Friday, September 3, 2010

I recently got an e-mail from David Bradley asking my opinion of Web 2.0 as it relates to science, where it’s heading, and how we can get scientists more involved in web 2.0 / data sharing / and the semantic web. I thought this would be a great topic for me to write a real post on since I’ve been involved in this field and trying to promote the ideas of web 2.0 in the sciences for the last 5 years.

For starters, I really have no idea what it will take to get scientists to be fully engaged with the on-line world. It's hard enough to get them engaged in the real world (I wish that was a joke…). I think for most scientists to get involved with a network, we're going to have to develop something that significantly increases scientific productivity, and I'm not talking just a free reference management site or being able to post lab retreat pictures to a profile. The last 4-5 years have showed us that scientists really are not interested in FaceBooks for science. The marginal success of ResearchGate, NatureNetwork, and LabSpaces can't be cited as triumphs because very little of what goes on at any of those places is what I had in mind when I first developed my site. And the failure (whether they’re willing to admit it or not) of many of the other sites that had a business model of being a facebook for science highlights pretty clearly the demand for such sites. Now, the sites that have been successful are so because they found a niche. ResearchGate’s groups and Job listings appear to be relatively active, Nature Network has its blogs and forum, and LabSpaces has the news and Blogs. However, It seems like at least for LabSpaces, I've just catered to the scientists who were willing to interact on-line, and for the most part that's by the way of blogging about their science or the scientific experience. Now, there's nothing wrong with that, but it's a far cry from the collaborative tool that I think a lot of us FB4Sci developers were hoping for. So how can we change this? How can we get more scientists involved in these types of collaborative websites?

Barriers to Science Social Networks

First, I think it's a good idea to put into perspective what the barriers are to getting a critical mass of scientists doing science on-line, sharing ideas, and collaborating on projects with people they’ve never physically met. These barriers are formidable and I think, in many fields, you'll never be able to get past them.

1. The culture – The culture of scientific fields varies significantly. There are some fields that lend themselves well to collaboration while others do not. It seems like the most willing on-line collaborators are the computer scientists while the least collaborative are in the life sciences. This may have something to do with each group’s familiarity with computer use or how comfortable they feel sharing data on-line. Many life scientists want to bleed their data dry before offering it up on the secondary market (and I think A LOT of scientists feel this way). Another factor is likely the way that science has been done. Many of the habits of science evolved a hundred years ago. There will have to be a major change in the dogma of the scientific culture for science to become web based, open, and collaborative. Do I see this happening tomorrow, or in 10 years? No. I wouldn’t ever chance tenure on open notebook science. Not because I don’t think it’s a good idea, but because the people who make tenure decisions are so ingrained with “how to do science” that doing science differently could have very negative results.

2. Secrecy – The whole premise of a facebook for science is information sharing. These networks are all about data! The types of data that can be found on social networks varies greatly. The type of data and information I’d like to see shared on a science network is probably the least likely to ever appear! I have gotten the feeling that a scientist would give up his social security number and credit cards on-line before he’d make his data open and shareable. Scientists are inherently secretive for so many reasons. BUT for the most part, I can’t argue with their concerns.

3. Getting scooped – As I said, social networks trade in data. Science social networks should trade data and ideas, but with that comes the concern of ownership. Who owns the ideas and data? Who is reading them? What’s the guarantee that I’ll get credit for MY data and MY work without someone stealing those ideas and scooping me? I’ve seen a lot of talk about this on a lot of blogs and over at friendfeed. People talk about data and blog timestamps, but when it comes down to it, if someone scoops you and publishes your amazing ideas in Nature, you’re screwed. No one cares about the timestamp and your competitor could easily say they knew nothing about your blog/post/forum and came up with the idea independently. That’s not too far fetched in fields that are oversaturated with researchers.

4. Patents – It actually surprised me that the first person who called me after my website launch was from the University of Iowa legal department. This was definitely something that I hadn’t considered when I developed LabSpaces, but he told me that I should make researchers aware that if they post ideas in a public forum that it may make them ineligible for patenting their products or ideas. It’s sad that this is the state of the law, but patents and intellectual property rights are huge for universities and biotech businesses, so this whole utopian idea of information sharing and idea collaboration in an open forum was doomed to fail from the beginning.

5. Access to Experts – One of the greatest benefits of a University system are your local colleagues. If you’re missing a reagent, you just walk down the hall and get it. If you’re having trouble learning a protocol, you talk to your neighbor/friend who is already doing the technique. What I’m saying is that many researchers already have many instant resources in walking distance. The benefit of an on-line network would be a diversity of information sources, but that doesn’t mean they’re better or faster than what a researcher can get by talking to the guy down the hall that has 15 publications. From personal experience on already established “help forums”, I’ve found the advice and knowledge level to be subpar at best with little understanding of WHY things are done (Sometimes the crowdsourcing crowd is full of morons). If push comes to shove, you can always email experts in the field for advice using publication searches. The expert network has already been curated, at least in the life sciences. Just find the expert and send them an e-mail for their protocol. I myself have done this many times and most researchers are extremely helpful. So why do we need a network or a forum when the experts are already at our fingertips in the literature?

Making Science Social

Given these barriers, what’s the best way to get scientists involved with a social network where the focus is discussing science and collaborating on various projects? You may think that based on my list of barriers and the opinions there-in that I have given up on social networks for science. I guess a part of me has. I don’t see scientists embracing new media or internet social networking en mass to share ideas and collaborate with strangers. It’s just not going to happen from a cultural perspective and from a technological perspective. I do think that the fields which lend themselves well to data sharing and collaboration will have no problem creating niche sites, but there will never be a “Facebook” that all scientists go to on a daily basis. The toolkit required for that to happen would be massive. How could a developer ever satisfy the needs of a geneticist, chemist, physicist and astronomer. I’m getting a headache just thinking about it. Maybe I’m jaded by the last 5 years, but I’m convinced that scientists aren’t interested in using social networks for work purposes (or I’m just not smart enough to figure out how to make this happen).

There has been a lot of talk about how the semantic web is going to revolutionize how we use the internet. I have heard and read so much about this over the past 5 years that I think it’s going to be hard to live up to the hype. From a scientific perspective, the semantic web sounds great. It’ll be a perfect way to discover new connections between people and data. The problem is getting the people and the data involved! And again, many people have spent the last 10+ years trying to socialize science on-line using forums and later social networks with marginal success. I’m not sure that making data semantic is going to revolutionize scientific interaction. We have a tremendous backlog of semantic data (the literature). We even have semantic search engines to probe the literature, and the results are unimpressive at best… I also think the data that lends itself well to semantic interaction is limited. This again points back to the culture of doing science. In summary, I’m not sold on the semantic web or semantic science, but I really hope I’m proven wrong.

Of course there are many different types of social networks out there, so we can’t give up on scientists just yet! I think what we need to do is to trick scientists into using social networks and then once we have them trapped, we can force them to understand the advantages of a web-based social network. I think sites like Mendeley or CiteULike have an advantage here because they have created a product that scientists can use to increase productivity. If they’re interested in making scientists more open to data sharing and collaboration remains to be seen, although I think Mendeley does have facilities in place for paper/manuscript sharing.

My approach to kidnapping scientists is going to be a little different though. I’ve become much more interested in the politics of science and increasing public science literacy. I’d like to see more scientists interacting with the public on a social network to help bring down intellectual barriers. By bringing in bloggers to talk about their science and their daily lives I think we've brought the whole profession of being a scientist down to a much more tangible level and have shown the general public that many of us aren’t self-absorbed elitists. I know that running LabSpaces has been invaluable in helping me to better communicate my science to people who ask me, “What do you do for a living.” I see fewer glossy eyes these days, and I hope to help other scientists break through that barrier by interacting with others in an open forum.

Social Science That Works: Citizen Science

But it’s not all doom and gloom here! Science data isn’t completely closed and we’re slowly seeing cracks in the secrecy armor. I’m really excited about citizen science initiatives because they have a two fold function. They get the general public interested and active in science AND they give a lot of groups access to cool data that would never see the light of day.

-The GalaxyZoo community has made some really cool discoveries by scanning back logged astronomy data.

-FoldIt is a game that lets participants try to find the best folded structure of a protein and researchers have found that gamers are much better at folding proteins than standard algorithms.

You can read more about projects like this or find out about other great science initiatives over at Science for Citizens.

Closing

Finally, I’d like to close by saying that I have high hopes for on-line science social networks. I don’t see the scientific community embracing a single scientific network like a true “FaceBook” for science. The attraction of scientists to a network is not going to be the network itself, that’s apparent, so it’s going to come down to developing a network around tools that help make a scientist’s life easier. The “problem” for developers is that the number of tools that overlap between fields and disciplines is too small for any network to really dominate the science social network landscape which could limit large scale investment in social network based free tools. I do however, see niche networks arising that provide specific tools for certain fields. A great example of this being the high throughput sequencing tool Galaxy out of Penn State