Wednesday, May 30, 2012

To Go Where No Math Has Gone Before


by Brian Peacock

We have many goals for FastFig, but one of the things that gets me most excited is the prospect of putting powerful and often complex math models in the hands of the non-technical crowd.  A little bit of background: 

FastFig will include a sharing system where users can share a model—an equation or snippet of FastFig code—with the world.  Each model will have its own page that will include information about it and a simple inputs-in-answer-back interface.  It doesn't matter how complex the model is, anyone will be able to use it by simply entering values for the different fields and pressing the equals button.  FastFig will spit back an answer.

Why am I so excited about this?  Because it will give EVERYONE access to computations that were once privy only to those technically trained (mathematicians, engineers, scientists, etc.).  What will people do with this new power?  I really can't wait to find out.  Can we use it to save energy?  Or perhaps better understand economic issues?  To visualize population growth?  Or to understand the impact of a of a construction project?  And researchers will be able to share their findings in a way that can be used by anyone NOW.  Using science in ways we never have before.  That's the prospect that keeps me working late into the night.  

Monday, May 14, 2012

What We Scientists Need in a Computational Tool


by Brian Peacock

At FastFig, we are a team of scientists and engineers that, frustrated with the existing computing tools, branched out to create something better.  In this post, we would like to share with you what we, from experience, feel is needed in a 21st century computational tool.  

1.  A Great User Interface - No one has time to waste on a lousy user interface, yet so many programs today have them.  With tomorrow's computational software it should be effortless to get started and help should always be one click away.  That's just good design.

2.  Easy Sharing - What use is creating anything if it can't be shared?  Modern software should be built for sharing whatever the application.  In science, we need to be able to communicate our results to both experts and non-experts alike and we believe software can help.

3.  Cloud Computing - Today we have vast computing resources at our disposal.  Yet, due to the technical complexity of accessing these resources, the majority of us are not able to use them.  Software should automate our access to the cloud.

4.  Social Interaction - It seems everything is social today, why not science?  And we don't simply mean adding 'like' buttons to every page.  We mean true social interaction between scientists at all times.  From validating results to project collaboration, we can make science better through social communication.

5.  Full Web Integration - The most incredible thing about the web is the ability for programs to easily work together by sharing data and computing resources.  The next scientific computing tool should play well with others.

What's important to you?  What do you want?  What do you need?  Let us know and we will build it.

Wednesday, May 9, 2012

Why Science Should be About Users not Readers

12_5_9by Brian Peacock

The way we create and share scientific knowledge has changed very little since the journals Philosophical Transactions of the Royal Society and Journal des S├žavans were first published on a regular basis in 1665.  From then on, nothing was really science unless it was published in a peer reviewed journal.  Today the scientific journal industry has grown into a wide and (some feel unfairly) profitable industry.  With the rise of the internet, little has changed about the way science is shared.  Journals may be online, better indexes have been developed, but at the heart of it all is static content that must be digested by a knowledgeable reader.  Note that I use the word reader.  We do scientific research with the hope of eventually using this knowledge.  The current system is geared towards the science reader and not the science user.  Below are 4 reasons that we need a more user-centric way to share science:

1.  Journals are data poor

Scientific journals have limited space for figures and data tables and typically only include the most significant data.  This is nice for the reader but what about the scientist or engineer who might want to use this data to support another argument?  Some journals do allow online attachments for this kind of content but many do not.  The interested user must therefore contact the author to attain this information which is often a time consuming if not fruitless task.

2.  Data is not easily aggregated

Test and then test again is a mantra of current scientific process.  Complete comparison between similar studies occurs only once someone has decided that it is worth their while to write a review paper summarizing the current thought on a topic.  While there will always be a place for these articles, it should be for describing processes rather than aggregating data.  In user-centric sharing, new data in a field should be tacked on to old data and analyzed as a whole.

3.  Equations and algorithms are not immediately useful

My field is in the area of environmental modeling.  As modelers, we use equations to describe the behavior of a physical system.  With current publishing practices, the model is published as a set of equations that only a skilled technical user can implement.  If the user is lucky, the author of the paper will have shared the code or stand alone program that can be used to solve the model.  However, in many cases, this code is difficult to implement as a part of another model, especially by the nontechnical user.

4.  Data from 'failed' or 'amateur' experiments is neglected

Data is useful in science.  Period.  Provided this data is accurate, the source is irrelevant.  We have two major neglected sources of valuable data: 'failed' experiments and 'amateur' experiments.  Failed experiments are not really failures at all but rather a discovery of what is not true.  The data that is collected in these experiments often goes unpublished despite its potential use in a related study.  Similarly, amateur experiments are rarely published due to a perceived lack of credibility.  There is still inherent value to the results produced by amateurs; the data simply has larger error bars.  For example, environmental science classes all over the world collect water quality data on a regular basis as an educational exercise.  This data is rarely reported despite the value that such data could have to environmental scientists and policy makers.

The internet has great potential to solve many of the problems associated with static journal content.  After all, the internet was first devised as a way to share academic materials.  Let's use technology to build a more user-centric scientific system.