Outsourcing research

Every now and then, I take a moment to read my unsolicited commercial email before binning it. (Note to eMarketers: This moment is generally used to mock and deride the goods and/or services offered in the unsolicited commercial email. Take me off your stupid mailing lists!) The other day, I came upon a message offering me a service that’s a new one to me: outsourcing research.
As a professional philosopher, I don’t have much call to outsource my research (which mostly consists of reading, thinking hard, pounding away at the keyboard, and swearing if MS Word crashes in the middle of a crucial sentence). But if I were trying to run a chemistry lab, the idea of outsourcing research might be very appealing.
And, it may surprise you to learn, I think there might even be positive effects for our body of scientific knowledge from trying something like this.

First, let me give you the crucial bits of the email, which was sent to me by a company called Amazelia:

Research Processes vs Business Processes – Can they also be also outsourced?
Traditionally Process Outsourcing has been limited to Business Processes. The compelling reasons being – the workflows follow a pattern and volumes are high. Process training can be imparted with relative ease and the investments can be amortized over comparatively shorter periods.
In contrast an academic pursuit like research is considered more of a creative, ‘one of a kind’ process with much lower volumes and thus not amenable to outsourcing. It is our submission that though research is a highly creative and individualistic process, it can be still be broken down into smaller process steps.
We believe that the highlighted process steps can be very effectively outsourced, freeing the in-house research resources for the more critical process steps of definition, exploration, identification and presentation.
We have significant expertise and experience in enriching data from any available form to the form required by YOU. Moreover, it provides solutions for delivering the data using different mediums including web and print. Amazelia also has expertise in developing workflow solutions that automate and ease flow of data as it goes through different processes

• Data Capture & Enrichment Services
• Repository Creation Services
• Data Delivery Services
• Editorial Services

Probably the first thing to note here is that the services Amazelia is offering here seem to revolve around data analysis and management. They are not, as far as I can tell, offering to set up labs to perform a gazillion runs of your experiments.
But what if they were offering experiment-outsourcing?
Talk to scientists whose labs conduct lots of research. A lot of them will agree that the interesting part of doing research is “definition, exploration, identification and presentation” — that is to say, dreaming up the experimental approach to answering the scientific question, working out a way to implement that approach, looking at the results to see what can be deduced from them, then putting those results into the context of the rest of what is known and discussing it with other scientists. The actual “fun” of getting the experiments to work, repeatedly, until a sufficient pile of data has been amassed, is usually saved for grad students and lab technicians.
Given that the principle investigator (aka “the brains of the operation”) is basically passing much of the labor of experimentation to someone else (students and techs) already, is there any reason not to “outsource” the work more completely — say, to a lab oversees?
Well, sure. This could have a fairly bad impact on the throngs of local science graduate students (and graduate student wannabes) — if their labor in the laboratory can be replaced with cheaper laboratory labor oversees, their research stipends go away and they get to spend more time teaching pre-meds (talk about inhumane!). Beyond the havoc massive experimental outsourcing might wreak on the lives and economic prospects of graduate students in the U.S., it could also, in the name of cost-cutting, rob them of valuable laboratory experience that is actually important in developing their skills at conceiving and designing experiments that can be implemented. (The “brains of the operation” needs to have a sense of what is do-able.)
There are also, of course, issues of laboratory safety to consider when offshoring experiments. It may be attractive to conduct your experiments in a place without a hard-ass IRB putting all kinds of restrictions on you. But throwing caution to the wind on toxic waste disposal, animal use, and keeping electrical cords out of puddles on the floor, even if it’s completely legal, might have bad consequences, too.
And, offshoring research gets rid of one of the PIs most useful tools: the ability to pop into the lab and see firsthand how things are going. This is not just a way to make sure the experiments are being run, but it’s also an opportunity for quality control (are notebooks thorough, strip-charts labeled, equipment and reagents in good shape, etc.).
So, you’d have to be insane to think outsourcing experimental research has anything going for it, right?
Maybe not. After all, one of the things scientists strive for are reproducible experiments. You want to get your data from an experiment that other scientists of reasonable skill can successfully conduct, too. If only you (or your lab) can get a given experiment to work, there is reason to worry about that experiment.
Moreover, when reporting your results, you’re supposed to describe the experiment that you used to get those results in sufficient detail that another scientist working from this description could set up the experiment and get it to work. Sadly, “materials and methods” sections in journal articles rarely deliver clear or complete enough descriptions that you can reproduce the experiments without a struggle.
And of course, there’s the problem of seeing the results you want to see. If you’re very invested in an experiment working (because you’ve been slogging away at it so long), there may be a tendency to see the data you produce in the best possible light. The bad data we got yesterday must be the result of an equipment malfunction or a bad reagent — we’ll just leave it out.
What if, after you dream up an experiment and do a preliminary run or two — to be sure that the experiment isn’t completely implausible — you wrote up a detailed protocol and sent it off to a neutral lab (perhaps in another state)? What if you sent it to two? You could find out:

• Whether two labs, working independently of each other, come up with the same results.
• Whether your experimental protocol was sufficiently detailed and unambiguous that the two labs were doing the same experiment (and if they weren’t, what piece of the protocol led to the difference in execution.
• What the data are from the point of view of researchers who don’t have an expectation of what the outcome is “supposed to be”.

To the extent that you’re interested in building objective knowledge about the system under study, and in presenting that knowledge clearly to the scientific community, indulging in a little “lab-swapping” might be a good thing.
There would be a lot of details to work out, of course. You wouldn’t want to send out your experiments to the lab of your closest competitor, since that lab could probably guess from the protocol just what you’re hoping to find. This would leave you vulnerable to being scooped (if the competitor’s lab rushed some protocols into existence to try to get to the results before yours) or sabotage (“Oops, the instrument broke again!“) But maybe there could be a system, parallel to that in place for refereeing manuscripts, by which you could identify the sorts of techniques in which the labs doing the experiments would need to be proficient and the labs which should not be asked to do the experiments because of potential conflicts of interest.
“Outsourcing” experiments locally would be a way to make sure everyone’s graduate students cut their teeth on real experimental work and pick up valuable laboratory techniques. It would, as it happens, separate the experimentation from experimental design and, later, data analysis. (Presumably, if you’re taking in other people’s experiments, you’re also sending your own experiments out to other labs.) I suspect, however, that the de facto division of labor in some labs is such that grad students are most focused on getting the experiments to work and get rather less conscious mentoring in the steps before or after running the experiments. Perhaps PIs would be more interested with chatting with students about experimental design and data analysis if these students are sweating over someone else’s experiments.
From the point of view of building a truly international community of science, international swapping of experiments could be a good thing, too. The key, though, would be in ensuring a reciprocal relationship rather than a unidirectional flow of experiments to the labs where labor and materials are cheapest. The point here is not to save money so much as to produce more objective and reliable scientific knowledge.
How crazy an idea is this, and is there any reason principled scientists shouldn’t look into trying something like this, at least on a small scale?