Saturday, May 25, 2013

What do you call a reliable Scientific Establishment?

Engineering:
Science, particularly academic science, is now a big business, and it is an unusually corrupt one that is primarily dependent upon the media and government funding.  It has no practical external limitations upon it holding its businessmen accountable. As Stapel's example demonstrates, there is absolutely nothing - nothing - reliable about it.  This point should be driven home every single time anyone makes the absurd claim that science is the best, or the only, arbiter of truth and reality.

Here is how bad the corruption is: Stapel was actually teaching a graduate seminar on research ethics. Notice too that all of the established academics who caught wind of the fraud not only looked the other way, but advised others to do so as well.
We have a word for real and genuine science that is reliable enough to be trustworthy.  Engineering.
You don't have to look to a Dutch Scientist who committed fraud on an industrial scale to see the rot.  There's the 90% of important scientific papers cannot be replicated by experiment.  And the scientists who published them cover it up:
In their Comment article 'Raise standards for preclinical cancer research', C. Glenn Begley and Lee Ellis (Nature 483, 531533; 2012) refer to scientists at Amgen who were able to reproduce findings in only 11% of 53 published papers. Several correspondents have asked for details of these studies, which were not provided in the article.

The Amgen scientists approached the papers' original authors to discuss findings and sometimes borrowed materials to repeat the experiments. In some cases, those authors required them to sign an agreement that they would not disclose their findings about specific papers. Begley and Ellis were therefore not free to identify the irreproducible papers — a fact that the Comment should have mentioned.
What do you call a scientific paper where experimental replication failed?  Falsified.

But the name of the game in the scientific community is publishing and getting grant funding.  If you publish something valid, than that's even better.  Perhaps I speak as an Engineer (Electrical, thanks for asking), but there's a clarity in engineering that no longer exists in science.  The clarity is that something works, or it doesn't.

I'm starting to think that a great place to start reducing the deficit is to slash the budget of the National Science Foundation.

7 comments:

SiGraybeard said...

This is extremely important stuff that gets virtually no attention. The fact that 90% of important medical studies can't be replicated is stunning and hints at epic corruption in the medical and pharmaceutical industries. Just like the tin-foil hat crowd charges.

At the least, it says modern medical science has enough replication and science behind as applying leeches.

Disclaimer: here's where it gets personal. My wife had a stem cell transplant, commonly (incorrectly) called a bone marrow transplant, for breast cancer because some scientist faked his results showing it helped survival rates. Later studies actually showed the first was faked and they have since stopped making women endure this treatment.

It appears you and I talked about this once before.

OMMAG said...

Excellent post!

I've been using this argument on scientific illiterates for decades when they start spouting off about articles they read in magazines or something they saw on TV.

It's not new and it is far to common that popular media uses junk as the basis of most published material.

Borepatch said...

Graybeard, it isn't the pharmaceutical industry, it's the research hospitals and Universities. It was one of the pharmaceuticals who exposed it, because they tried to replicate some very promising research and came up with essentially nothing.

Anonymous said...

Thankfully the rot is not nearly so bad yet in the hard sciences (in my personal experience, at least).

Vox and others are correct in pointing out that a main reasons for corruption is the intimate relationship between the University complex, "Big Science," and government funding. Another important factor, though, is the sheer size of the scientific establishment nowadays.

In the past, the top of most fields consisted of twenty or thirty individuals, with perhaps several thousand others filling in the gaps. Science was a craft, a vocation, and that outlook really shines through when you read primary documents from, say, the period during which quantum mechanics was developed. These people were deeply dedicated to truthseeking.

These days, science is a gigantic profession, and not only is the average intelligence of scientists therefore lower, but the culture is different. Things have become cliquish and frankly somewhat cultish; more about how many followers you can amass than about finding the truth.

This was driven home to me in undergrad when I attended a lecture by an eminent physicist; one of the chief proponents of superstring. Not only was he disrespectful of the audience, but his main reason why criticisms of superstring were wrong was "oh, they (the skeptics) are stupid!" delivered in a snarl. All of his justifications were social/shaming; none scientific.

They are, it increasingly seems, not stupid.

This experience helped me make up my mind not to go into research physics as a career.

AndyN said...

We have a word for real and genuine science that is reliable enough to be trustworthy. Engineering.

Even this isn't entirely correct. Engineering is only trustworthy as long as engineers are subject to unfettered peer criticism and unmanipulated market forces.

I'd be willing to bet the engineers at Solyndra were absolutely sure - or could at least offer a compelling argument, whether they believed it or not - that the proprietary design of their solar modules was based on sound engineering. Had they been forced to make that argument in a free market, it's likely nobody would have ever heard of Solyndra. Since engineers, like all other scientists, have to sell their ideas in a market with the heavy thumb of the government resting on one side of the scale, we're all footing the bill.

Granted, some scientists are less honest than others, and some scientific disciplines lend themselves more readily to fraud than others, but let's not pretend that there's no way for a crook to use engineering to justify his fraud.

Anonymous said...

AndyN,
Reliable doesn’t necessarily mean immediate. The difference between science and engineering is that engineering comes with a built-in day of reckoning for fraud. It may not happen right away ( Solyndra took six years to go from start-up to bankrupt ) but it will come. With science, there may never be any feedback that a fraud has been committed.

AndyN said...

reallyroscoe,

I read somewhere once that the second greatest feeling as a scientist is to discover or prove something that nobody else had ever been able to discover or prove. The greatest feeling as a scientist is to see a rival discover something nobody had ever discovered before and then to prove that they were wrong. As long as science is a human endeavor, there will always be feedback because someone will want to make a name for himself by knocking someone else off their pedestal.

Even if that wasn't the case, any science that has any impact on people will have to be exposed to scrutiny at some point. One of the largest and most powerful political entities in the history of the world has thrown its full weight behind the theory that cow farts have more of an impact on global temperature than that huge flaming ball of gas in the sky, and yet even the enthusiastic support of the UN can't prevent people from seeing that more cows are farting now than ever and yet the earth isn't warming. Someone publishes research that says 90% of published studies claiming that a proprietary mix of pixie dust and happy thoughts will fight cancer are hokum, and while it's nice that he's quantified the extent of the fraud, it was already the case that anybody who tried to treat cancer with any of those proprietary mixtures of pixie dust and happy thoughts would fail.