A thing we have done for going on the last 25 to 30 years is go to music concerts at the Carnegie Music Hall in Pittsburgh, the one attached to the Carnegie Museums and Library in the Oakland part of town. Lots of non-local name collisions there.
There are a lot of parking options for these buildings, but in a pinch the closest lot is the one behind the museums off of Forbes Ave. What you would do is drive up to the booth, hand the bored student $10 or whatever and then park your car. After the show they leave the gates up because everyone in the lot paid the bored student when they came in. So it would take 5 or 10 minutes to clear the whole space, even if everyone decided to leave at the same time.
This scheme worked fine for everyone, except perhaps some dipshit administrator at the Carnegie Museums who decided that the way to make the organization more profitable would be to do revenue extraction from the parking operation. So they made two classic mistakes.
Fired the bored student in the booth.
Installed a computer to do with student’s job.
So yesterday after a sublime and inspiring concert of symphonic wonders presented by the CMU Philharmonic (side note: if you are after a low pressure and low cost way to begin to navigate classical concert going, these shows are the best thing in town) we spent almost a full half an hour trying to get out of the lot because the parking robot would not let people leave.
Here’s how the system is supposed to work.
Instead handing $10 to the bored student, you get a ticket from the robot.
In tiny little type on the ticket there is some text declaring that you can pay your parking by following the QR code on the ticket. No app required!
The QR code on the ticket takes you to a payment web site built in a late 1990s style where you have to fill in your credit card number and billing address into a tiny little form.
Having pre-paid the ticket, to exit the lot you drive up to the exit robot and, in the dark, put the ticket under the QR code scanner in just the right orientation in order for the robot to let you leave.
Let us review all the things wrong with this scheme.
They have made entering the parking lot fast and exiting the parking lot slow. This is exactly the wrong way around. No one cares if it takes a a minute or two extra to park their car. They care a lot if late at night after a long show they can’t fucking leave the parking lot.
For some reason they think interacting with this god forsaken web interface on a tiny phone screen is a thing that people want to do in the year 2025 instead of handing $10 to a bored student. I have used and programmed computers for more than 40 years and I can assure you that not a single person wants this.
The system does not make it clear that if you do not pre-pay the ticket you will have to sit in front of the exit robot for upwards of 10 minutes trying to pay the ticket either using their shitty web site interface in your phone or using their shitty kiosk interface in front of your face from your car. Either way 25 other annoyed people who just want to fucking leave the parking lot will be breathing down your neck while you try to do this. Again, not a single person wants this.
Finally, even if you paid the ticket the exit robot requires that you present the QR code to its face in a very specific orientation that you must figure out while staring at it in the dark while 25 other annoyed people who just want to fucking leave the parking lot are breathing down your neck. Again, not a single person wants this.
So how did we end up here? While it’s pure and unjustified speculation on my part I think the root cause for this failure is as I said at the top: some dipshit administrator thought this would be a great way to do a bit more revenue extraction for the museums. After all, this is America and in America arts organizations have to make a profit. If the customer here is really the parking lot instead of the people using the parking lot, then it all makes sense.
But let’s say the people who installed this system were actually trying to do a good job, instead of just trying to obtain a few extra pennies of parking lot revenue for their balance sheets. In that context, what went wrong here?
This is a classic case of installing a computing system to automate a manual workflow without understanding what made the manual workflow work in the first place. The number one rule of building computer systems is:
Understand your requirements.
What people want from a parking lot is for payment to be easy so that exit is fast. What this Carnegie Museum system has done is exactly the opposite of this. They have made payment inscrutable so that getting out of the parking lot is hard and slow. This is a pretty fundamental mistake, but is of a sort that is all too common when people do grand projects to automate “simple” tasks. Probably all of the great failures of software engineering happened because ultimately people did not understand the problems they were trying to solve, or through some other misunderstanding spent years solving a problem no one cared about 1.
Sadly this aspect of automated parking payment seems to be universal. Probably this is because the problem that the people running the lots see is “how do we get more money” instead of “how do we make the parking lot not be a hellscape”.
The garage we use downtown also works this way, but is not quite this bad because it’s more clear that you have to pay ahead of time to get out. Mostly this was because there used to be this guy in the lobby of the garage who would spend 45 minutes saying “keep your ticket and pay ahead of time” at the start of every event night. Recently they upgraded the machines in that garage and got rid of that guy in the lobby, but they still have to employ people to help people exit the garage so that it does not take all night on busy event days. The QR code scanner on their exit robots is also a lot more clear than the one at the Carnegie lot. But I doubt this really helps if people are in big trouble (like they forgot to pre-pay).
So the second mistake the Carnegie Museums have made is making this system completely unattended, because they wanted to stop paying the bored students to sit in the booth. When computers are involved the one thing that you know will happen with 100% probability is that there will be confused users who cannot figure out your stupid UI. This is true even for high quality systems, and it’s doubly true for the contract programmed 1990s web page hell-scape that was on offer in this new parking lot system.
Thus, last night, we got to sit in the parking lot and watch vehicle after vehicle get stuck in front of the parking lot exit robot in the dark while 25 other annoyed people who just want to fucking leave the parking lot are breathing down their necks.
No one wants this.
Summary: if you are going to a show at the Carnegie in Oakland, don’t use their parking lot for a while. It’s sort of a shit show right now.
Notes
I was going to add an appendix about how the above mentality is also indicative of the idiocy that is going on in the current LLM bubble. But I will spare you that extra 800 words because it sort of writes itself. And what’s the point anyway.↩︎