The Oracle Responds to Kevin Marschke's Rant

Back to list

OP:

*MySQL* - No new project should be using MySQL, you should be using the more feature rich, more updated, and more efficient PostgreSQL instead. Old projects using MySQL should also consider switching to MariaDB.

 *React/Angular* - It's my professional opinion that new webfront projects should use Vue3 instead. Provided, React/Angular technologies are very much supported and not outdated by any means. I used to prefer React and Angular, and avoid Vue2 at all costs. But since Vue3's composition API, its taken reign as my preference. Not only from a personal-use stand point, it's also much more manageable in a team setting as well. I'll note that react native is still my preference with mobile apps, as it's still the most impressive project for that use.

*Node.js backend* - This shouldn't have been used as a backend in the first place. JS was solely engineered to be a webfront FE language: it doesn't have threads, it doesn't have buffers, it doesn't have binary, it doesn't compile. The framework is just to basic for a sophisticated backend. Not to mentioned the over-reliance on external modules to do even the most basic tasks (note: external mods aren't objectively bad). New projects should use Python/C#/Go/Rust instead (depending on need).

*Java* - I expect that Java will continued to be used for decades on projects already using it. But Oracle continues to drop the ball on their tech (MySQL included). No new project should favor Java over the much more supported, more versatile, and more open-sourced C# or JetBrains' Kotlin.

*Windows Server* - I have not met a single developer that prefers windows over linux (that's worked on both). There's too many security issues, too many moving parts, it's not open sourced, not as well engineered (IMO), and not as documented. New projects shouldn't lock themselves into the Windows environment, especially with window's native linux modules that can run linux-based builds either way.

And Sam Altman's baby says:

Criterion	What I Observe (incl. counter-examples)	Recommended Rewording (example)
Tone & Absolutes	Heavy use of “never/shouldn’t” may read as dismissive. Counter-examples: massive, successful stacks on MySQL/MariaDB (WordPress/Wikipedia), Node (Slack/LinkedIn mobile era), Java (banks/Android).	In my experience, PostgreSQL is more reliable/feature-rich for new builds, so it’s my default over MySQL.
Audience Perception	Risks alienating devs constrained by company standards. Many enterprises mandate Java/Windows Server/AD and still succeed.	While many teams thrive on MySQL or Node, my default choices for longevity are PostgreSQL with Go/Python.
Nuance in Trade-offs	Implies silver bullets. Node excels at I/O-bound real-time apps; MySQL is simple/ubiquitous; Java has mature tooling/LTS; Windows Server is required for AD/Exchange in many shops.	Node is fantastic for real-time (chat/streaming). For CPU-heavy services, I lean toward Go or Rust.
Professional Branding	Reads like a rant; hard bans can signal inflexibility. Plenty of orgs succeed with mixed stacks; pragmatism is valued.	These are my defaults, not rules—I still choose tools to fit constraints, team skills, and hosting.
Engagement Strategy	Concludes as a lecture; misses a prompt for dialogue. Questions usually boost reach and quality of replies.	That’s my take—what trade-offs have you seen picking DB, FE, and backend this year?

Tip: Hover to invert colors. Click any cell to copy its text. Click anywhere outside or press ESC to clear selection.

Copied to clipboard