On Innovation

How to build Data Science teams that foster innovation

“ … the development of the bulb followed the usual path of innovation: incremental, gradual, serendipitous and inexorable”

— Matt Ridley in ‘How Innovation Works’

There may be no definitive recipe to bring about innovation, but a study of history reveals that the eras and locations where innovation flourished, had a few factors in common that stand out:

i. Liberty: Agency for people to think freely and experiment. Innovation seldom happens in a centrally directed and planned manner. Instead, it is a messy, incremental, bottom-up phenomenon.

ii. Networks: Ability for ideas to meet and mate¹ to produce new ones. Ideas come into being as fresh combination of what already exists².

iii. Property Rights: Acknowledgement from society of a person’s intellectual contribution, and a willingness to collaborate³.

iv. Culture: Appetite for risk and a benign attitude towards failure.

There are also other logistical aspects like capital to fund innovation⁴ and sufficient density so that infrastructure scales at a sub-linear rate, while the economic products scale at a super-linear rate⁵. But, they aren’t directly pertinent to this post.

These nation-sized factors can also be incorporated locally at a smaller scale within teams, to foster innovation.

In this post, I list a few ways I have tried to do this. The list is definitely not exhaustive, but has delivered consistently for me.

Also, while any team can implement these, my focus is on Data Science teams (DS) within modern technology organizations, who due to their proximity to the modern day gold — data, are uniquely positioned to bring about innovation. In these organizations DS is (or should be) to Product, what physics is to engineering, what biology is to medicine, or what economics is to policy-making. And, if not for this focus on innovation, they can easily slip into being what accounting is to finance.

We think of innovation as singular eureka moments by genius heroes. In reality, it is a long-drawn, incremental, iterative, evolutionary processes that takes collective inspiration and perspiration of many. The Edison bulb [in pic] is an excellent example of this. PC: https://www.fi.edu/en/history-resources/edisons-lightbulb

***

Steps to foster innovation by DS:

The steps mentioned here are listed in the order a team is built ground up, from sourcing candidates to interviews to periodic team ceremonies.

1. Mention the intent to actively innovate in your MISSION statement

Mission statements are infamous for being vapid words strung together over expensive offsites. But, outside of Dilbert comics, articulating the team’s purpose through a mission can be useful to:

i) Clarify why the team exists

ii) Communicate to leadership, what the team will deliver

iii) Call-out how the team will achieve this vs their adjacent functions

As a DS manager, it is best to emphasize active participation in innovation in the team’s mission statement. This could take the form of:

• “To find efficiencies in system X…”
• “To incorporate intelligence into Y products …”

Instead of something generic, passive, and useless like :

• “To make the team data-driven …”
• “To help inform decisions …”

2. Build a DIVERSE team

To ensure creativity and innovation, the team must be diverse⁶.

A team made of diverse identities will do better than a homogeneous team at solving complex problems because of their wider cognitive repertoire gained through varied life experiences. The improvement in throughput of the team is always big enough to offset any headwinds from the difficulties of working in teams.

Note: This is different to (but not against) the normative and demographic representation arguments for introducing D&I.

Personally, I try to build teams in the image of the Santa Fe Institute⁷. Irrespective of the problems we are asked to solve, my team comprises the usual suspects — computer-scientists, economists, and industrial-engineers; but also — biologists, physicists, political-scientists, MBAs, veterans, ex-entrepreneurs, liberal-arts majors, who speak different languages, grew up in different countries under different circumstances, pursue varied hobbies, and identify themselves in a variety of ways.

A biologist working on customer survival analysis, alongside a political-scientist inferring the true impact of a bug, with a computational-scientist finding the most efficient way to run a simulation — can generate output that is far more powerful than what any of them can do individually.

Sometimes, the breadth of cultures can create uncomfortable situations. But, over a period of time, the benefits go far beyond any short-term adjustment costs.

3. Interview to CHECK for innovation quotient

To ensure you hire innovators, budget >40% of the interview schedule for questions that check curiosity, creativity, and perseverance.

In practice, this could be as simple as asking a broad question and then letting the candidate elicit the problem and potential solutions. Some popular examples are:

• “What features should go into building an algorithm that does [X]?”
• “[X] metric spiked on mm/dd. What do you think happened?”

With each clarifying question, you can open up a new branch or close one, giving you signal on the density of threads the candidate can think of.

With each answer, you can ask ‘What else?’ to stretch their creativity and also get signals on their perseverance.

4. Say NO to bullshit work

Once you have created your dream DS team, you must protect them from the corporate version of resource curse, that comes in the form of:

• “Can you pull me data x for mm/yy? Actually, can you also add z to that and run it again?”

The more common practice many DS succumb to, is to provide answers to specific questions 80–100% of their time, and work after-hours on identifying opportunities (the innovation part). This model of being a martyr is not sustainable and is always bad for business in the long run.

Few strategies to deal with this:

• Map DS individuals to a product or product area vs having a centralized team of consultants.
• Index their performance to the performance of the product vs on the volume of analysis done, questions answered etc. Let the individual DS determine what they work on using this incentive in the marketplace of problems, questions, and opportunities.
• Encourage them to embrace the random ad-hoc asks, since sometimes they may lead to a breakthrough⁸. But, also, support them in being able to say a hard ‘No’.
• Make incremental improvements to self-serve infrastructure vs a perfect oracular system that answers all questions.

5. Encourage the team to USE the product they are building/improving

Innovation happens when one rearranges the world into a new form not seen before. In order to do this, it helps to understand the ways in which the world can be bent through experiential process.

Those who work on physical or digital technology products are at an advantage to make use of this phenomenon. To find a breakthrough, it helps to play around with the system relentlessly and then some more.

Personally, I have been able to find opportunities by experiencing the products I helped build in-situ and then expanding on them in-silico. The teams I led are usually the highest-frequency users of their respective and adjacent products and this has helped them find creative solutions that they may not have thought of without the physical product experience.

6. Create a platform to SHOWCASE and SHARE

Innovation can be a messy process. To provide some order to this non-deterministic phenomenon, it’s important to create some rituals.

I have found success by setting a periodic meeting to ‘pitch’ ideas, where the idea generator is prepared to be quizzed and brings sufficiently convincing arguments for their thesis. This forum helps by:

• Fixing a (mostly non-negotiable) preset time that one can look forward to — to prepare for and showcase one’s ideas or listen to others’ ideas.
• Creates a marketplace of ideas — where supply of ideas is met by demand for new projects, and the currency is one’s reputation as an inventor, presenter, and leader.
• Encourages continuous innovation and generates a healthy pipeline of ideas ready to work on, when the opportune time arrives (for instance, after budget planning).
• Leverages the diverse viewpoints of the team to check the validity of the idea and/or add features to it before it goes in as a candidate for the product roadmap.

There are some risks with this ritual, that can be averted:

• There may not be enough supply of ideas to meet the frequency of the meeting. With time, this can lead to attendees losing interest and the purpose of the meeting losing ground. This can be averted by deciding on the right cadence, trying out with a smaller group before exposing to a larger audience, indexing performance reviews to ideas presented, and pushing to get ideas presented in the forum.
• There are too many [not-useful] viewpoints that hinder the kickstart to the idea. This can be mitigated by inviting the right audience, and compering the meeting towards a clear conclusion.

Similarly, to leverage the diversity of the team, a forum should be setup to share specific skill-sets and methodologies. The objective should be to make most of the team ‘jack of all, and master of (at least) one’.

7. Create the right INCENTIVES

Innovation takes more than invention. If invention is the idea, innovation is the business of turning the idea into something practical, affordable and reliable that people will want to use and acquire⁹. The most successful products coming from Silicon Valley were built on the graveyard of inventions. This success was made possible by innovators who recycled ideas and scaled it.

To keep the innovation engine running, the right behaviors must be differentiated and rewarded.

Behaviors to reward:

• Value innovation and not just invention. It should not be enough to find an insight. DS should be rewarded to market it, convince leaders to fund it, corall a team to work on it, and see it through the goal post.
• Reward for outcome and impact, and not just volume of work done. This is hard to implement because of the time it takes to implement ideas, and the dependencies on others. Yet, using this model roughly, creates the right long-term behavior.
• Fail fast and celebrate attempts. At a country-wide level, the greatest strength of the US is, as a culture not associating any shame to failure. Applying this at the team level results in the same impact.

***

Innovation emerges at the edge of chaos and order. Be it a small team or a large nation, to drive innovation, the leader of the group should focus on creating a compelling vision, a good-enough platform, and the right incentives; then let the individual agents of the system adapt and evolve in an organic bottom-up way.

***

Notes:

1. The phrase ‘meet and mate’ is from ‘How Innovation Works’ by Matt Ridley.
2. The discussion on new ideas standing on the shoulders of their predecessors can be found in ‘The Nature of Technology’ by Brian Arthur.
3. In the book ‘Civilization’, also summarized over a TED talk, Niall Ferguson discusses the power of property rights in development.
4. The book ‘Americana’ by Bhu Srinivasan, goes into great depth about the role of venture capital in the formation and prosperity of the US.
5. In the book ‘Scale’, also summarized over a TED talk, the complex-systems scientist and physicist Geoff West, describes how efficiencies develop with scale in biological, ecological, and also social systems with the neatness of a mathematical equation.
6. In the book ‘Diversity Bonus’, complex-systems and political scientist Scott Page, provides the theoretical derivation to how diversity in teams brings about an additional output (the bonus).
7. The Santa Fe Institute is a multi-disciplinary research institute , who per their mission, are searching for order in the complexity of evolving worlds. You can find more about their history in the book ‘Complexity’ by Mitchell Waldrop.
8. Inserting randomness can help in getting to solutions faster. This is more obvious with algorithms, as mentioned in this Quanta article but can be applied to a broader surface area.
9. Matt Ridley describes the difference between ‘Invention vs Innovation’ in his book (1) and in this interview with Naval.

You may find phrases or themes from the following books in this post:

10. ‘The Rational Optimist’ by Matt Ridley on specialization and exchange.

11. ‘Harnessing Complexity’ by Axelrod & Cohen on the process of evolutionary development process of variation, interaction, and selection.

12. ‘Why Nations Fail?’ by Acemoglu & Robinson on the importance of strong institutions.

13. ‘Saving Capitalism from the Capitalists’ by Rajan & Zingales on embracing competition.

14. ‘Leave Me Alone and I’ll Make You Rich’ or the ‘Bourgeois trilogy’ by Deirdre McCloskey on how liberal values make a prosperous world.