Perspectives on Innovation

Hello Readers!  I am writing to tell you about some exciting news from the Client Services group at InnoCentive.

Just last week we launched an exciting new Challenge entitled Predictive Data Analysis. This $100,000 Challenge asks Solvers from all backgrounds to build a predictive model based on a complex dataset.  I know, $100,000 on its own makes this Challenge quite special. But, there is another really cool feature that’s so noteworthy – The Prodigy. This website feature allows enables Solvers to get instant feedback on how well they’re doing in comparison to other Solvers.  Sound exciting? Let me tell you a bit more.

First, a bit of background. The Predictive Data Analysis Challenge asks Solvers to build a model using a large dataset in order to estimate the relative performance of various breeds of an organism.  We have provided the molecular and performance data on 100 breeds of the organism and ask Solvers to estimate the relative performance of an independent set of 150 breeds based on their molecular data.  The Prodigy allows Solvers to upload the relative ordering of the breeds and then it will instantly be compared to the known answer provided to InnoCentive by the Seeker.  After submitting, the Prodigy will provide Solvers with their score, their ranking and if they are within the top 10 best submitted scores thus far, their username will appear on the leader’s table.

This feature is obviously not amenable to all of InnoCentive’s Challenges. However, we think that it will encourage you to continue if you’re on the right track and to go back to the drawing board if you aren’t!  We recommend that you log into the Predictive Data Analysis Challenge and let us know what you think about the Prodigy. We’d love to hear from you in the comments section below.

Happy Solving and good luck.

Gabriel Eichler
InnoCentive Client Services

This is the first in our Blog Series “Help a Solver Succeed” (HASS), where we ask InnoCentive experts to provide resources that they think might be helpful to you in solving Challenges.  Today’s post is from Innovation Development Manager Gabriel Eichler, who is a member of our Client Services team.

Our blogging team has asked me to write a piece for the first issue of the “HASS – Help A Solver Succeed” blog series. This section is dedicated to profiling enabling technologies, services or information that may help our Solvers be more successful at either Solving our problems or be more productive at doing your own work on a daily basis.

Since my educational background and Challenge writing specialty is almost exclusively focused on computational, bioinformatic or statistical Challenges I find it apt to write about a programming language.  I have decided to dedicate my HASS entry to the programming language R.

I came to know R during my PhD research at the US National Cancer Institute.  Previously I had written extensive amounts of code in Matlab – my previous programming language of choice for rapid prototyping or computational experimentation. Though Matlab has a more sophisticated look and feel, and I knew it quite well, I was instructed that learning R would be essential to my graduate studies. Digging in I learned that R was first distributed in the spring of 1997 by Robert Gentlemen and Ross Ihaka and it resembles the closed source, commercial language S in many ways.  However, from the beginning Gentelmen and Ihaka have made R an open source language that thrives off a community of volunteer developers. From nearly the very beginning, R has maintained the Comprehensive R Archive Network (CRAN) resource for everyone to publish their own R extensions or libraries. This brilliant step quickly made R a force to be reckoned with.

I find R to be the best way to quickly model statistical questions, create powerful graphs or even super compute a difficult but parallelizable problem. The interface and kernel are extremely lightweight so your computer is left with maximal resources to compute on what you want.  Beyond that, the CRAN resources make R an even more powerful resource because thousands of people have created hundreds of packages meant to assist you in performing complex tasks.  In fact, in my nearly 3 years of continual use of R, I have rarely (if ever) encountered situations in which I actually had to write complex procedures for any standard statistical or machine learning algorithm.  For example, I was able to develop a multiprocessed, Random Forest based algorithm using mostly code pulled from CRAN.

In summary, I’m a huge fan of the R programming language. If you haven’t already done so I would encourage you to download a free copy and play around with it. I’ll be the first to admit that it’s not as slick as a commercial package such as Matlab or S, but the power of open source has elevated R to be one of the most useful and valuable languages around.  Plus, isn’t it kind of cool to participate in InnoCentive’s Crowdsourcing process by using a resource that is, in and of itself, a product of Crowdsourcing?

Thank you, R.

Gabriel Eichler, PhD.

Giving grades is often cited as the biggest downside of teaching.  In too many cases, it reduces the importance on the knowledge imparted in favor of a contest to see who can repeat the teacher’s words most precisely.

Professor Cathy Davidson of Duke University thinks she’s found a solution:  handing the power over to her students.  Per her blog: “this year, when I teach ‘This Is Your Brain on the Internet,’ I’m trying out a new point system. Do all the work, you get an A. Don’t need an A? Don’t have time to do all the work? No problem. You can aim for and earn a B. There will be a chart. You do the assignment satisfactorily, you get the points. Add up the points, there’s your grade. Clearcut. No guesswork. No second-guessing ‘what the prof wants.’ Clearcut. Student is responsible.”

If the grading students determine that an assignment hasn’t been completed satisfactorily, the student has a chance to resubmit the assignment, for another chance at the points.  If all assignments are deemed satisfactory, the student gets 100 points, for an A in the class.

According to Davidson, every study on peer review among students shows that students perform at a higher level, and with more care, when they know they are being evaluated by their peers than when they know only the teacher and the TA will be grading.

Comments to Davidson’s proposal are mostly supportive, though one raises an example that illustrates a downside – gaming the system.  A professor from Buffalo tried this form of grading and found that 2 groups emerged, one composed of fraternity brothers, the other a group that had self-formed within the class.  These groups each determined that they would vote each other up and the other group down – regardless of the quality of work.  When the teacher intervened, she got complaints of “you set the rules, you can’t change them now.”  To be fair, she was grading on a curve, which she admits may have been a mistake.

What do you think?  Would you trust your peers to grade you fairly?  Can this be done, as long as safeguards are put in place to prevent things from getting personal?

To read more on this story, check out this article in Inside Higher Ed.

BonnieJean Butler is the newest member of our Client Services team, and is managing the Water Problems Affecting People in Developing Countries Challenge.  BonnieJean spent time in India and offers a unique perspective on the impact of this Challenge.

You and I can help save a life. Yes, you and I. And it’s easier than you think.

During my world-wide travels, I saw hard-working moms walk for miles in the hottest, most humid weather you can imagine. The destination? The closest stream. Why? To fill a bowl of water and walk back those very same miles to bring the water to their babies.

Her goal seems like a simple one, but these moms know the water has something in it that may make their children sick and possibly die. They’ve seen before; so many other babies have already died. Maybe this time it will be different. Maybe this stream is cleaner. Maybe my baby is stronger than the others. Maybe not.  So why do it? Simply, their babies will die quicker without water. What a horrible choice.

In some countries, more than 20% of children die before they reach 5 years old and high bacteria content in water is a major contributor. It’s hard to believe or even imagine if you haven’t seen it. People world-wide are dying because of bacteria-filled water.

We wonder, “can’t they just fix it”? Great question! Here’s your chance to help, and it’s easy! Not like “lose 50 lbs in 3 weeks” easy, but really and truly easy. We are seeking to identify these water related issues and you can help identify and solve them. Submit your own ideas or forward this Challenge to your family, friends, or strangers; whatever it takes.

You don’t have to irrigate the Sahara. Smaller incremental changes or improvements are usually more actionable. Get creative with solutions that use limited funding, but create a big benefit for a population. Think about cost efficient solutions can be deployed quickly and have a wide ranging impact.

Please get involved with this Challenge. You really can make a difference and maybe even save the life of a child. So think about it… And pass it on.

BonnieJean

In today’s InnoCentive Insider, Peter Lohse talks about the curious fact that many Challenges are awarded to Solvers with expertise outside of the Challenge field.

It has been more than six years since I joined InnoCentive and I continue to be fascinated by the business model and the success we have with delivering solutions to our Clients. Our success rate overall was around 30% at the end of 2005 and is now quickly approaching the 50% mark (on average, for Theoretical and RTP Challenges). We do not have an empirical basis yet for comparing this outcome with the effectiveness of internally focused solution efforts. However, considering that many of the Seekers had been unsuccessful in finding a solution to these problems on their own, I would say the solution rate is quite spectacular.

Much of the praise for this success goes to our Solvers. They are the brains, experimenters and composers behind the winning proposals. The Client Services Team at InnoCentive is in the privileged position to be a first hand witness to our Solvers’ tremendous creativity.  Each of us has seen hundreds of successful submissions, hence we have a pretty good understanding of how a proposal should be formulated to have the potential for winning an award. These learnings are available to all of our Solvers through the InnoCentive newsletter or through this Blog. For example recent postings from my colleagues Lisa Reinhold, Eugene Ivanov and Michael Albarelli provide valuable insights in this regard.  While we believe that Solvers who follow these guidelines will submit proposals which are more likely to be successful, we recognize that factors other than the form of a Solver submission will have an influence on winning an award. It’s some these other factors that I would like to discuss. (more…)