Section 1: A Blank Page and a Million Readers - Problem Understanding & Clarifications

Executive Summary: The prompt sounds like the most familiar system on the web. It is really two systems wearing one logo: a durable writing tool for a few hundred thousand writers a day, who trust it never to lose a word, and a read-mostly publishing network for tens of millions of readers, who expect instant pages that Google can find. We decompose the prompt, run a clarification exchange where every answer removes work, write the assumptions down, and meet Maya, whose post runs through all twelve sections.

The interviewer reads it off a card:

"Design a multi-user publishing platform like Medium. Every writer gets a personal publication where they draft, edit and publish posts; readers read those posts, follow the publications they like, and get a feed of new posts from them. Assume tens of millions of readers."

Your first reaction is relief. It's a blog. That relief is the trap, and it has two shapes.

Two Ways to Get It Wrong

The WordPress Install. One database, one app server rendering every page on every request, a cache bolted on at the end "for scale." No thought about where reads land or about the editor. Fine for one writer; not for ten million publications sharing one read path.

The Twitter Clone. The candidate hears "follow" and "feed" and spends forty minutes on fan-out. The feed is real (Section 7's Deep Dive covers it), but it's the side dish. The editor writers live in, and the article page readers land on from search, are the product.

Key Insight: Both failures come from pattern-matching the prompt to a system you already know instead of asking where the work actually is. Decompose first, then ask questions whose answers visibly change the design.