Friday, November 13, 2009

F=ma [Made in China]

We have a confession to make, gentle reader. We're guilty of hubris. We used to think we were no lightweights at this business of re$earch trickery. Why, on a bright day we might even have ventured to put on our Sherlock Holmes hat, and just like Master Detective does with mud-stains in a 50 mile radius of London, we might have dared to claim we can instantly identify which murky alleys of fraud a published result has passed through. But the Grand Reviewer up there probably didn't like our active voice, and sent this whopper that has made our assessment even dimensionally incorrect. The impact of the realization is so large that we are even contemplating sanyasa from the exciting, high-stress life of research and settling down in the serene, peaceful life of a pit trader or an investment banking MnA analyst instead. But perhaps we should begin at the beginning.

A labmate casually remarked today that a very well known professor in a well known university has all his grad students in the US, and all his postdocs in Singapore.

Also, the lab was getting its experiments on mouse cells done in China.

BY GOD, and the entire editorial panel of Nature crossbred with the Economist's! It struck me like the loud report of the starter pistol did to Yeddy when his political race went Ready-Steady-Reddy: Outsourcing research! That must be the most ABSOLUTELY FANTASTIC IDEA EVER! How could we be so blind to it for so long? For shame, for shame! Just take a look at the advantages, which I outsourced straight from an article on the benefits of outsourcing:

Advantage #1: Outsourcing can save you money.

Hell yeah! Customer satisfaction is always our #1 priority. The huge number of  research studies which end up with unremarkable conclusions is a tremendous waste of grant money. If for 1/10th the price the people over at Wipe-pro can provide value-added solutions to the Poincare conjecture, with relentless effort towards quality and project delivery excellence (PDE), leveraging their time-tested experience (TTE) of project management, industry best practices (IBP) and internally developed project management software (PMS), who are we to meddle with the free market? High quality results and 100% customer satisfaction guaranteed. All platforms, including NSF, NIH, DoE, ARC are fully supported. 150% moneyback policy if Tier-2 journal rejects work, Conditions Apply. Why needlessly spend money on costly grad students, supercomputers and lab facilities? All a grad student would do with his stipend would be live, eat and watch porn on the lab's projector or  Apple HD Cinema Display anyway.

Advantage #2: Outsourcing can help you share risk.

No shit Sherlock! Who wants to run a 3 month experiment only to find the correlation is only 0.4? Plus, universities have developed stiff-nosed Victorian standards of conducting research and it's impossible to be creative under such terrible constraints. Procedure, they harp, Procedure and Hypotheses and Validation and Double Checking and Consistency and Backtesting and yadayadayada. You don't want all that, you want a good life. Take a hint from the investment banker down the road with the Ferrari, and share your risk. Maybe if you imitate him well, you can hope for a Camry. You don't need to do Science the way old European fogeys did it. Peering at the skies for 30 years without a telescope to record observations accurate to 1/60th of a degree, and dying of a bladder burst, pooh! It's 2009 now. Risk is taboo. The seller will be eager to please, and you have the pleasure of perusing results with none of the pangs of Procedure.

Advantage #3: Outsourcing can help accommodate peak loads.

Of course! It was micro and stem cells a few years ago, it's bio, nano, energy and cleantech today, and it will be pervasive soc-nets and Cloud computation tomorrow. You never know when a funding source will chance by one day, so it's always best to be prepared with research power on the bench. GPU computations for resonant MEMS cantilevers? Our Korean team will handle it. Solid Oxide Fuel Cells for Clean energy for 2-stroke engines? Our brand-new benched team, which can't tell apart electron emissions from nocturnal emissions, is right on it, and wants to know what in colour you want the prototype painted. Long-term environmental impact of sulphur effluents on river systems? No problem, our India team has specialized on-ground expertise, and can deliver historical data, analysis and predictions all in a low-cost, 5 papers guaranteed package. Hurry, order before Thanksgiving and get a conference-friendly Java GUI FREE!

Advantage #4: Outsourcing can help develop your internal staff.

Very deep. First, you won't have wasteful grad students asking poking questions loitering around. Neither will you have pesky postdocs. Only you, tenure and a growing stack of publications. Bliss. You can finally take that long-deserved vacation without the headache of that algorithm not converging, or that shady simulation not coinciding with experimental results. It's Someone Else's problem now.

So what are you waiting for? Just pick up the phone and call 1-800-MOAR-PAP3RZ now!


(Update, Jan 6) Hark, unbelievers! "Chinese academia ghost-writing 'widespread'"


If our humble efforts have succeeded in piquing the interest of the gentle, generous reader, might we beg that he humour a fleeting thought toward a modest venture embarked upon by the One Post Phenomenon and your humble author, Soopper Turbo Suttifiers GmBH? We offer superior quality German acronyms beside our name, and ensure such high standards of service that you shall always leave our premises with a light heart and a lighter purse.


Gratifyingly, the website which I outsourced the bullet points from, Sourcing Magazine, 'the world's leading online content provider for the Outsourcing community', has its homepage stuck in an endless redirection.



Mahesh Mahadevan said...

Recursive outsourcing in references - sweet! Maybe (on a blown up scale) that is what happened to the magazine's website.
Maybe if you imitate him well, you can hope for a Camry. Holy Godmother of Inside jokes!
On a more serious note, it will invariably be passed as "collaboration" and "international relations" and ignored until Somebody Else's problem turns into yours.

displayname said...

The Onion had a video a while ago: More American Workers Outsourcing Own Jobs Overseas.

I don't see the problem: I've heard that many professors treat their graduate students as the equivalent of outsourced labour already. ;-)
It's also true that many of the things we/scientists do is technician-work (doesn't mean it's not useful) and can rightfully be automated or outsourced where possible. As Andre Weil is reported to have said after he had got a great expert in the field of elliptic operators to solve a problem that cropped up in his research — "when I have a problem with electricity I call an electrician, when I have a problem with ellipticity I use an elliptician".

Vinayak said...

Looking at the above comment, I recall the discussion on how that Grad school is a proving ground for international students. In that case, I guess its true that overseas grad students are just another form of outsourcing ( get more hardworking math guys from asia and put them to work for a cheap pay which they consider handsome salary).

I too read the article about a famous lab having its research on mice done in china - looking at the bigger picture I can see that both the sides are happy.. not taking into account the social and environmental effects such things may have..

So does that mean outsourcing is good for either parties ?? or is it bad ?? I guess there is not right answer.. there are too many parameters to be considered

wanderlust said...

why even bother OUTSOURCING when you can crowdsource? you can just post the job you want done on and someone or the other will do it for a few cents. no minimum wage, no workers' rights, very one-sided system.... and you'll have atleast ten people doing the same job. research proves that four-five mturkers equals one expert.
better, no?

Mohan K.V said...

@Thalai: Thanks!

The recursive outsourcing is a bit of a below-the-belt hit, I confess. It's happening because of a misconfigured Apache server, which dies when it doesn't see a 'www.'. Quite the same happens with the IITM homepage. But it's one of those happy coincidences that it happened to be an infinite redirection, instead of a CNAME drop like IITM :)

@displayname: Thanks for the links, all three were delightful. I had a faint liking of ONN before, now I'm positively addicted.

Labour extraction - That is gospel truth. When I say that, the force of my conviction (derived from the momentum of first, second and higher hand experiences) is so strong that if anyone claims otherwise, I am confident I can 'No' in a more suave fashion than the most interesting man in the world.

My beef with monetizing research is that there's so little in terms of measurable, verifiable progress. The examples you quoted are perfectly valid, but are special because any claimed results can be very easily verified. Not quite the same with, say, an experiment to measure the strength of a material, or a numerical model for turbulence. It's much easier for cargo cultism, reinforcement and plain fraud to happen when work is split up across people, continents, time zones and cultures.

There are hand-countably few people who comment on my discourses, and I could gladly employ a Pirahã accountant to keep track of people who include hypertext. I'm guessing you're a new avatar of an old chest-mark, and if you're not, welcome stranger!

@Vinayak - Absolutely, Prof. H told that right to my face. For a more candid take, read Jon Katz's 'Don't Become a Scientist'. It was written a decade ago, but the numbers are still the same.

And of course both sides are happy, no hypothesis that people would want to be true will ever be invalidated! On a very broad scale, yes, you're right, there is some advantage in getting someone else to do the donkey work.

@wanderlust: A most admirable proposition. (Out+crowd)-sourcing has been proven to work, and combos are especially interesting. But alas, extracting work from everyone would have our already-distressed venture scraping the bottom(s). But we're old hands who know Talent when they see it, and STS GmBH hereby bestows upon you the title of Grand Honorary HR Consultant!

Anonymous said...

Maybe it's just a bias because I'm in the theorem-proving side of things (or maybe it's the other way around), but I don't think most "scientific" research should be trusted unless it can be easily verified anyway. Except for some facts that have been verified repeatedly (either because they were surprising, or because of fierce opposition, as in the case of evolution and global warming), most one-time experimental results are at risk of errors and should be treated with scepticism. For example, most behavioural studies are hard to trust because they often ignore conditional probability and usually involve a very biased sample from WEIRD societies — usually undergraduates, and the more-reported research is often from even more selective (less representative) universities. :-)

And for research that can be verified — like tweaking various parameters until you find one that can be verified to fit :p — it's fine to just outsource it (and mention who did the work...).

[Yeah, the displayname above was me: that happens when I forget to sign in with OpenID and let Blogger pick the default (Google Account). I had chosen "displayname" as the Display Name a long time ago when I was afraid of turning up in search results, but now I've stopped caring. Anyway, as here, everyone always knows when I'm commenting, so it's doesn't matter much either way. :-)]