Ricochet is the best place on the internet to discuss the issues of the day, either through commenting on posts or writing your own for our active and dynamic community in a fully moderated environment. In addition, the Ricochet Audio Network offers over 50 original podcasts with new episodes released every day.
Computerization of Everything Makes Everything Worse
I write code. I write code with a purpose: replace humans and free them up from administrative tasks, so they may do other things.
This is a noble pursuit. I free humans from actually doing the administrative work of filing, of confirming that the boxes are properly filled. I make sure tasks are done the same way, the right way, every time.
There is a System. First, the file goes to John, and then John sends it to Chris, who checks against an Excel spreadsheet. After the right notation is made, Chris sends it to Angelina, who edits a row in a different spreadsheet. Angelina goes off and makes a note in the latest report and sends the file (and report) on to John, who sends the file back whence it came with a note at the top: “Processed.”
Let’s imagine there actually is a system that works this way. Wouldn’t it be nicer if a computer did the work instead?
John could receive a file and click a button marked “Process.” The computer checks for problems. The computer notes the relevant Excel spreadsheets, makes changes, and adjusts the reports. All the while, it looks for any problems Chris and Angelina have been trained to flag. Life is good, Chris and Angelina are off doing other things instead of the white-collar equivalent of plunging a toilet.
What if there are any problems? Don’t worry. John’s manager is sure the process is perfectly defined. The person who wrote the “Process” button is similarly certain; nothing can go wrong because every contingency has been planned.
Claire Berlinski https://claireberlinski.substack.com/p/global-eyes-2f3
This Global Eyes will be comparatively brief because I had to write off yesterday and most of today (and formally declare them “Admin days”) because I’ve been engaged in the kind of quest Captain Ahab would recognize to persuade the organs of the French bureaucracy of this fact: J. Claire Berlinski and Claire J. Berlinski are one and the same woman.
I’ve explained now until tears rolled down my cheeks that no, there are not two Berlinski women, sharing the same date of birth and the same apartment, both of them missing a critical year in their tax files, but one, and were you to merge them, voilà!—a single, dutiful, filer of the paperwork, nay, a perfect one!—would emerge. There she is: Exactly the kind of woman one would like to allow to continue to live in France!
What if there are any problems? What if the report Angelina prepares takes the name of the person who submitted the file, and the computer takes that at face value?
If John is still working there, John could probably realize what happened. John could reach in, tweak some knobs, and then the system will continue to work. In this way, the automation is saved from itself.
John won’t be working there forever. John will retire, and now the job is much simpler. You can now get some guy named Gerald just out of college, who has not worked in this office for 30 years. Gerald can press Process just fine, thank you very much. The job, as described, is simple. The system, as described, is perfect.
But one day, someone will have the wrong initial in the wrong place. The system will break. Gerald will click Process, and the system reports a problem. Not with itself, mind you; the system is perfect. The problem is with a tax evader. Gerald knows the system is right, or perhaps he sees that the system made a mistake. Gerald can’t do anything about that though. “I’m sorry, Ma’am. I can’t help you.”
The mortgage system is broken. People who can make rent of $2,000 every month can’t show the federal ratings standards they can afford a mortgage at $1,000. The system as described has very specific things it wants you to demonstrate. Specific documentation, with certain possible alternatives. If you can’t show that documentation, you can’t get a mortgage.
Air travel is not necessarily broken. If you don’t have ID, and can’t get validated in the TSA background check system before you fly, you can pass a 10-30 minute screening when you pass through security. You can be quizzed on your neighborhood, and the TSA official right there gets to decide if you are who you say you are.
I have a noble goal. I free humans from actually doing the administrative work of filing, of confirming that the boxes are properly filled. I make sure tasks are done the same way, the right way, every time. Computerization makes this possible.
Computerization also makes lives worse. Sometimes, the system as described, the system that is encoded in software, is wrong. The day you find out a system is wrong can be years after the system was implemented. The people who can fix the system might never realize the system is broken, or might never be able to fix it.
Look out for computers. Make sure they never become our masters. For many aspects of our lives, we’re already too late.
Published in General
If the database can’t be reconciled with the person listed in the database, wouldn’t it be cheaper to just eliminate the person?
Too right!
How quickly we jettison our old systems for computers. Library card files, phone books, checkbooks. Add to these computerless cars, and cars that don’t have to be charged.
Sometimes the power goes out. Sometimes the power stays out. And more and more things grind to a halt when that happens.
Which one? The one they can still get money from? Or the one that already paid?
wouldn’t it be cheaper to just eliminate the person?
Computers tell the kid at the fast food window how much change to give back to the customer. If it doesn’t work, they have absolutely no idea. Everything stops.
Computers don’t make mistakes. The people programming them did.
Chesterton done said somethin’ ’bout that.
Doesn’t matter. No person can be worth all the trouble it takes to revise the procedures. Eliminate the person.
Among my duties the last several years of my working days was managing some administrative record-keeping processes in a law department of a large corporation. My boss was an enthusiast for trying to off-shore the work (send it to India). In the process of exploring off-shoring, we learned that we had to specify every little detail of the processes – all of the variables and “what ifs.” Which caused me to realize the step from off-shoring to automation was minimal. But, specifying all those variables turned out to be for us an unreasonable amount of effort. It impressed me how many places we found in our processes that looked binary, but actually depended on human judgment.
One of our daughter’s roles in her first job with her math degree (at a business-oriented financial services company) was to develop (computer) models for the company sales agents to use to explain to customers options for reaching certain financial goals. The models were set up for the sales agents to input certain information from the customer in order to run. Since the sales agents were generally not math majors (or even math competent), developing those models required they be “idiot proof.” Our daughter was amazed at how often agents would mess up even simple variable inputs. If you don’t understand what is going on, you won’t recognize if you’re going off the rails.
I’ve noted many times that my late father taught mechanical engineering. He often required students to estimate results before doing detailed calculations, so that the students might recognize when a calculation had gone wrong.
Even when our children were little (30 years ago) we once caused the whole fast food ordering process to stop when we didn’t know that particular chain’s name for its children’s meals, and the person behind the counter could not translate our stated desire for a “child’s meal” into whatever the brand name on his computer button was.
For years I liked patronizing an old school chicken restaurant that had only old mechanical cash registers, so the teenaged counter clerks had to add up on paper the prices for each order, and to calculate the change due (as the place did not take credit cards there was lots of calculating that took place). I always enjoyed watching the teenagers doing arithmetic. The place was torn down about ten years ago.
It is not all bad.
We notice what does not work. Most of the time, things do work.
Part of the problem is that human beings are not given the power to use their judgment to make a change that is needed at their level. The Power is reserved for someone higher up hard to get too. The person the customer talks too has no power.
It is like IT support. The Tier 1 person has to follow a script. It is the Tier 3 people with real power to make changes.
One of my pet peeves of computerizing makes things worse is automobile navigation systems. The available computerized navigation systems do not factor in enough of the variables that matter to me, the driver. If route A is one minute faster than route B, the system directs me to route A but does not tell me if there are more traffic lights or it’s through a dodgy area of town, and many systems don’t tell me when it says “faster route” whether I should expect to save one minute or twenty minutes. It allows me to reject or to use toll roads, but it’s a binary for the entire route.
Driving from my house to my brother’s house (90 miles) I have two sections for which a toll road is an option. One of them is very expensive and at the times we typically go to my brother’s saves us only a couple of minutes, so I don’t use it. But at rush hour the calculation might be different. But there’s currently no way for the navigation system to respond to my preference (which will change day to day) of for example, “only if the toll road will cost less than $10 (the toll varies by time of day and day of week) and saves more than 15 minutes but less than 30 minutes, take the toll road.” Making my own decisions, I usually do not use the toll road in that section because using it saves me only a few minutes at a cost of more than $10. But on another section of the same trip, a different toll road saves me fifteen or twenty minutes at a cost of $5, a tradeoff I consider worthwhile. So in one trip I use one toll road but not another based on my judgment. I know a computerized navigation system could factor all that in, but so far none do.
On the ten hour drive to our daughter’s, I often take a road in one section of the trip that takes 15 minutes longer than its alternative, but is much more scenic, has less traffic, and thus is more enjoyable for me to drive. But that decision is based entirely on my judgment and preferences of the moment. A computer would have a very hard time anticipating and calculating those judgments and preferences.
A local pizza place had a young man at the electronic cash register who apparently didn’t how to use it, or maybe he didn’t have a user account. In any case he would pull out his cell phone to make change. Haven’t seen him in a while…
If you eliminate the flesh and blood person, the identities remain in the computer forever, unbalanced, free-floating, seeking to be united, resentful and angry. I’m not a coder, but I’ve been told this is how computer glitches start on the first place.
“The identities remain in the computer forever” reminds me of a funny learning we had when we switched databases. The computer code field names in the old database never made sense to us, and trying to match them up with field names in the new database was therefore challenging. But finally we learned that the code for the old database had originally been written in French (we had bought the database from a second or third generation developer, so until then we hadn’t known about its French origins). So once we knew that, and with the help of a person who understood the French language, we had much more success navigating the computer code field names. The French in the database lived on long after all the French developers were gone!
The problem I find myself encountering more and more often lately is that the Tier 1 people don’t seem to even know that Tier 2 or Tier 3 even exist, let alone how to get to them.
The differences in timing and tolls etc might seem easy or even obvious, but the computer systems don’t have any of that information unless it is provided to them. Which means someone – perhaps a sizable number of people – doing nothing but keeping track of toll information by hour and day of week and holidays and road construction and toll-both-worker union strikes and… and… and… Well, I hope you get the point.
Mostly, nobody wants to spend that much time and money so that YOU can save time and money.
Yeah. I noticed that Garmin didn’t show the church at which I worshiped. I supposed that it would be good customer service to allow users to submit updates, changes, and additions to their database. But I was wrong.
I expect they get buried with user-submitted updates, much of which is not very useful or worse, and would take too much time and personnel to sort through.
Modern toll paths are a place where computerization helps
Mostly the state.
Sometimes it’s a computer system–sometimes it’s an all-human bureaucracy–and sometimes it’s a combination of the two. See the examples in this post: The Reductio ad Absurdum of Bureaucratic Liberalism.
When I was in the Air Force, I worked a lot on a radio system that had many, many parts. It was hard to make sure that all the pieces were working. Basically the FLR-9 was an elaborate antenna way to ‘see’ 360 degrees, and if an ‘enemy’ signal was heard, two of them across the world could triangulate a location. The signals from the 360-degree array could be blended to ‘focus’ on a specific direction.
It was a massive array of antennas, wires, and boxes that blended signals from the antennas. If one part of it failed you could lose anything from a small area to a significant part of the 360 degree reception. The problem is that our operators were looking for signals that may or may not exist; so if they didn’t hear anything, there was no way to know if a signal was absent or if they couldn’t hear anything from that location.
I proposed a way to rate the performance of each of the FLR-9’s: a computer program could process all the test scores from individual tests of the components and blend them into a performance score – an overall percentage readiness rating for the whole array. We needed to know how well this system was working, right?
My bosses loved that idea (that should have been my first warning) and I started working with a systems development guy to work out the code. Somewhere along that process, I started to re-think the problem. It occurred to me that this score was almost meaningless. We had a FLR-9 at Clark AB in the Phillipines; nothing to the south of them was of much interest. We had one in Misawa, Japan; nothing to their east was of much interest. And there were similar live and dead zones around the world.
It might have been possible to account for this with a weighting system, but the problem was pretty complex already, and the weights could always change. So I put on the brakes. I met with the systems guy weekly but never really advanced the project much. I never added much to assist him with the program. I was about to exit active duty, and I successfully stalled this misbegotten computer project until I left. I’m pretty sure nobody else understood it enough to take up the torch.
So I guess you could say that my final – and maybe best – achievement in my USAF career was to prevent the use of a misbegotten computer analysis.
Oh, and here is an FLR-9. It’s really big. None of them function any more, but this one at Elmendorf is still up as a sort of a museum, last I knew.
Brilliant solution! We need to get down to 1 billion people anyway to make this planet sustainable.
I took a reservoir engineering course for economists from the incomparable Dr. John Lee, chairman of the department at Texas A&M. He required that you be able to model a reservoir with equations on a single piece of paper before you built the massive simulation in the supercomputer. He said if you can’t model on one piece of paper, you have no idea if the outputs from the simulation make any sense.
At least he was creative!
We ran into the problem that you describe while hurriedly driving to a family wedding some 90 minutes away.
I had already scoped out the route. I liked that it was very straightforward.So I knew ahead of time that when we arrived at a complex group of off ramps and overhead hiways and byways that the thing to do was to stay on the straightaway.
But my spouse was riding shot gun and he had just installed the GPS app on his phone. He intended to use it. So exactly at that point where the highway complexity went from 0 to 105, he yelled that I had to take the next exit “Or we won’t make it on time.”
I complied by cutting off another car in the process. Then we went up a ramp and then filtered across an overhead section of road only to zoom down again… Then the next command came from my co pilot to me “We need that exit right there!”
Of course the layout was such that with all the several on and off ramps semi converging almost on top of each other that now the only way to do an exit was to plunge across two lanes of heavy traffic, hoping we didn’t kill ourselves or others in the process.
We then ended up on the same highway we had left prior to the hysterics. But in doing these death defying measures we had saved something like 1/3rd of a mile.
Thanks GPS. Just thanks!
Plus since the Tier 2 and Tier 3 people are smart alecks, they have told the Tier 1 folks to insist to customers that they are also Tier 3!
I wrote a bit about technology in some other post. It was more pro-technology and more anti-human. Still not sure what to think about AI though.
Years ago, i was buying something at a convenience store and the total came to $5.25 I handed the young man a $10 which he entered as ‘amount tendered’ and the register indicated ‘change’ of $4.75. As that was happenung i found a quarter in my pocket which i offered and the fellow insisted that he had to give me the 4.75. He’s probably an economist at the Fed today. Nice kid, though.