If the self-taught candidate can pass an interview screen at the same level of a new grad, then how does hiring the former become more riskier than the latter?

Passing trivia interview screens is useless, can’t tell you anything useful. You have to dig recursively into technical details about past experience or projects, and for university new grads this would be their education work. What will the self-taught candidate do at this point?

> What will the self-taught candidate do at this point?

Generally self taught people will have built something. I'd generally consider that the definition of self taught and would not include people that have only done self directed learning.

That would be fine, but I think it’s weird that you’re trying to make the phrase “self-taught” stretch to include also building something. I think it’s far more common for people to believe taking a self-directed online course and getting a certificate entitles them to equal consideration for a job as someone who put in 4 years of a structured university program.

Maybe I'm out of touch with how it's used today, but self taught used to be someone that built things, W3schools was the closest thing to online courses years ago. If so I'd argue the modern usage is wrong though, a self taught mechanic is someone that tinkers with cars, a self taught electrician is someone there own cables, a self taught sysadmin typically fell into the role by doing stuff, a self taught investor has invested money. So I see know reason why a self taught developer shouldn't be someone that's built software.

I would hope they demonstrated competency for the position. I'm sure there are other avenues other than education work that hold the same weight.

If you can't tell whether someone else is skilled or not, you aren't competent enough to be making decisions.

Well, asking CS trivia isn’t going to tell you they have skills to do the job. What do you propose?

In my decade+ of experience building engineering teams, the only useful signal has been probing technical details of education and past work experience.

If a candidate has neither, they simply cannot emit any signal you could possibly use to decide if they can do the job. Quizzing them on sorting algorithms or devops buzzwords is not going to tell you anything.

...where did I suggest that you should be quizzing people on trivia?

If you need to work out whether someone can do something, ask them to do it. Talk to them about things they have done as a first step. Don't overthink it.

Where I am, no-one asks DS&A questions. Managers still manage to fuck it up by focusing too much on inane questions - a propensity which seems to be inherent to human existence - and discounting just sitting down, asking someone to do something, and then getting them to explain why they did it that way.

> “If you need to work out whether someone can do something, ask them to do it. Talk to them about things they have done as a first step. Don't overthink it.”

I agree very much. It’s why I would be substantially less likely to hire someone for a software engineering job if they don’t either have (a) a somewhat standard bachelor’s degree in a related discipline and communicate clearly about knowledge gained through coursework or (b) previous work experience already delivering software engineering solutions.

If instead they have self-taught academic certification and no professional software experience, it’s too unlikely they can do the job when I ask them about doing it, regardless of their ability to talk about CS topics traditionally associated with tech interviewing.