Tweet This
A widely held belief in the software industry is that top-tier programmers are significantly more productive than their counterparts.
These are a few of the notable statements from the past 20 years:
• "Most things in life have a dynamic range in which average to best is at most 2:1. Now, in software, and it used to be the case in hardware, the difference between the average software developer and the best is 50:1, maybe even 100:1." (Steve Jobs, 1995)
• "The variation between programmers is so great that it becomes a difference in kind. In every field, technology magnifies differences in productivity." (Paul Graham, 2004)
• "There are 10x productivity differences among programmers. The studies have collectively involved hundreds of professional programmers across a spectrum of programming activities. Specific differences range from about 5:1 to about 25:1." (Steve McConnell, 2011)
Although the magnitude of difference remains a debate, Silicon Valley recruiters are forever in search of these unicorn engineers.
Data scientists, who combine statistics, programming, and story-telling, display a similar range of talent as that of developers.
LinkedIn is committed to data science because much of the LinkedIn website consists of data products built in collaboration between data scientists and product managers by using profile information, the social graph connecting members and other entities, and website user engagement data to power models or recommendation systems. Examples include the Homepage Feed, PeopleYouMayKnow, Course recommendations, and JobsYouMayBeInterestedIn.
The company also relies on data science when making decisions for the business. An insight, product recommendation, strategy deck, or dashboard is used to improve the business or a product. For example, a 2009 user behavior analysis led to a wholesale change in mobile strategy, shifting from one design for Android, iPhone, and mobile web to the development of separate apps tailored to the unique drivers of each platform. This decision enabled members to more efficiently complete the tasks they wanted to do and was in keeping with LinkedIn's mission to make the world’s professionals more productive.
Yael Garten, Director of Data Science at LinkedIn, shared the following list of behaviors and qualities that distinguishes exceptional data scientists from the average performers.
1. Request context and envision the answer before you start the work. Average data scientists do what they're asked and begin building prematurely when often the real question to answer or data product to build is still not understood. Great data scientists, on the other hand, act as true owners and partners, and solicit more context, asking “assuming I had the answer to that question, what would you do with that information? Assuming I built this data product, what then? How would you use it? What would it look like? Who would it impact? How would it integrate into existing workflows or products or projects or other knowledge?” This probing ensures the question is worth answering and the product is worth building and clarifies the data and methodology and timeline needed to complete the work.
2. Iterate and collaborate. Some data scientists lock themselves away in a hole only to return much later. Great data scientists iterate, soliciting feedback, brainstorming, and breaking work into chunks as they go along. Academics at times work for months with no milestones or peer review. Great data scientists avoid this by iterating, partnering, and sharing, all of which enable them to course correct and ensure the output of the work is integrated because it makes sense, is timely, and is supported by the broader organization.
3. Be proactive. Data scientists have to stand as the voice of the data. Sometimes this is hard. But great data scientists are vocal evangelists who, via excellent work and earned respect, are able to drive change by being proactive and true to the data and the needs of the problem at hand.
4. Solicit help and feedback. Data science is hard, and no one person is great at everything. Top performers identify what they're great at and ask for help in other areas. They improve and others have been able to contribute to creating sound output. It's win-win.
5. Trust their gut on importance, impact, and feasibility. When they're dealing with massive amounts of messy data, multiple objectives, and a fast-paced world with deadlines, the key to impact is having a gut for what will be impactful and for what the best way is of getting to a prototype that validates whether there is something worth pursuing -- before continuing to invest resources.
6. Exude humble self-confidence on how to use the data. A winning combination is a data scientist that blends strong opinions with personal humility. I encourage my data scientists to do what is right for LinkedIn members by having a clear perspective on what action the data is telling us to take, conveying a compelling recommendation to business partners, and at the same time acknowledging the additional considerations, contexts, or priorities that they may be missing.
7. Prioritize. Recent graduates lack the necessary experience to do this well in industry. But any practice in doing this helps, and awareness that this should be a focus area when entering the workforce is essential. Great data scientists ruthlessly prioritize what they spend their time on, find ways to automate whatever can be automated or optimized, and carve out sacred time for innovation, exploration, and opportunity identification.
8. Display technical virtuosity and resourcefulness. Great data scientists are resourceful. They know how to leverage tools, technologies, and methodologies in accurate, rigorous ways. They accomplish their goals in the most context-appropriate ways. To do so, they are flexible with rapidly evolving technologies and know that the level of investment in code and tools will differ depending on the business context.
9. Communicate clearly and persuasively. Clarity of insights is especially key for the subfield of data science that is advanced analytics. Great data scientists recognize their audience, translate a problem and its solution appropriately, and tell the story of the data concisely such that it is delivered as a clear, actionable plan. Skillful data visualization is often integral to this process. Business partners need to trust that the work is correct but don't need the all technical details, caveats, and nuances of the data.
10. Possess grit, passion, and a great can-do-attitude. Vague context, messy data, roadblocks, cross-functional teamwork, tight timelines, competing priorities, rapidly evolving technologies, and demanding clients are commonplace in this field. Business partners expect the “unicorn" statistician-analyst-programmer-hacker-product person to figure it all out, and that's when grit, passion, and a can-do-attitude help make the work exciting and fun.
These habits have helped Garten and her teams have a lasting impact on the product questions they've been asked to answer.
" Much of the value of data science comes via unplanned innovation ," Garten claims. "When data scientists immerse themselves in a problem area, opportunities come from exploring what's possible and what's impactful. That differentiates good from great."
Follow me on Twitter.
