Tuesday, July 31, 2012

Data Scientists Should Be Design Thinkers

World Airline Routes

Every company is looking for that cool data scientist who will come equipped with all the knowledge of data, domain expertise, and algorithms to turn around their business. The inconvenient truth is there are no such data scientists. Mike Loukides discusses the overfocus on tech skills and cites DJ Patil:

But as DJ Patil said in “Building Data Science Teams,” the best data scientists are not statisticians; they come from a wide range of scientific disciplines, including (but not limited to) physics, biology, medicine, and meteorology. Data science teams are full of physicists. The chief scientist of Kaggle, Jeremy Howard, has a degree in philosophy. The key job requirement in data science (as it is in many technical fields) isn’t demonstrated expertise in some narrow set of tools, but curiousity, flexibility, and willingness to learn. And the key obligation of the employer is to give its new hires the tools they need to succeed.
I do agree there's a skill gap, but it is that of "data science" and not of "data scientists." What concerns me more about this skill gap is not the gap itself but the misunderstanding around how to fill it.

There will always be a skill gap when we encounter a new domain or rapidly changing technology that has a promise to help people do something radically different. You can't just create data scientists out of thin air, but if you look at the problem a little differently — perhaps educating people on what the data scientists are actually required to do and have them follow the data science behind it — the solution may not be that far-fetched as it appears to be.

Data scientists, the ones that I am proposing who would practice "data science" should be design thinkers, the ones who practice design thinking. This is why:

Multidisciplinary approach

Design thinking encourages people to work in a multidisciplinary team where each individual team member champions his or her domain to ensure a holistic approach to a solution. To be economically viable, technologically feasible, and desirable by end users summarizes the philosophy behind this approach. Without an effective participation from a broader set of disciplines the data scientists are not likely to be that effective solving the problems they are hired and expected to solve.

Outside-in thinking and encouraging wild ideas

As I have argued before, the data external to a company is far more valuable than the one they internally have since Big Data is an amalgamation of a few trends - data growth of a magnitude or two, external data more valuable than internal data, and shift in computing business models. Big Data is about redefining (yet another design thinking element, referred to as "reframing the problem") what data actually means to you and its power resides in combining and correlating these two data sets.

In my experience in working with customers, this is the biggest challenge. You can't solve a problem with a constrained and an inside-out mindset. This is where we need to encourage wild ideas and help people stretch their imagination without worrying about underlying technical constraints that have created data silos, invariably resulting into organization silos. A multidisciplinary team, by its virtue of people from different domains, is well-suited for this purpose.

What do you do once you have plenty of ideas and a vision of where you want to go? That brings me to this last point.

Rapid prototyping

Rapid prototyping is at the heart of design thinking. One of the common beliefs I often challenge is the overemphasis on perfecting an algorithm. Data is more important than algorithms; getting to an algorithm should be the core focus and not fixating on finding the algorithm. Using the power of technology and design thinking mindset, iterating rapidly on multiple data sets, you are much likely to discover insights based on a good-enough algorithm. This does sound counterintuitive to the people that are trained in designing, perfecting, and practicing complex algorithms, but the underlying technology and tools have shifted the dynamics.

Wednesday, July 18, 2012

Learn To Fail And Fail To Learn




"I have never let my schooling interfere with my education" - Mark Twain

In a casual conversation with a dad of an eight-year old over a little league baseball game on a breezy bay area evening, who also happens to be an elementary school teacher, he told me that teaching cursive writing to kids isn't particularly a bright idea. He said, "it's a dying skill." The only thing he cares about is to teach kids write legibly. He even wonders whether kids would learn typing the same way some of us learned or they would learn tap-typing due to the growing popularity of tablets. He is right.

When the kids still have to go to a "lab" to work on a "computer" while "buffering" is amongst the first ten words of a two-year old's vocabulary, I conclude that the schools haven't managed to keep up their pace with today's reality.

I am a passionate educator. I teach graduate classes and I have worked very hard to ensure that my classes — the content as well as the delivery methods — are designed to prepare students for today's and tomorrow's world. At times, I feel ashamed we haven't managed to change our K-12 system, especially the elementary schools, to prepare kids for the world they would work in.

This is what I want the kids to learn in a school:  

Learn to look for signal in noise:

Today's digital world is full of noise with a very little signal. It's almost an art to comb through this vast ocean of real-time information to make sense out of it. Despite the current generation being digital native the kids are not trained to effectively look for signal in noise. While conceited pundits still debate whether multi-tasking is a good idea or not, in reality the only way to deal with an eternal digital workflow and the associated interactions is to multitask. I want the schools to teach kids differentiate between the tasks that can be accomplished by multitasking and the ones that require their full attention. Telling them not to multitask is no longer an option.

I spend a good chunk of of time reading books, blogs, magazines, papers, and a lot of other stuff. I personally taught myself when to scan and when to read. I also taught myself to read fast. The schools emphasize a lot on developing reading skills early on, but the schools don't teach the kids how to read fast. The schools also don't teach the kids how to scan - look for signal in noise. The reading skills developed by kids early on are solely based on print books. Most kids will stop reading print books as soon as they graduate, or even before that. Their reading skills won't necessarily translate well into digital medium. I want schools to teach the kids when to scan and how to read fast, and most importantly to differentiate between these two based on the context and the content.

Learn to speak multiple languages:

I grew up learning to read, write, and speak three languages fluently. I cannot overemphasize how much it has overall helped me. One of the drawbacks of the US education system is that emphasize on a second or a third language starts very late. I also can't believe it's optional to learn a second language. In this highly globalized economy, why would you settle with just one language? Can you imagine if a very large number of Americans were to speak either Mandarin, Portuguese, Russian, or Hindi? Imagine the impact this country will have.

A recent research has proven that bilinguals have heightened ability to monitor the environment and being able to switch the context. A recent study also proved that bilinguals are more resistant to dementia and other symptoms of Alzheimer's disease.

Learn to fail and fail to learn:

"For our children, everything they will 'know' is wrong – in the sense it won’t be the primary determinant of their success. Everything they can learn anew will matter – forever in their multiple and productive careers." - Rohit Sharma

As my friend Rohit says you actually want to teach kids how to learn. Ability to learn is far more important than what you know because what you know is going to become irrelevant very soon. Our schools are not designed to deal with this. On top of that there is too much emphasis on incentivizing kids at every stage to become perfect. The teachers are not trained to provide constructive feedback to help kids fail fast, iterate, and get better.

Our education system that emphasizes on measuring students based on what and how much they know as opposed to how quickly they can learn what they don't know is counterproductive in serving its own purpose.

Learn to embrace unschooling:


Peter Thiel's 20 under 20 fellowship program has received a good deal of criticism from people who are suggesting that dropping out from a college to pursue entrepreneurship is not a good idea. I really liked the response from one of the fellows of this program, Dale Stephens, where he discusses unschooling. He is also the founder of UnCollege. Unschooling is not about not going to school but it's about not accepting the school as your only option. Lately if you have looked at the education startups, especially my favorite ones — Khan Academy, Coursera, and Codeacademy — you would realize the impact of technology and social networks on radically changing the way people learn. Our schools are neither designed to comprehend this idea nor to embrace it. This is what disruption looks like when students find different ways to compensate for things that they can't get from a school. This trend will not only continue but is likely to accelerate. This is a leading indicator suggesting that we need a change. Education is what has made this country great and it is one of the main reasons why skilled immigrants are attracted to the US. Let's not take it for granted, and let's definitely not lose that advantage.


Originally, I had written this as a guest post for Vijay Vijayasankar's blog

Photo courtesy: BarbaraLN