Halloween Edition: Spooky algorithms, poetic visualization and more
Historical bias in algorithms, AI superpowers, poems visualized
Welcome to the 3rd edition of Cross Sections 🎊- a newsletter on data science, data viz & communication.
👻 Spooky Algorithms
Many algorithms are benign but some are spooky. Just a handful of them have set back social equity and encroached on people’s civic lives in the not-so-distant history. Here’re some examples taken from The Weapon of Math Destruction by Cathy O'Neil.
Discriminative stats
In 1896, a statistician at Prudential Life Insurance, Frederick Hoffman published a report with extensive data. He indicated that race strongly associates with life expectancy and that the lives of Black Americans were so precarious that they were uninsurable. The analysis was flawed: 1) it mixed correlation with causation 2) it didn’t stratify the result. Yet it set back racial equality in the US. Until prohibited by law in the 20th century, insurance companies identified risky groups and neighborhoods not to invest. This offers a more detailed view of the racist incident.
Manipulated Opinions
Our opinions on what to buy may be influenced by ads and marketing. Political campaigners use the same strategy. One of such strategies is micro-targeting to send targeted messages to different groups. This was used to spread misinformation. And people don’t know what messages their peers are exposed to, which hinders different groups from having meaningful dialogs.
Tyrannical metrics
There have also been other examples of how teachers’ performances were judged on metrics and those who don’t game the metric lose out at work, and how universities adopted different tactics to rise up the ranking at the expense of higher tuition fees.
📚 Recent Reads
This is a section that captures my reading and reflections
AI Superpowers by Kai-Fu Lee
The author was this book was certainly very inspiring as a scientist/entrepreneur/VC. Instead of writing a summary, I will reflect on a few items instead.
Things I appreciate about the book: it offers a balanced view including an Asian perspective and addresses certain Western misconceptions of Asia. It explained how different environments drive different strategies among tech companies built in the US vs in Asia. Its viewpoint on US internet companies’ attitudes towards and struggles in Asia, such as resistance towards localization, and the career bottleneck experienced by Asian professionals in US companies’ branches, certainly struck a chord, as it reminded me of what I have witnessed. I also appreciate its reflection on the importance of human connection during a machine-powered age through his own cancer-surviving story and caution on potential inequalities created by how AI redistributes wealth.
One thing I don’t fully agree with: the perspective on certain groups as being more accustomed to surveillance - having voice/image data captured - this might be a homogenous view. If such surveillance is already ingrained in the system, just like the body scan as the defacto norm at airports for security reasons, then how are the silent majority gonna resist them?
One thing I didn’t know before reading the book: There is an economic concept called GPT - general-purpose technology - not to be mistaken with Generative Pre-trained Transformer. This concept describes any technology that affects the entire economy, for example, electricity, steam engine, or AI.
🔦 Kaleidoscope
This is an ad-hoc section on curious finds. This week I will bring you on a tour of artistic text visualization.
Poems are hard to translate, and texts are hard to visualize. Both charts and text are condensed ways to encode information. This project, Trees of Translation by Baltazar Pérez, tackles them both by visualizing the writing process of translators working on poems of American poet Emily Dickinson and novel Chilean poet Victoria RamÃrez.
🔖 Re-cap
How Cross Sections was started
If you have any suggestions or feedback, please don’t hesitate to drop a message.