You're rejecting candidates because they didn't reverse a binary tree in 12 minutes. On a whiteboard. While someone watched.
That's not evaluation. That's a hazing ritual with a MacBook.
You want to know if someone can build your CRUD app. So you ask them to implement Dijkstra's algorithm from memory. In a language they don't use. With no IDE.
(This is how we find the best people, apparently.)
Meanwhile, actual job requirements: "Update a button. Ship it before Thursday."
Here's a thought: Have a conversation.
Ask them about something they actually built. Let them explain a decision they made—and why they'd make it differently now. That's how you learn if someone can think. Not whether they memorized Big O notation for the interview and will forget it by lunch.
"But how do I know they're technical enough?"
If you can't tell from a 45-minute conversation about their actual work, the problem isn't them.
Your questions suck.
Ask about a system that broke. Ask what they'd do differently. Ask them to explain something complex to you like you're not technical. (You'll learn more in 10 minutes than any LeetCode gauntlet ever taught you.)
Stop quizzing people on CQRS like it's a personality test.
Stop rejecting senior engineers because they didn't use your favorite framework.
Stop pretending a timed algorithm puzzle predicts job performance. It doesn't. It predicts who practiced timed algorithm puzzles.
That's not hiring. That's pattern-matching on anxiety tolerance.
Talk to people. Like they're people.
Radical, I know.