Alan Woodward, security professor at the University of Surrey, said it was a "positive move" that he hoped would help "set the tone" across academia that research was "about discovery, not making money".
That's not going to go down well with our lords and masters.
All very nice, but...
when you are looking at a study of, say, time course studies from 2000 zebrafish brains, scanned at sub-cellular levels on a high speed, multi/bi-photon matrix lightsheet microscope, generating around 200Tb of data alone, then ongoing storage costs are a real issue. How long should these data be made available? 5 years? 10 years? Forever? OK, it's nice a cheap to stick them on tape or freeze them in AWS glacier, but even then you are looking at a good few thousand pounds a year, and it's hardly freely accessible.
And putting it all on ice gets even more appropriate when you consider, say, keeping a set of plasmids in liquid nitrogen for the next 20 years. Space in the -80°C freezer, anyone? And you have to keep track of the samples as well. Barcodes and QR tags on everything, then there's the company that's going to rake it all in with a laboratory management solution - unless there's a reliable and flexible open source solution for that too. Has to be cross compatible with other institutions of course, because researchers are a mobile lot, and when they go, they tend to take their samples with them.
All told, this, although it's been on the horizon for a while now, is going to provide some real challenges.
Welcome move, code-wise
(I recognise the difficulty in many other areas)
I am often hampered by not having access to code written by other scientists. If I want to test my shiny new algorithm against state-of-the-art efforts from others, it is a right pain to implement their methods from scratch. Besides, even if I write a working implementation (or get a student to do it), how can I be sure our implementation of the algorithm is the same as the original. I remember a Spanish colleague being very surprised at the timings we reported of his algorithm, because my student's implementation was a lot faster than his own, on a very similar machine. I also frequently just want to use a particular method in some image preprocessing step, and having some state-of-the-art method to work with directly saves a lot of time. People are getting much better in sharing academic code, but seeing a more concerted effort to retain access to code is useful.
Re: Welcome move, code-wise
Agreed. We had code for a breadboard light sheet microscope, and I was called in to have a look at a dropped frames issue. I got all the stepper motor pulses on a dual beam scope, as well as the camera and shutter trigger pulses,it all looked good, but at some seemingly random point in time they went out of synch, skipping a frame and re-synching part way through the cycle. All the instructions and scripts were supplied by a collaborating laboratory. The set up of the script for the D to A convertor was fine, but they did have a bit of code they'd written themselves for the image acquisition, which they hoped to commercialise, so they only provided the executable. Turned out that it was that code that was causing a resource blip all down to a silly little timeout waiting for a response from piece of hardware that was superfluous in our rig and so wasn't on there. Took months and months and months to sort that one out by a process of trial and error, mainly because although that bit of code was the only one left where there might be a problem, they refused to acknowledge that there was anything wrong with it and in their opinion not having this bit of hardware wouldn't be an issue. We had the D to A system shipped back to the States for a full diagnostic, which took up a lot of the time. Turned out that WAS faulty as well, but not in a way that should have caused that fault. The frame dropping was still evident even with a replacement known, checked and recalibrated D to A.
Re: Welcome move, code-wise
I am often hampered by not having access to code written by other scientists. If I want to test my shiny new algorithm against state-of-the-art efforts from others, it is a right pain to implement their methods from scratch.
I am generally in favour of sharing code, and try to make my code available as widely as possible. Doing so (hopefully) optimizes the overall outcome for the field.
However, there is a flip side to this coin, closely connected to the tragedy of the commons.
The currency of the academic-career world is recognition, which comes in two main flavours: publications and citations. By making (a good and useful) code freely available, I generally decrease the number of publications I will get: otherwise, researchers needing the code will have to collaborate and co-publish with me. It gets worse: When I do a good job writing and documenting the code, so that it can be installed, used, and modified without my help, I get fever collaborations still: why invite me, when the code works? Sometimes, it gets worse again: scientific codes are rarely written as a one-off: you typically expect to amortize the cost of writing the code over several related publications. But, if you make the code freely available, another group might scoop you by studying a system you had in mind and publishing the results before you do - so you end up doing the heavy lifting without accruing the benefit. Note that I am not talking in hypothetical: I've had all three scenarios playing out with my own codes.
In theory, I am supposed to be compensated by the increase in the number of citations to my work. In practice, the outcomes vary: some research groups will cite you every time they use your code. Others will cite you the first time they publish the results obtained with your code, then just cite their own first publication instead - even though they are still benefiting from your work. And then there are those who'd never cite you at all - even though it is clear that they must have used your code to get that they report.
For me personally, this is mostly a source of a minor irritation and sometimes of a major amusement at conferences: I have a permanent position which matches what I want exactly, and have no plans or desire to move elsewhere. I can therefore afford to be magnanimous about it. Unfortunately, my situation is more of an exception - and getting no publication/citation payback on a multi-year investment of developing a code can easily ruin the career of a junior researcher.
Re: Welcome move, code-wise
I think that AC has a very valid point here. Although I'm overjoyed that the Wellcome Trust takes on this challenging pioneer role, there are some fundamental conditions and practices that might hamper reaching their observed goal. Working in the pharma/ medical device space, I've to put forward that the healthcare system we have currently is driven by commercial incentive. This on itself is nothing new; for years the discipline of Translational Medicine has focussed on this. And, despite firm academic, regulatory, and financial backing, not reached a (more) open environment. As AC mentions, the drivers and metrics of developement remain for example publications and IP ownership. Which translates to money. And this is where industry comes in. And, be honest, excel. And no matter how utopian your vision for the future is, one has to recognise that reaching a result that really helps people/ patients, the commercial construct we've got now is indispensable. First, academia isn't good at developing, registering, large scale manufacturing, and distributing products. Second, parties involved now in the industry now will not give up their advantageous position. Examples of this run from scientific publications pay-walled by publishers with financial interests, to industry dropping development in some therapeutic areas because return doesn't justify initial investment. Or healthcare payers steering reimbursement driven by financial, and not therapeutic considerations. Third, thinking that all development comes from academia is a misconception. Frequently academic research is done for the sake of doing research, or learning and understanding something. However, in for example medicine, this does not (always) translate in something that helps patients. Most of those development are done by industry themselves. Paid by industry. On their conditions. And thus, granting organisations have no leverage in these development processes.
nevertheless, despite being an old and grumpy person, I think the Wellcome Trust needs to be commended, and hope that many grant organisations realise that they need to set similar/ more conditions like this before they hand over the money. After all, you can only peel the label of a beer bottle by starting at the corner...
"Matthew Woollard, director of the UK Data Archive and the UK Data Service, "
Who has ever heard of this guy, and this organization?