Oct 16, 2014 · 20m · mad
John Rauser, Pinterest // Big Data at Pinterest // Data Driven NYC (Hosted by FirstMark Capital)
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
In this Data Driven NYC presentation, John Rauser demonstrates how software engineers can use basic programming concepts—iteration, logic, and random number generation—to master fundamental statistics without getting bogged down in complex mathematical formulas.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. Matt holds 6.5% of the talking time here. How this is scored →
speaking balance: gold is Matt, purple is the guest (3 minute bins)
When audience member Kevin Keenum gives a statement about Bayesian statistics, Rauser bluntly highlights that there was no question in the comment before giving a brief reaction.
Hardest push from Matt ▶ 16:21 Host pressing for specific recruitment practicesMatt Turck follows up on Rauser's general answer by pressing specifically on how Pinterest finds or trains data scientists from non-traditional backgrounds.
Biggest teaching moment ▶ 8:35 Demystifying classical sampling distributionsRauser illustrates how traditional stats education obscures key intuition behind degrees of freedom and sampling distributions under layers of mathematical formalisms.
Matt holds his own ▶ 14:43 Host steering presentation toward practical engineering hiringMatt Turck immediately grounds Rauser's theoretical talk into real-world business context by asking how Pinterest balances software skills against statistics background during hiring.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | Matt as informed peer | Guest teaching | Guest disagreement | Matt pushing back | Why |
|---|---|---|---|---|---|---|
| Background at AWS and 'Faking It' in Statistics | 0 | 5 | 2 | 0 | In this opening monologue segment, Rauser shares his background at AWS and sets up his thesis by pointing out that many software engineers fake their understanding of statistics. The host does not speak, requiring zero host scores. Rauser gently challenges the audience while introducing the theme of technical self-education. | |
| The Case Study: Beer and Mosquito Attractiveness | 0 | 5 | 1 | 0 | Rauser introduces a concrete case study from PLOS ONE examining whether beer consumption increases mosquito attractiveness. As a monologue presentation, host metrics remain at zero. Rauser smoothly lays out the experimental setup and the basic numerical gap between the test groups. | |
| The Analytical Approach: STAT 101 and Welch's T-Test | 0 | 7 | 3 | 0 | Rauser critiques standard STAT 101 pedagogy by walking through Welch's T-test, degrees of freedom, and sampling distributions. He educates the audience on why conventional mathematical formalisms confuse even experienced practitioners. Because this is a continuation of the presentation monologue, host scores are strictly zero. | |
| The Computational Approach: Random Permutation Test | 0 | 7 | 2 | 0 | Rauser demonstrates how computer iteration and random permutation tests solve the statistical problem intuitively, backing his approach with a citation from R.A. Fisher. The segment remains a pure solo presentation, keeping host scores at zero. Rauser authoritatively argues that programming grants direct access to core statistical ideas. | |
| Concluding Remarks on Programming Superpowers | 3 | 5 | 5 | 2 | The session transitions to Q&A where host Matt Turck asks practical questions about Pinterest hiring and data tools, and an audience member offers a comment on Bayesian stats. Rauser displays mild combativeness by candidly pointing out that the audience comment lacked an actual question. Turck plays a light facilitative role without pressing hard on claims. |