Caught my ear – March 2021

Desire to avoid more screen time during Covid WFH has shifted most of my reading / learning to podcasts. These are a few conversations and podcasts that caught my ear in the last few months. Will simply list these and associated information I looked up – without my interpretation or meaning making.

Will leave why they caught my ear unsaid …

In July 2020 Amazon kicked off a pilot with Crossover Health for Employee Health centers and an alternate primary care model. This March Amazon announced they are expanding to two more states. This conversation with the founder of Crossover Health Dr. Scott Shreeve gives you a first hand walk through of how they are trying to reimagine Healthcare with a combination of primary care, proactive care and analytics. Crossover Health was founded in 2010 as a solution for self-funded employers to help curb healthcare costs. They’ve built a roster of clients including Apple, Microsoft, Amazon and LinkedIn.

There may be parallels to this from history of healthcare in US. Employer-sponsored health insurance incentives during WW II were at the heart of growth of health insurance system in 40s, 50s and 60s. Structure of the private health insurance system has significant influence over rest of the healthcare industry. These new commercial models may be seeds for a new design for the healthcare system.

Dr. Bob Kaiser visits with Dr. Scott Shreeve, founder, CEO and board member of Crossover Health. Shreeve, an emergency medicine physician, explains how he found a way to scale his vision for a new primary-care model that utilizes technology — an approach often dubbed healthcare 2.0 — to improve access for patients and enhances their quality of life.

The company Shreeve founded now counts Apple, Amazon, Comcast, Facebook LinkedIn and other large employers among his customers.

Click below to listen
The Business of Healthcare Podcast, Episode 81: Healthcare 2.0: A Case Study

Human decision making is fraught with cognitive errors. Decision making in complex systems results in unintended consequences (often due to hidden causal interactions, feedback loops and delayed reactions). These traps are well documented in economics, environmental studies and public policy under systems thinking. One of those is Goodhart’s law.

Goodhart’s law points to the whack-a-mole game we end up playing by picking metrics of interest. You may have heard of this law as “Any observed statistical regularity will tend to collapse once pressure is placed upon it for control purposes” or “When a measure becomes a target, it ceases to be a good measure“.

Hal Ashton, a PhD student, wrote a paper (Causal Campbell-Goodhart’s law and Reinforcement Learning) on whether popular reinforced learning models in AI (the algorithms behind self driving cars and Google Alpha GO) have cognitive errors, like humans. Hal uses a simple decision tree example of a dog in Scotland looking to go for a walk in the right attire – predicting the weather using a manipulatable barometer !! Hal was inspired to write this paper after reading The Book of WHY by Judea Pearl.

Hal Ashton, a PhD student from the University College of London, joins us today to discuss a recent work Causal Campbell-Goodhart’s law and Reinforcement Learning.

Campbell-Goodhart’s law relates to the causal inference error whereby decision-making agents aim to influence variables which are correlated to their goal objective but do not reliably cause it. This is a well known error in Economics and Political Science but not widely labelled in Artificial Intelligence research.

Through a simple example, we show how off-the-shelf deep Reinforcement Learning (RL) algorithms are not necessarily immune to this cognitive error. The off-policy learning method is tricked, whilst the on-policy method is not. The practical implication is that naive application of RL to complex real life problems can result in the same types of policy errors that humans make.

Great care should be taken around understanding the causal model that underpins a solution derived from Reinforcement Learning

Click below to listen
Goodhart’s Law in Reinforcement Learning

Traditional Data Warehouses are being challenged with the emergence of AI / ML. The use cases of AI / ML are creating new demand for complex feature engineering, as well as creating new ways to process incoming difficult or complex data using algorithms (NLP, image analytics etc.). I recently framed this argument in a talk you can watch online (Next-gen data management through the power of Data Science).

Some argue that these needs and capabilities are why Data Lakes arose in the first place and the split of Lakes vs Warehouses is irreversible and irreconcilable. There are others who argue that new technology paradigms always start addressing niche or specialized use cases, but eventually make it back to the base platforms as new features. So relational databases and data warehouses will eventually rule again !

Lakes v. warehouses, analytics v. AI/ML, SQL v. everything else… As the technical capabilities of data lakes and data warehouses converge, are the separate tools and teams that run AI/ML and analytics converging as well?

In this podcast, originally recorded as part of Fivetran’s Modern Data Stack conference, five leaders in data infrastructure debate that question: a16z general partner and pioneer of software defined networking Martin Casado, former CEO of Snowflake Bob Muglia; Michelle Ufford, founder and CEO of Noteable; Tristan Handy, founder of Fishtown Analytics and leader of the open source project dbt; and Fivetran founder George Fraser.

The conversation covers the future of data lakes, the new use cases for the modern data stack, data mesh and whether decentralization of teams and tools is the future, and how low we actually need to go with latency. And while the topic of debate is the modern data stack, the themes and differing perspectives strike at the heart of an even bigger: how does technology evolve in complex enterprise environments? 

click below to listen
The Great Data Debate

We all have scar tissue from bad interviews and hires. There is a strong argument to be made that traditional models of interviewing don’t work. Yet, organizations and many of us do it for the lack of alternative options. Yes, hackathons and coding assignments are an option for technical roles – but even those are limited in evaluating critical success factors such as team work, openness and decisiveness. Adam Grant – Wharton Prof. and organizational psychologist – summarizes research, findings and success in reimagining interviews in his popular podcast WorkLife with Adam Grant. Yes, Adam Grant looks very young. Because he became the youngest tenured professor at Wharton aged 28.

“We’ve all answered our share of cringeworthy interview questions—and watched managers pick the wrong person while rejecting the right one. Is it time to delegate hiring decisions to algorithms?

Find out what the experts recommend, and meet a pair of leaders who have reimagined the interview process by ignoring credentials and refusing to look at resumes.

Reinventing the Job Interview

The next conversation is about two management techniques practiced in Amazon – that we have all heard about – ‘narrative memos in place of PPT‘ and ‘starting any product journey with press release‘. Authors Colin Bryar and Bill Carr talk from their first hand combined experience of 27 years at Amazon about the nuances, back stories and ways to bring these ideas alive in other cultures. Their book Working Backwards: Insights, Stories, and Secrets from Inside Amazon is now in my listening list !!

When you hear stories about Amazon’s “invention machine” — which led to a company with not just one or two products but several successful diverse lines of business — we often hear about things like: Memos, six pages exactly and no powerpoints at all!; or, the idea of just “work backwards from the press release”; and other such “best practices”…

But what’s often lost in hearing about these is the context and the details behind them — the what, the how (as well as their origin stories) — not to mention how they all fit together.

Knowing this can give us insight into  how all companies and leaders, not just Amazon and Bezos, can define their cultures and ways especially as they scale. After all, Amazon was once a small startup, too.

Amazon Narratives: Memos, Working Backwards from Release, More

Share your interpretations and reactions to these podcasts. Keep listening. Will try and post this at least once a quarter.

Published by SridharTuraga

A dad. A partner. A son. A problem solver. A learner. A teacher.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: