A hiring manager at a decently desirable company posts a job online. Within a week they are likely to get around 50 applicants, at which point they basically stop looking at new ones. Probably 30% of those can be eliminated immediately for not having any relevant experience or coursework whatsoever, having glaring mistakes in their resume, etc. Now the hiring manager still has 35 applicants to evaluate. How do you propose that they narrow the pool in a way that doesn't take weeks of labor? Remember that any multi-round interview, skills test, whiteboarding session, etc. is going to take significant time from the team as well for every single candidate.
This is now an optimization problem. The hiring manager and team have X hours they can allocate to reviewing resumes and interviewing, but X is not nearly enough to properly evaluate the skill level of every single applicant in the pool. We need shortcuts to narrow it to a reasonable level, so what options do we have? Experience is one, so is pedigree (prestige of school, certifications, etc.), maybe we could also look at GitHub commits or spend a maximum of 5 minutes per candidate looking at referenced projects/websites. But we need something that can be evaluated quickly, and experience is an easy proxy since it at least indicates that a person has been around a thing for a while and wasn't so bad at it that they got fired for incompetence. And in many cases that's better to look at than having no indicator at all (i.e., 0 years experience).
It's also worth noting that experience implies more than just technical competence. There is value in having worked on a team in a real-world work setting using skill X that goes beyond how well you can solve technical problems. Experience means you've seen things go right, and also seen them go wrong, you've worked with various kinds of people and personalities while doing X and have the benefit of knowing what to do when you encounter a situation that has come up before.