[rachofsunshine]

In the case of people who needed us to vouch that they deserved a job, yeah. But:

- In borderline cases, we didn't want to risk our credibility and companies didn't want to skip tech screens. So competent-but-not-amazing engineers got shut out.

- Our quiz is not perfect, so engineers who didn't fit what we were quizzing got shut out.

- Not everyone needs us to vouch for them. Companies (reasonably) trust that someone with a degree from a top school and years of prestigious experiences can probably do fizzbuzz, and demanding that they prove it was a barrier to them using Triplebyte.

We think it's better to have more granular ability for companies and engineers to decide what mattes to them. A prestigious company can say "we'll only talk to people who took the quiz and got a top score" (and that score is us saying "this person deserves a job"). One that desperately needs the headcount can talk to the borderline cases or decide they don't want to put the extra barrier of a quiz in the way.

Like most markets, we think the hiring market on our platform works best when it's able to respond to local conditions.

[thaumasiotes]

> (and that score is us saying "this person deserves a job")

If you're not willing to refer people to companies based on their score, how can you interpret the score as you saying "this person deserves a job"?

[rachofsunshine]

The difference is in the bar for forcing companies to set aside parts of their screening process. To get them to do that - as we needed to under our old model, because we didn't have a notion of "recommended but without the requirement to skip tech screens" - we needed to be making very strong recommendations.

[thaumasiotes]

I'm a little bemused at the notion of evaluating someone "strong tech skills, but can't pass an interview" and determining that the appropriate recommendation is "talk to this guy, but -- unlike with most of our candidates -- don't skip the tech screens". The recommendation seems like the opposite of the diagnosis.

How does the new ability to make that recommendation address the original problem?

[ianwalter]

Because getting some leads with less certainty is better than getting no leads with higher certainty in a tight market? Wasn't this addressed in the post?