01 July, 2012

Keynote about Open Science

At Akademy, about an hour ago, the keynote by Will Schroeder from Kittware was finished. It was a very nice talk - and I've collected some notes, see below!

What is this thing called science?

After introducing Kitware and what they do ('all things scientific computing related'), Will starts to talk about science: where does it come from?

You might remember this Descartes dude. He questioned everything - and that is where it started. Nullius in Verba, "take nobody's word for it", that was the thought behind this movement.

And realize that this did not go down easily! People where locked up for this, faced jail time for their convictions. They were the hackers of their time, trying new things, finding new ways. And sharing knowledge.

Because that is what science was (and should be) all about. The way it worked was as follows:
A scientist wrote a paper, a letter. This would go to the Royal Society or another 'science institution'. There the experiments were replicated and verified. Once verified, the letter, paper or book was replicated and distributed through society.

Things changed since then

But commerce took over and now, the process goes from scientist to commercial publisher where volunteers do peer review and then the article gets published in a journal.

This looks like the same process, but it is not. First of all - in reality, replication of experiments does not happen. There is a number of reasons for that, some practical (huge computational requirements, growth time of tissue samples) but often it is also lacking data, details on how the experiment works or closed, unavailable software or procedures.

The thirst for {fame\power|control|money} has tainted science: we've lost the search for truth. It is "publish or persih", career pressure is huge and scientists are afraid to share knowledge because it might loose them a paper or even patents and licensing income.

Meanwhile, according to a case study, licensing revenue on patents is about 2 billion, but if you substract the costs the university breaks even. And the push for patents is corrupting and damaging science and creates resistance to collaboration.

The results

And it shows: Nature published a study showing that more than 90% of papers in science journals describing 'landmark' breakthroughs in preclinical cancer research are NOT reproducible and are thus just plain wrong.

Will gives a computational science/medical imaging example. It is quite complicated - but boils down to the fact that we can't reproduce the result because we lack knowledge of how it was obtained.

So there is a huge pressure on scientists to do bad science and nobody checks up on the results. Meanwhile, journals take easily 2 years and hundreds of euro's to publish their articles and you also have to pay thousands to read the results - which were peer reviewed by volunteers!

Our data is unavailable or put in proprietary formats, publishers control the flow of information and closed and proprietary software is used to do analysis and controls how scientists work.

What we need

What we need is open science: open access to knowledge, open access to data and open access to source.

It is a real tragedy that we have to put the word 'open' in front of science!

But we have to. Science, part of this three hundred years old tradition of hacking and sharing knowledge, has been corrupted and locked up.

The good news is that things are changing. Universities realize that the status quo does not benefit society and change their policies. Harvard now asks professors to publish in open access journals and the UK is going to only fund research which ends up in the open.

Of course, we already knew that: both society and business show a clear trend. Open is better and will take over closed!

note that these are personal notes and not reviewed: no guarantees about the correctness!


  1. I see this a lot in bioinformatics: unmaintained software, or worse, stuff that requires expensive licenses (SAS, Matlab, Oracle) just to run.

    1. Yup. I was real glad that this topic was raised at Akademy...

  2. This is the path competition seems to lead, first the increase of efforts followed by shortcuts and lost of focus.

  3. A PhD student here. Agree with everything in your blog. Reproducibility of results has gone down considerably. Non-availability of equipment is one thing (eg: not everyone has access to LHC :P). But for computational experiments, non-availability of source code for many papers and lack of info about the system/software config is not excusable.

  4. I have investigated Open Access as well and I totally agree. It is ridiculous that scientific research, mostly entirely funded with taxpayer's money ends up in journals which require us to pay their publishers a second time to access the knowledge.

    As you point out, mandatory Open Access policies see more and more adoption. The share of Open Access publishing is rising [1]:

    "Since the year 2000, the average annual growth rate has been 18% for the number of journals and 30% for the number of articles. This can be contrasted to the reported 3,5% yearly volume increase in journal publishing in general. In 2009 the share of articles in OA journals, of all peer reviewed journal articles, reached 7,7%. Overall, the results document a rapid growth in OA journal publishing over the last fifteen years."

    Yet I think more should and could be done to stimulate Open Access. Governments finance science and could impose tougher requirements for Open Access on universities and such. I wonder if Will Schroeder gave some thoughts on how to solve the problem?

    [1] http://www.plosone.org/article/info%3Adoi%2F10.1371%2Fjournal.pone.0020961

  5. Commerce is in all probability _the_ worst thing to have ever happened to science.


Say something smart and be polite please!