Digital advertising is no longer just a promotional tactic — it is a foundational layer of the modern economy. What began as a way to monetize attention has evolved into a deeply embedded infrastructure that connects content creation, user behavior, and business growth in a seamless, data-driven loop. The tripartite model of advertisers, publishers, and users is now supported by a vast, automated market system that reflects the very architecture of the internet itself.

As user behavior shifts, privacy norms tighten, and new technologies like conversational AI reshape entry points to the digital world, advertising continues to reinvent its role — not by standing apart from the platforms we use, but by integrating into them. Whether in the form of native content, sponsored search results, or monetized chat interfaces, advertising is becoming increasingly ambient, personalized, and intelligent.

And yet, at its core, the principle remains the same: advertising follows attention, and where attention flows, economic infrastructure follows. In this way, digital advertising not only funds the internet — it structures it. As this system continues to evolve, those who understand its mechanics will not just follow the future — they will help build it.

My Blog Post: https://medium.com/@shiyinw/digital-advertising-as-an-economic-infrastructure-0cf02b7a369d

Yes, we know TikTok is on the verge of being banned in the US. But why are #TikTokRefugees flocking to RedNote?

In my latest article, I delve into the Chinese social media platform RedNote (⼩红书), exploring its decentralized approach to content creation. Unlike traditional platforms dominated by influencers, RedNote empowers everyday users with innovative features like impression purchasing and a distinct recommendation system that promotes diverse content and fosters user engagement. I also introduce a novel statistic—the ratio of MAU to the most subscribed account—to compare RedNote's model with previous social media iterations, highlighting its potential to democratize content creation and address the shortcomings of influencer-driven platforms.

My Blog Post: https://medium.com/@shiyinw/%E5%B0%8F%E7%BA%A2%E4%B9%A6rednote-democratizing-content-creation-in-a-post-influencer-world-7ca9b8378f00

The phrase “over-mentored, under-sponsored” comes from my favorite panel at the recent Grace Hopper Celebration, where executive leaders discussed the challenges faced by female technologists. This statement resonated deeply with me because it perfectly captured feelings I had but never fully expressed.

This episode discusses X's recommendation algorithm, which determines the content users see in their timelines. We analyze how Twitter uses data like user interactions, social connections, location, and even phone contacts to personalize recommendations. It relies heavily on graph mining to represent the virtual community as a mirror of the physical world, emphasizing local and popular accounts. There are concerns about information cocoons, where users might be trapped in a bubble of their own predicted interests, potentially limiting exposure to diverse viewpoints. Notably, we further explore how Twitter aims to balance content regulation with freedom of speech, using machine learning to filter inappropriate content while acknowledging potential biases in the training data. Ultimately, the episode praises X's transparency in open-sourcing the algorithm, emphasizing its value for learning about computer architecture and the complex inner workings of social media platforms.

My Blog: https://medium.com/@shiyinw/twitters-recommendation-algorithm-is-now-open-source-what-does-it-tell-us-2d964f79f4e0