Thinking Beyond 0's and 1's
327 | Sat 03 Aug 10:45 a.m.–11:30 a.m.
Presented by
-
Nisha is a Software Engineer at Oracle Cloud Infrastructure. When she isn't troubleshooting the cloud, she contributes to various open source Software Bill of Materials (SBOM) projects, most notably SPDX. In their free time, they like to solve human problems by making things.
Abstract
Open Source won. We see it in the large number of software projects created and used by other software projects. Most of our modern day software, including AI, runs on a large number of open source software projects. Working in a cloud company that produces and deploys software at scale, I see a lot of phenomenon that look very much like what I used to see when I worked in semiconductor manufacturing an age ago. Examples of these are drift from the norm, heisenbugs, emergent properties, and just ¯\_(ツ)_/¯ things.
The physical world is full of these types of phenomenon. We deal with it by using probability and statistics - accepting that we can't give a "true" or "false" answer, but settling for a continuous "maybe". This is a talk about looking at software production at a larger scale than just the single artisanal "app". We will apply probability and statistics to open source software at scale, and use some "Machine Learning" to get some insights into how the single app is the product of, and part of a somewhat unknowable whole.
Open Source won. We see it in the large number of software projects created and used by other software projects. Most of our modern day software, including AI, runs on a large number of open source software projects. Working in a cloud company that produces and deploys software at scale, I see a lot of phenomenon that look very much like what I used to see when I worked in semiconductor manufacturing an age ago. Examples of these are drift from the norm, heisenbugs, emergent properties, and just ¯\_(ツ)_/¯ things. The physical world is full of these types of phenomenon. We deal with it by using probability and statistics - accepting that we can't give a "true" or "false" answer, but settling for a continuous "maybe". This is a talk about looking at software production at a larger scale than just the single artisanal "app". We will apply probability and statistics to open source software at scale, and use some "Machine Learning" to get some insights into how the single app is the product of, and part of a somewhat unknowable whole.