Sunday, May 25, 2008

Science protocols - Recipe for success?

I enjoy cooking and baking, and while I own my share of well-thumbed cookbooks, on a day to day basis I am likely to find my recipes on my favorite cooking websites. There are a number of good ones out there, including Epicurious and Food Network, but the one I go to 99% of the time is AllRecipes.

Why do I prefer this website? For one, the look and feel is inviting, intuitive, and informative. There is no barrier to entry and novice and expert cooks alike will find what they need easily without intimidation or pandering. A nice perk is the ability to search by ingredients, helping you find recipes that will use what you have on hand. But the most important feature is content - the community at Allrecipes is substantial and helpful, not only providing the recipes themselves, but also feedback on the recipes that is often corroborated multiple times. Tips like decreasing the number of eggs or doubling the sauce, roasting at a lower temperature for longer, cutting out the salt, or adding more lime juice can truly be the difference between a successful dish and not.

There has been a lot of discussion on science social networking sites and on whether the promise of "web 2.0" is being delivered in science yet (see David Crotty's post at CSH, Bora's question, and musings at The Scientist over on NN). Some reasons why so-called Science 2.0 hasn't been catching on include the fact that scientists are extremely busy and don't have time to invest in familiarizing themselves with new online networking sites or web tools that have no immediately obvious benefit to them (though some disagree with the claim that scientists are busier than those in other fields). As David points out, much of it boils down to inertia: if we already have a way of doing things, the only way we'll change is if the new way is obviously advantageous and it doesn't take too much effort to adopt it.

It was while reading these related discussions that I started thinking about scientific protocols and how much added benefit could be derived from community content. There are protocol websites out there (OpenWetWare, CSH Protocols, etc) which are a great start, but for the most part these are put up by the original user or published by a journal and rarely generate feedback that could be useful to those looking for a particular protocol - such as slight temperature changes, buffer modifications, or other tweaks that either led to better results or fixed problems. Although it's been a while since I've worked in a wet lab, it seems that a lot of fine optimization goes into a protocol before it produces what is eventually published, and this can often take months to refine.

Given how similar protocols are to recipes, is it that far of a stretch to imagine a protocol version of AllRecipes giving similar benefits? Just as you save time, money, and ingredients by learning from other cooks, you would save time, money, and resources in the lab from other researchers. Granted, this assumes that scientists aren't the type who would say, "What - give other labs a head start by learning from my mistakes? Are you crazy!?" but instead would say, "Think of how much this could help science in general if we all helped each other do experiments more efficiently!" Imagine going to a protocol website, searching by your requirements (protein name, species, type of assay, perhaps), going to the highest rated protocol, and reading a number of reviews that unanimously suggest tweaking one particular step. Or imagine finding the quickest (30-min meals)/most efficient (10 dinners for under $10!)/most popular (95% of people choose this recipe)/best (rated 5 stars by 500 users) protocols for doing X Y or Z as reviewed by scientists like you.

Some might find this kind of crowdsourcing offputting for the scientific domain, others might say it's about time. I know the picture is not so simple, but it just seems silly that we're not benefiting from what other fields (like cooking) have already embraced. Funding is scarce and time is a precious enough resource as it is - why waste both by banging our heads against the same wall others have banged on when we can move forward by finding the door?

I'd be interested to learn if there are actually any protocol websites out there that more fully resemble the types of recipe websites I mention. The solution isn't to create an AllRecipes for protocols (as David mentions in his post) but to provide a service that is useful to scientists and that encourages them to participate. Since the application area is more focused than general science collaboration/networking sites, are the benefits more obvious and will it gain traction more easily?

And for something that is neither here nor there, what is it about science that keeps it from exploiting and embracing the web the way practically everything else has done? (I have inklings but would enjoy hearing others' opinions.)


coturnix said...

Well, JoVE is trying to do something like that, showing videos of how stuff is done. Like a Food Channel for science.

shwu said...

That does look really useful! It's great when people ask questions and are able to get answers (from the "authors" or others), too. I suppose it can get awkward for people to offer suggestions but it would be helpful if, say, things don't work quite the same at higher altitudes, or if someone found that some slight tweak led to significantly better results - or even just to say "make sure you do this part exactly how they say - or else everything fails!" Some words of wisdom from others who have been there and done that to reinforce the best protocol.

Crowdish said...

"what is it about science that keeps it from exploiting and embracing the web the way practically everything else has done?"

While I am not overly familiar with the scientific community, could it be simply that knowledge is power and that many scientists believe that by keeping the sauce secret they will remain separate from the rest of us with opposable thumbs?

Look at some of the effects of the Internet: 10 years ago, people paid $275 commission for a stock trade, now they pay $10. Brokers lost their jobs. Look at the real estate industry today, RE agents try to maintain as much control over the transaction as possible, keeping that 6% commission intact (even though we are in the midst of a massive downturn in real estate values).

Knowledge is power...and a job? Not sure, just my $.02 on the crowdsourcing issues.

Lorrie said...

At OpenWetware we do get interesting comments and discussion on some of our protocols, for example this one:

We eventually hope to build our protocols repository into what you envision: AllRecipes for research protocols. We're also thinking about how to offer helpful feedback, ratings, and search tools for each protocol. I'd be interested in hearing your ideas on what "AllProtocols" would look like on OpenWetWare, and how we can make our current collection better and more useful.

Cameron Neylon said...

Shirley, love the analogy (and I like cooking so that's good as well)

crowdish - I think its rather the reverse, there is no money (or glory) in it for scientists so they can't be bothered. To me the solution is to tip that on its head - if we can't get the reward culture to change (and it is but slowly) then we just have to pay for those contributions.

I would agree with Lorrie that OpenWetWare is a good place to build such a thing, or perhaps rather OpenWetWare would be a good platform on which to build such a thing. Think of the Wiki as the back end document store with a WebApp built on top for the ranking etc.

Another thing which would be good here is something that Kulius Lucks suggested when he and I met with the people from PLoS Comp Biol in Cambridge. Imagine if there was an autolink process between OWW and PLoS papers so that the 'supplmenatry methods' was hosted on OWW. Then imagine an autolink script that checks how many different papers reference this method.

'This methods features in five papers in PLoS ONE and two in PLoS Biology. It is cited in 24 other papers'

That's the real measure of a protocol's success. Whether it is usedin published research. So using that as the criterion could be powerful (and doesn't require people to do the rating - which the y usually don't in science)

Jean-Claude Bradley said...

Just like coturnix, the first thing to come to mind was JoVE and second OpenWetWare. Describing protocols is a major use of OWW.

I also thought of Nature Protocols, but that is not free...Nature Precedings would be an alternative.

To my mind, the biggest hurdle to doing this properly is the significant time burden. In my field, if I go to the trouble of crunching through my lab notebook I might as well write a full paper (unless I'm using JoVE). That may not be the case in fields like molecular biology.

David Crotty said...

It is a bit of a conundrum, and if you come up with the answers, I'd be thrilled. I think one obvious reason why you don't see that many protocol repositories is that people don't automatically think to write up their protocols in a formal manner. The vast majority of what you read in CSH Protocols is commissioned material. Once you have a need for commissioning, you have a need for an editor, and a support staff, and thus you have costs that need to be paid rather than an open site.

You've also got me thinking about the nature of social networks, and whether they're relevant or useful in professional communities. Sure, the kids love their Facebook, and job sites like LinkedIn and SciLink serve a purpose when one is hiring/looking for work. But beyond that, is there a point? What other professions communicate with each other in this manner? More thought needed, I'll probably write more about this on my own blog soon.

shwu said...

@Lorrie and Cameron, that example is a great one - I love that someone included a sample plot! Also the fact that the original author included some background information on the assay can be very helpful. Although the wiki format has its advantages, ideally I think you should be able to do things like collapse sections or see only the X most recent comments, maybe allow tagging of comments so you can categorize by type (is this a comment suggesting a modification, an endorsement or otherwise, asking a question, or something else - like sample data?). Without these filtering mechanisms the pages will quickly become impossible to navigate. We also have to make clear that the rating system is geared towards the protocols themselves and not the original contributors of the protocols.

With the kind of protocol attribution Cameron describe, I think it could really gain some traction. It would be useful even if they wait until after the paper is published. And linking from the paper itself is a great way to get a baseline amount of traffic. Great idea!

@crowdish, it feels to me that scientists fear not the general populace but other scientists. Which is ironic because science also depends on collaboration. For some reason, while we (in the royal sense) can imagine all the benefits we'd glean from data and everything else being open, we balk at making our own work open.

@David, to me Science takes aspects from both types of networking. It can be purely professional but not necessarily goal-oriented and time-restricted the way job searching is; scientists are always looking for answers to questions, people to work with, interesting problems to solve. In this way, a scientific online collaborative community could be indefinitely sustaining, and also useful to its participants, not merely a diversion. In my opinion, science is unique among the professional areas because for the most part it thrives on openness and collaboration but has yet to realize these fully in the internet era. (Some domains are more complicated, such as the medical domain.) It is one of the few places where some type of online networking could have a real impact on the work you do.

David Crotty said...

I think the problem is (and has always been for anyone who has tried in the past to build these community centers) incentivizing people to join in. I can see the incentive for someone asking a question, looking for help, but where's the incentive for the person answering? It's always good to be nice to your fellow man, but it's never going to be much of a priority, particularly in this age of extremely limited funding and job availability.

One thing I should mention, is that CSH Protocols pays a small royalty to authors of protocols (or they can choose an open access option). It's not something we've publicized much, we're waiting to see how it goes as we send out the first year's worth of royalties. It's not a huge amount of money but I know that as a graduate student, I certainly could have used an extra hundred bucks here and there. I'm not sure how much of a difference this will make, but we'll see, and it's certainly an interesting experiment in profitsharing with the community as an alternative publishing model.

bgood said...

Since I saw it presented at the Web2.0 Expo in 2007, I've thought that Instructables would be a fantastic platform to build a scientific recipe book with. They've really got all the pieces assembled. With a little more stodgy-scientist and a little less potato-gun-builder look and feel it would be good to go. If you are in the bay area, you might just drop by, I bet they would would be interested in talking.

rpg said...

I think the problem is (and has always been for anyone who has tried in the past to build these community centers) incentivizing people to join in. I can see the incentive for someone asking a question, looking for help, but where's the incentive for the person answering?

One might make a similar argument about peer review. I think people do it because, consciously or otherwise, they know there's a return on the investment.

So. Hmm. Yes. I help other folks in the lab because I'm that sort of guy, but sub-consciously maybe it is an investment and I expect help in return, whenever I need it?

Jean-Claude Bradley said...

rpg makes a good point - anonymous peer review shouldn't work because the incentives are quite minimal (the main one probably making sure the reviewer is cited :) and it takes time. But it does "work" since publishers can find enough people to do it for free.

David Crotty said...

As far as peer review goes, it's often an arduous process for an editor, and one regularly goes through multiple requests before one finds someone willing to review. The most common response is "I don't have the time right now", which is likely to be the response of most scientists to the heavy demands of social networking.

I'd also make the argument that there are issues with expediency for the person asking the question. Sure, you can throw your problem out there and hope someone happens along who has the answer for you, but you're much more likely to get that answer with a more directed approach. And most people can't sit around for weeks with their experiments on hold hoping someone will solve their problems.

shwu said...

@David, I'm not sure if this would completely solve the expediency problem, but having "supernodes" in such a network (something Cameron brought up during an earlier discussion) - i.e. people who are adept at recognizing appropriate connections and can bring the relevant parties together - might be key. Not everyone has time to monitor the network for people who can help them or who they might help, but if there are some who's "job" it is to do the monitoring then we may very well see the network working on useful timescales.

Of course, there is still the question of "what's in it for them?", but it seems that so-called supernodes often enjoy their role as a connector, and we'd need relatively very few of them, so most likely reward wouldn't be an issue. And it's possible that "Supernode at Big Science Community - enabled hundreds of successful collaborations" could become an acceptable bullet in a CV at some point? ;)