I think this group is quite poorly served by our current system.
Adults with a newfound motivation are valuable and they know that. But just saying so isn't enough.
But leaving during undergrad (or just never going to college) then excelling in self-learning and industry seems to put you in a much more awkward position when it comes to continuing onward with education. You are far beyond most undergrad-level courses, but in most cases you need that bachelor's degree to even be considered for grad-level programs--or at least I believe that's the case? You end up needing to waste years of your life and tons of money just to qualify for the courses that would actually teach you something new.
So perhaps the problem is as much that there just aren't compelling tracks offered to people who don't fit neatly into the lines as it is about admissions flexibility.
That is precisely what signalling means. Top colleges pick only does who have signalled being over 99.99% of the population in things like intelligence, conformity and work ethics, then they filter out a bunch of them that never end up their degrees, and now their having their degree signals that you are a top 99.999% worker.
I think the sort that get accepted to these programs also tend to be the sort that would consider just sitting on the beach sipping martinis to be something that'd rapidly become hell. It's just boring. So you'd have a group of ultra highly driven people with 4 years of undirected self study, but also access to pretty much all the knowledge they could ever want in the era of the internet.
It's really impossible to predict, but I do think the results would be interesting and not necessarily what you're implying. In today's era of assumption of education = skill, we often forget things like the Wright Brothers were a highschool grad and a highschool dropout -- yet they effectively invented aeronautical engineering with some skills they picked up operating a bicycle repair shop. There's some major dissonance between what we, collectively, ought be doing today - and what we actually are doing. I think it's a mixture of exponential leaps in entertainment and a general worsening in the condition of entrepreneurship (imagine how laws you'd be breaking by trying to fly your own homemade airplane now a days!) Your idea would at least certainly help combat the latter simply because that's all there'd be to do!
This doesn't explain why a dropout (or even just someone with an acceptance letter but who never joined) isn't worth as much in the labor market as someone who finished the degree.
Ability. You’re talking about ability. Caplan estimates that on average the return from education is ~70% signalling and ability, 30% changes in human capital if I remember right.
Those two seem to me to be vastly different things.
Just because I build a cabinet out of good, knot-free straight boards doesn't mean the cabinet is all about the boards.
How did this person do in high school? Given high school is orders of magnitude more trivial, we can assume they were likely at or near the top of their classes, and probably would have shown remarkable results in skill assessment exams. And with that sort of motivation he also probably would have been involved in immense extracurricular and other such events. This person would likely have been able to get into any university he ever wanted.
Top universities do provide a certainly better than average education, but their main strength has nothing to do with their quality of education. It's the quality of their student body - which turns success stories into a self fulfilling prophecy. Imagine you start an 'basketball school' and only accept people that are at least 6'6", highly athletic, can dunk from x feet, run 100 meters in y seconds, etc. Go figure -- you're going to 'produce' a disproportionately huge number of NBA quality players simply because your admittance is already heavily biased to individuals who are already headed in that direction.
The point of this is that none of this has anything to do with signaling, but it also has very little to do with the quality of education received. The value of an e.g. Princeton degree is that you're the sort of person that could get accepted into Princeton which would be comparable to the sort of person that could get admitted to 'Basketball U'. Regardless of what happens during those 4 years, you're already almost certainly going to be ahead of 99% of the rest of the population. The degree just works as 'proof of filtering'. E.g. even if our basketball university had a pretty bad education system, you'd still see NBA quality players emerging from it at a way way higher rate than the population of non-admitted individuals.