Suggest modification to this talk

Title

Description

by John Paul Ashenfelter
Many of the greatest achievements in the history of computers are based on lies, or rather, the strategic sets of lies we generallly call “abstraction”. Operating systems lie to programs about hardware, multitasking systems lie to users about parallelism, Ruby lies to us about how easy it is to tell a CPU what to do… the list goes on and on.
One of the primary “strategic lies” of the internet is the presentation of each service as though it were a discrete, cohesive entity. When we use GitHub, we think of it as just “GitHub”, not a swarm of networked computers. This lie gives us the opportunity to build high availability applications: apps designed to never go down.
Let’s take a tour through the amazing stack of tools that helps us construct high availability applications. We’ll review some of the incredible technology underlying the internet: things like TCP, BGP, and DNS. Then we’ll talk about how these primitives combine into useful patterns at the application level. I hope you’ll leave with not only a renewed appreciation for the core innovations of the internet, but also some practical working knowledge of how to go about building and running a zero-downtime application.

Help us caption & translate this video!

http://amara.org/v/GVgn/

Date

Summarized using AI?

If this talk's summary was generated by AI, please check this box. A "Summarized using AI" badge will be displayed in the summary tab to indicate that the summary was generated using AI.

Show "Summarized using AI" badge on summary page

Summary

Markdown supported

In this video titled 'Learning Statistics Will Save Your Life', John Paul Ashenfelter discusses the importance of statistics for programmers and developers, particularly in the context of improving decision-making and validating assumptions through data. Key points of the talk are as follows:

- **Audience Engagement**: Ashenfelter introduces the topic of statistics humorously, acknowledging that few in the audience have advanced degrees in statistics, which sets a relatable tone for the presentation.
- **Statistics Misuse**: He references a blog post by Said Shaw highlighting how developers misuse statistics to support incorrect conclusions.
- **Descriptive Statistics**: The foundation of his discussion is descriptive statistics, which include concepts like mean, median, and mode. He emphasizes that a single statistic can misrepresent data if not contextualized with measures of dispersion such as standard deviation.
- **Real-World Example**: Using data from his previous job, he illustrates how the average number of badges earned can be misleading without considering the variance among users' achievements.
- **Statistical Errors**: Ashenfelter explains the distinction between type I (false positives) and type II (false negatives) errors in statistical testing, emphasizing the significance of using rigorous methods to avoid incorrect conclusions.
- **Performance Testing and A/B Testing**: He discusses practical applications in software performance testing using statistics and the challenges of A/B testing, particularly the required sample sizes to detect significant changes.
- **Bayesian Statistics**: Ashenfelter introduces Bayesian statistics, which allow data to inform hypothesis testing in a dynamic manner, contrasting it with traditional A/B testing, which can lead to significant delays in decision-making.
- **Fermi Estimation**: He explains Fermi estimation as a method for making quick, reasonable guesses in areas that appear complex, reinforcing statistical reasoning in domestic and business contexts.
- **Conclusion**: The talk culminates in a call for developers to adopt robust statistical practices in their work to make informed decisions and draw correct conclusions from their data. Ashenfelter is thankful for his patrons and provides recommendations for further reading on statistics and machine learning.

Ultimately, the video serves as a practical guide for developers to enhance their understanding of statistics, not only as a theoretical subject but as a vital tool for informed decision-making in their applications.

Cancel