A seasoned developer argues that new programmers should build real, personally useful projects rather than generic portfolio exercises like snake clones or toy interpreters. In the LLM era, such artificial projects feel even more pointless since AI can generate them instantly. Instead, the author recommends projects with lasting utility: a personal screenshot hosting server, a CLI email reminder tool, a Discord bot for your friend group, self-hosted game servers, and a lightweight web analytics tracker. The post has a webdev slant, with the author noting that web-based projects have the lowest barrier to reaching real users. It closes with a philosophical reflection on how AI is devaluing code itself, making the 'build something real' argument even more important for motivation and learning.
Sort: