration and politics, than being good at research.
While we're on administration... the number of administrators (not teachers or researchers) has skyrocketed over the last 2 generations.
Ultimately, I think that the "scientific method" in practice is in large part embodied the academic publishing system. ...the algorithm determining how science works.
Really? Maybe inside academia itself that's true, but I think people outside are fairly eager to criticize academia and put down academics.
I feel like anytime a study with even a slightly intriguing/controversial abstract gets circulated, one of the following happens (in order of frequency):
1) Narrow criticism of something very specific (usually sample sizes) with very little understanding about why that criticism might be misinformed (again, usually relating to sample sizes and how statistical significance works) which they use to discredit the whole paper
2) Pointing to previous papers that may (but often don't) contradict what the paper in question is suggesting - and "suggesting" is a loose term here since people tend to draw all sorts of conclusions that the author often doesn't. This is then used to basically throw hands up and say "it's unknowable."
3) Just plain dismissal, especially when the paper's implication is something that the reader doesn't want to hear. This comes in all forms, from the pedigree-based "oh who cares what someone from Montana State says what do they know" to the more direct ones like "why the hell would I listen to Bernanke talk about business, he's an ivory tower economist"
In any case, I hesitate a little, for the above reasons.
Criticizing papers is..erm.. good and proper, I think. Sometimes the criticism will be crap, but a critical instinct isn't a bad thing here. Bias is, c'est la vie.
Critising academics... I feel we're far too comfortable dishing out personal criticism in the internet era. Leave that one there.
Criticising the system... All those "flags" that I raised are actually common and recurring criticisms of the system, within academia. But, they seem to fall short on how meta they go. Publish-or-perish & grant systems, the insider criticism sounds more like "my boss is a moron" talk, cheap shots.
What I'm hoping for is big ideas from academia, I guess. Ideas about how to structure the institutions of academia: publishing, grants, whatever else determines what research gets done, by whom. Something that promotes efficiency.^ Something that promotes completion/conclusion so that we end up with knowledge, not a collection of research findings. Negative results and data pooling. Seperation of data & interpretation. Replication. This is not an excel-vs-R problem. It's an "incentives-within-academia" issue.
^Remember, there're problems that these grant bureaucracies are clumsily trying to solve. This is a market, and a bad market can be a lot worse than a good one.
From an article about a sociologist and anthropologist who studies science and technology, Bruno Latour: http://en.wikipedia.org/wiki/Bruno_Latour "In the laboratory, Latour and Woolgar observed that a typical experiment produces only inconclusive data that is attributed to failure of the apparatus or experimental method, and that a large part of scientific training involves learning how to make the subjective decision of what data to keep and what data to throw out. To an untrained outsider, Latour and Woolgar argued the entire process resembles not an unbiased search for truth and accuracy but a mechanism for ignoring data that contradicts scientific orthodoxy."
A quote from another academic, Brian Martin, involved with Science and Technology Studies: https://web.archive.org/web/20100221213343/http://www.suppre... "Textbooks present science as a noble search for truth, in which progress depends on questioning established ideas. But for many scientists, this is a cruel myth. They know from bitter experience that disagreeing with the dominant view is dangerous - especially when that view is backed by powerful interest groups. Call it suppression of intellectual dissent. The usual pattern is that someone does research or speaks out in a way that threatens a powerful interest group, typically a government, industry or professional body. As a result, representatives of that group attack the critic's ideas or the critic personally-by censoring writing, blocking publications, denying appointments or promotions, withdrawing research grants, taking legal actions, harassing, blacklisting, spreading rumors. (1)"
From David Goodstein, who was Vice Provost of Caltech: http://www.its.caltech.edu/~dg/crunch_art.html "Peer review is usually quite a good way to identify valid science. Of course, a referee will occasionally fail to appreciate a truly visionary or revolutionary idea, but by and large, peer review works pretty well so long as scientific validity is the only issue at stake. However, it is not at all suited to arbitrate an intense competition for research funds or for editorial space in prestigious journals. There are many reasons for this, not the least being the fact that the referees have an obvious conflict of interest, since they are themselves competitors for the same resources. This point seems to be another one of those relativistic anomalies, obvious to any outside observer, but invisible to those of us who are falling into the black hole. It would take impossibly high ethical standards for referees to avoid taking advantage of their privileged anonymity to advance their own interests, but as time goes on, more and more referees have their ethical standards eroded as a consequence of having themselves been victimized by unfair reviews when they were authors. Peer review is thus one among many examples of practices that were well suited t...
There is one statement you make that I do want to directly comment on. The point of more research is needed is simply, that more research is needed. It goes directly to your statement of "do we know more about human behavior due to the ..." We learn more with all research and the way research currently gets disseminated is through publication. More research is needed publications as well as null result research is extremely useful in the research community and often does not get the attention outside (or inside sadly) the scientific community that it deserves.
When I say "build knowledge", I'm speaking colloquially. For a contrived example, we know traits are inherited by animals via genetics. We're confident enough to treat this as knowledge and design other expirements that take this as an assumption. Are we building this kind of knowledge in social science, currently?
Does "more research needed" lead to more research being done, typically in a field? If not, why? Why are we moving on to a new question, when the last one is still unanswered.
Take a "willpower is a limited resource" hypothesis (replication issues aside). It does not seem impossible to answer this question. Nothing in science is "knowledge" in an absolute sense, but we can get to some level of confidence.
So.. do we want to know the answer to this question?
If not, why do the the first expirement? If yes, then what would it take to get to an answer? We can answer that partially in advance. Independant replication will be part of the answer. Sufficient sample size also. If we know in advance that these things won't be done, what is the point?
I'm an outsider and I realize I may be throwing unwarranted dirt, so if/where I am, apologies. I think that in some fields we systematically do not do the things necessary to get to useful knowledge, a sufficient accumulation of evidence to treat a result as more than anecdote.
Personally, my hunch is that the root cause for most of them is a somewhat misguided focus on "excellence" combined with it being practically impossible to measure researcher performance with a reasonable amount of effort. Then again, the problems are so intertwined and have so many different actors, that finding the root cause is really hard, so who knows whether I'm correct.
But, we can have systems that promote performance without measuring it. I think those are the kinds of ideas we should look at.
Publishing... This is obviously a centrepiece institution. It's central to replication/quality issues, to publish-or-perish career incentives and to funding allocation. So, I guess any radical ideas here will likely impact on a lot.
Can we modernize publication, make it more open, and update the peer review system? One step is to move away from seeing "peer reviewed" as a binary. Another might be increasing the role of peer review. Maybe papers should have critical "reviewer notes."
Maybe publications could take more of a 'director' role, influencing follow up, replications, being critical of practices...
..If a paper concludes that "more research is needed," does this mean anything? Will the research be done? What exactly needs to be done (eg, same study @ 5X the sample size). Is this likely to yield a result? Is that result important?
A big part of the replication problem (from my perspective, safely on my couch) seems to be that very little replication is done, because the incentives are all wrong.
If institutions were different (eg capitalism, for example) I could imagine the problem being reversed: replication and further work could be overinvested in, impoverishing exploratory science instead of the other way around. Replication is a lot more concrete (and money men love concrete).. you know what you need to do specifically, and what you might achieve by doing it. The problem in academia is that original work, even unreplicated an inconclusive is valued higher than subsequent work. We have to incentivize replication. In social sciences, we're sitting on a mountain of inconclusive results. Unless this work is done, we can't build on.
Funding... Measuring researcher output well enough to base funding on is not possible, imo. But, maybe we'd have better luck with portfolios. Ie, structure funding bodies as competing portfolio managers, looking to impress based on portfolio performance. This might give us a better mix of predictable returns (eg replicate these 175 studies) and long tail risky stuff.
I very much agree that it would be great if "peer reviewed" was not seen as a binary - because the research it represents is not binary. However, basing funding decisions on a lot of portfolio's full of non-binary research results is still, I'm afraid, too much work for e.g. tenure committees.
(The annoying thing here is that just because a system like what you describe is still far from perfect, people will stick with the system that's in place even though that's far worse...)
> I think people are hesitant to over-criticise academia.
is a message to people here. Of course, there's all the "anti-global warming, intelligent design flat earther republicans" who have always criticized academia, but that's not what you're talking about (I think).
I think you mean that people here - i.e. relatively educated, orthodox liberals for the most part - should really start to look hard at academia.
I'm someone with a traditional academic pedigree (double major in Econ + Math from MIT, most of my friends have PhDs) but I, too, am growing increasingly wary of much published research.
For example, there's this eye-opening blog post[0] by Andrew Gelman, a professor of statistics at Columbia who collaborates on a lot of psychology research, about the problem of bad statistics in research in psychology. There's this [1] famous paper by the Fed about problems of replication in what I studied, economics. Or how the NIMH "is re-orienting its research away from DSM categories".[2] And then there's the endless pop-science ideas that show up and get shot down like Power Pose, Air Rage, or Learning Styles. Or the constantly changing recommendations on nutrition.
I don't really take for granted anymore that academia is generally working towards greater understanding of the world. It's an imperfect system that has done some incredible things, but it definitely is dealing with some serious issues at the moment. I hope that initiatives like journals for null results, changes in the way grants are apportioned (like in this posted article), etc, can put it back on track.
[0] http://andrewgelman.com/2016/09/21/what-has-happened-down-he...
[1] https://www.federalreserve.gov/econresdata/feds/2015/files/2...
[2] https://www.nimh.nih.gov/about/directors/thomas-insel/blog/2...
The problem, it seems to me, is similar to VC: The funding agencies can spread their risk across a pool of ideas, but the investigators generally have to commit to one or a small number of projects for some time, so they risk spending several years without much to show for it if the idea fails.
More seriously, the problem with approaches like this is the people who are chosen would be mad to actually work on the idea they put up. At most the funding is for 18 month and what happens if at the end of that time you have nothing?
If you want this idea to work the funding has to be for at least 10 years so that people can afford to take real risk and have enough time to build a track record if the idea doesn't pan out.
If you have 4 or 5 years you really only have 2 years before you have to switch to low risk if things aren't working out to give yourself time to get enough papers out before you need to apply for the next grant.
What I did when I was an academic was spend 70% of my and my students/postdocs time on low risk activities and 30% on high risk. This keeps the risk down and still allow for some dreams.
- I debated about the value of ideas in HN. They took me as a fool. "Ideas are worthless, Execution is king".
Now, I have a brain used to find good ideas quickly. My fuzzy process of analizing something complex is well advanced. And no venture capitalist, coder or entrepreneur will be able to compete with me cause I have years of advantage.
I've also trained emotional resilience and self-confidence cause I knew I was right even when the successful people didn't agree.
I don't even need to read the article to comment on this. Ideas are my area of expertise. I spent 7 years without working and letting my mind free, thinking about thinking.
Hacker News shepple: venture capitalists and entrepreneur sheeple: FUCK OFF!!!
Anyway, I know this sheeple will find another way to attract money for projects, cause they say just what others repeat and what people wants to hear. No what people need to hear.
Nature or another scientific magazine will publish an article in 2 years time about the power and value of being frank, blunt and honest. And how it helps the mind. Meanwhile, I will be taken again as a fool again.
Really: FUCK OFF!!!!!!!!! I will do my fucking projects without any external money.
'They called me mad! MAD! Well I'll show them! I'll SHOW THEM ALL!'
On the novelty front, reviewers are trained to focus on finding flaws or absence of detail or preliminary data. The presence of any of these requires that we score down - so it's just about impossible for something truly new to get funded. Until we do the project, we don't necessarily know what problems we'll encounter or what the data will look like. The small number of NIH grants aimed at funding high-risk work by earlier-career folks (eg, the DP2 mechanism) are a great development but a tiny fraction of the portfolio. The key thing to recognize is that any move to distribute funding towards younger folks or higher-risk work causes the old guard to scream bloody murder - "what do you mean I can't have 5 R01's at a time?". For study sections, I've observed first hand that old habits favoring established investigators and low-risk, incremental work are hard to break.
If you plot the number of NIH grants per investigator, there are a small but significant number of investigators out in the right tail with 4+ R01's. In general grants beget grants, such that bigger and better-funded labs are better positioned to get the next grant. Details re R01 mechanism are here [1].
A recent proposal that would cap the number of grants an investigator could lead at one time was shot down by the old guard [2]. NIH continues to struggle for alternatives, because unless the overall budget grows, which it will not in the foreseeable future, any strategy will involve taking from the rich(er) and giving to the poor(er).
[1] https://grants.nih.gov/grants/funding/r01.htm [2] http://www.sciencemag.org/news/2017/06/updated-nih-abandons-...
https://projectreporter.nih.gov/reporter.cfm
I left for Google, and got more done in 20% time (research-wise) than I ever did using 100% time in academia.
When I did sit on study sections, I worked hard to help NIH understsand why the older groups asking for closet clusters on a cloud grant weren't helping. But NIH is very slow to move- it took at least 7 years to get to the point where people could even apply for cloud credits instead of closet clusters.
On the other hand, due to limited resources, I get the feeling that most academics are operating in the risk-averse mode. See, for eg [2] -- it seems to be solid advice for succeeding in academia (I'd like to hear counterpoints to that from people who've had an extended/successful academic career), but it seems wholly focused on small predictable improvements.
I wonder whether this problem is a consequence of enforcing "accountability" in a regime dominated by fat tails.
Maybe we've picked up all the low-hanging fruit in research and we can only hope for slow incremental progress (which is what the management structure is set up to optimize), and we need to come to terms with that and stop wistful discussions about not having black swan successes.
[1]: http://paulgraham.com/swan.html
[2]: http://www.stochasticlifestyle.com/tips-google-summer-code-a...
In my own circumstance and what I know from my colleagues, CS researchers rarely write their best ideas in grants---not because they are afraid the ideas are too bold---but rather because those ideas are often not fully worked out, and nobody wants to just give those away to a review panel full of top-rank scientists who might make connections faster than you!
The problem with "blind review" is that project proposals can rarely be anonymous because just explaining the work and citing the relevant prior work leaks a lot about the possible author. So making review blind can often be an advantage to the higher profile researcher and a disadvantage to the capable but slightly less well known one.
That said, as my own experiment, during my next NSF review, I am going to tear away the first pages of all the proposals, and make my first pass without knowing the authors to see if it makes a difference.
This should be better said as, "Fund fresh insight, to find fresh insight." And while that sounds tautologically obvious, almost no one is doing it.
I think there's a fetishization of complete blind objectivity in so many fields right now, which has at the root of it trying to iron out inequality. A laudable goal, but you shouldn't let that ideal completely upend the practical question to yourself of, where am I going to place my bets?
An unavoidable fact is that ideas alone are not enough to make some technology or venture successful. Sure, some one-off inventor may have a brilliant idea, but the idea is just like 1% of the problem.
If your model is to fund a team to develop an idea, the track record / personalities of the team are quite important.
Most people in this world are barely able to get their own lives under control. What's the likelihood that a random person with a great idea can take that to something commercially viable? There's a reason that who it is matters, and unfortunately, that still leads to people being selected who reflect the starting set of less-than-diverse people.
This article isn't saying that anyone should be able to get an NIH grant, but rather that grants should be awarded less due to politics and more based upon potential merit and the promise of an idea. I.e., This editorial is being published in Nature, not viXra.
The central issue is not one of team dynamics, or rooting out inequality on a societal level, but rather about rooting out inequality among equally capable candidates. Currently, many academic fields are dominated by cronyism and that is what the article is advocating against.
I'd agree with arguments that this is true more because of network effects than whether others can do the job or not.
If you're only looking to repeat success, skilled folks are the best bet.
But it leaves others who want to learn behind. And a lot of the folks with skills got there because someone propped them up.
I'm just totally done with the idea our economy needs to be this fucking "always grow, always invent" scheme. So much of it is pointless shit that doesn't add to human discovery, rather it serves to change values in a bank account.
I'd rather we focus on getting the basics right; healthcare, food, shelter, for the masses. Make that our daily focus, and give people time otherwise to find their talent or cool idea.
But we're too addicted to universities telling us economics are like laws of physics and thus we must abide and live by them. Which is not much different than churches telling us the Bible is like laws of physics and the only model for living and we must abide.
Smart folks get lost in their discoveries and pretend they apply to all of us. And manipulative folks will take that and make it the law of the land for their benefit.
innovation produces billions of dollars of wealth . Invention produces trillions. Are we sure we have our priorities straight as a species?