If I understand correctly, they hired someone who seemed to have a good grasp of algorithms but when it came time to code, took forever to write code and took forever to type something out?
Isn't that the exact thing that code interviews are good for? Finding out if someone can write legible, working code quickly?
Code interviews are terrible for finding out if someone can do deep thinking on a very difficult problem over a long period of time, but the exact problem scenario this hire had should have been caught.
I'm not sure how having them write/debug some slightly more complex code in IDEA vs an online Vim approximation is going to suddenly keep someone similar to the 'candidate' referenced from slipping through.
I'd also argue that bootstrapping a project is very different than managing a great code base. I'd be very skeptical of the idea that evaluating someone's toy project tells you how well they'll do in a massive code base.