This is terrible - right now the random 17 year old middle-class kid at least has a small chance of getting a somewhat reasonably priced ticket to a popular show. In your model, they have zero chance.
Auction models are good for price discovery but this isn’t a price discovery problem, it is a supply problem. Believe it or not, artists don’t always want to maximize revenue from a ticket, they want fans from lower income brackets to be able to attend as well.
Suppose, for simplicity, that you've got 2 types of people: price-sensitive (your 17yo) and price-insensitive (let's call them rich). In reality it's a gradient, complicated by emotional aspects, but I think just 2 cohorts is sufficient for this explanation.
If face value is constant over time (i.e., the current model), scalpers can buy at any time that original tickets remain available. If they predict huge scalping margins, they'll buy up tickets ASAP, competing with the 17yo buying ASAP. And scalpers are more likely to have bots/scripts to help get in the moment tickets go on sale, putting them at an advantage over the 17yo. The 17yo probably ends up finding that the show has sold out to scalpers, so now tickets are too expensive. If the scalpers overbought, they'll eventually let the tickets go at reasonable prices (maybe even below face value) as the event nears, so maybe the 17yo has a chance that way, and many seats will be empty. If the scalpers underbought, great.
If face value decreases over time (i.e., my proposed model) from the original seller, then you've sort of got the exact same thing going on in terms of the rich buying early and the 17yo buying late, except the 17yo has one less middleman to contend with. Less chaos. Authoritative information about how many seats are yet to be filled. Bots that simply react to slow price changes like a human could, instead of bots that rush the release of tickets faster than humans.
In either model, the rich get their way and the 17yo gets whatever is left. But when well-controlled, this gap can be filled through need-based programs, student programs, etc. -- Broadway has some examples. These programs can be layered on top of, as they are orthogonal to, eliminating the incentive to use bad bots for scalping.
I don't have all the answers, but as someone who has been a musician for 30 years, programmer for 24, FoH audio engineer for 21, stock trader for 15, booked several shows, and buys tickets to shows every month or two, this is something I truly think could benefit the ticket-buying experience without excessive downside. The prices don't need to be astronomical (i.e., for the richest of the rich) when they first go on sale. They just need to be set, and reset continuously over time, so as to have N% be sold equal to N% of sale window elapsed, with the window ending at doors; sales would be almost exclusively to genuine show-goers because scalpers would almost exclusively be bag-holders.
Let's make this easy. There's 1000 people who want to go to a concert and there are 100 tickets. Of the 1000 people who want to go, every 100 of them is willing to spend $10 more, starting at $100. So the first 100 will pay $100, the second will pay $110, etc.
If the concert is priced at $120 there are 200 people completely priced out, but there are 800 people who will pay. The tickets are released, they all scramble for the tickets, some from each of the 8 cohorts willing to pay the price are able to attend.
If the concert is priced the way you described it, all 100 tickets would be bought before the price ever goes below $200.
There is simply more demand than supply. The only way to fairly distribute that, if that is what you want to do, is by the lottery-esque system we have now.
Auction models are good for price discovery but this isn’t a price discovery problem, it is a supply problem. Believe it or not, artists don’t always want to maximize revenue from a ticket, they want fans from lower income brackets to be able to attend as well.