i can no longer recommend personal projects

I used to encourage people learning to code to make personal projects as practice, to not worry about whether it’d be useful, and to publish them even if they’re unfinished to build up a portfolio. That’s essentially how my website came to be. Yet, not many people I know did this, even though they have plenty of time to practice leetcode. I’ve always thought it’s because they feel the need for an idea that is practical and not a waste of time, or they feel their work is too unpolished to publish. Or we got older and less creative.

Nonetheless, nowadays, I’ve 180’d from my position. I can no longer recommend people work on personal projects because I worry they’ll just vibe code them. It feels demoralizing to say this since it’s always fun to see what other people come up with when they’re not so focused on flattering colleges or employers.

Not that it affects me personally. LLMs haven’t really deterred me from pursuing personal projects, though they’ve taken away any hope of them being useful.¹ Perhaps that’s why lately I’ve been working on what I consider to be complete wastes of time, like scraping UCSD’s class schedules—I simply enjoy the mere act of programming by hand, and I care not how useful the end result will be.

Career-wise, I think that human programmers will still be needed in the long term, but way fewer than before, especially after the overhiring spree committed during Covid. After all, there’ll still be industries where code, for safety or legal reasons, must be written by a human. Plus, with CS enrollment dropping,² surely the decrease in SWE supply will be in my favor. I can only hope that I can stay ahead of the curve, which is why I still program manually.³

As for people learning to code, what would I recommend?

• If you haven’t settled on CS yet, then don’t do CS. I feel like software engineering has never been a role deserving of respect. It may pay well—too well—but the industries it works in have often been questionable at best: data brokering, recommendation algorithms, defense, and data centers.⁴ If you started CS now, you’d end up biting the apple of vibe coding and be trapped in a career that’s being eaten away by AI.

• If you’re already committed to CS but can’t find a job, then focus on networking. Blindly mass-applying hasn’t worked in years, and with AI-generated resumes and extensions like Simplify, companies are experiencing a deluge of questionable quality candidates that obscure you in the haystack.

  University is essentially an overpriced networking experience, so use that time to befriend soon-to-be alumni, and hopefully they can drag you into a job. Or, if you’re from the Bay Area, your parents are probably in tech and know people who have founded startups that could take you in.

1. My long term personal projects have been to make a web-based video editor (successor to OpenShit) and Minecraft clone (Doufu). I still want to make the web Minecraft clone, but the video editor I’m less motivated for because of the number of vibe coded web-based video editors that now exist. ↩

2. Except for my university, UCSD! ↩

3. Or maybe I’m just coping. ↩

4. I’m fortunate that my job helps address climate change, but I’m still serving to benefit someone questionable. ↩