Yesterday, I faced a situation that any commercial real estate agent or property manager will find very familiar: I needed to make a set of duplicate keys. I had three originals, and needed a number of duplicates.
For me, the usual solution for this task is to head down to the hardware store. There I find the friendly neighborhood key cutter and wait for a bit as he or she selects the right blanks, grinds out the duplicates and then buffs them using a wire wheel, eyeballing for a good match all the way.
But this time, the friendly neighborhood key cutter was a vending machine.
Installed in the doorknob aisle, I found an eight-foot tall, chartreuse-colored robot. A screen on the face detailed the process and a credit card receptacle beckoned. Like it or not, I was about to begin my first robotic, self-service key duplication chore.
We’ll call my original keys A, B and C. A and B were padlock keys, C was a deadbolt key. The A and B originals used the same blank type.
I started with A, a padlock key. I inserted the original into a reader slot on the front of the machine, and it read the key’s teeth and shape. Through a window in the front of the machine, I could see the moving machinery and the tray of blanks it would pick from. A friendly voice and a well-designed software experience guided me through the process, which boiled down to a) insert key b) wait about two minutes for the machinery to read, select the right blank, cut the key, buff it, and drop the completed duplicate into a slot to be grabbed by the user.
Minutes later, I had my first duplicate. It had problems.
Above, you can see the matchup. The silver key in the foreground is the original, the brass colored key behind it is the duplicate. You can see the teeth are aligned, but the blanks are in fact different shapes toward the base of the keys. Here’s an extreme closeup showing the difference:
The blank selected by the robot actually left more key material on the duplicate than there is on the original. I’ve highlighted the difference in red. I expected this would produce a misalignment — even though the teeth appeared accurate, I expected that this duplicate would not travel into the lock deeply enough to work.
I moved on to B, my second padlock key. Remember, B and A are of the same type — they both have the same number stamped on them at the blank factory. I inserted the original into the machine and waited.
The result was interesting. Even though originals between A and B were cut on the same blank, this time, the robot refused to cut anything: it apologized and said on a message on the screen that it could not duplicate this key.
So far, I was 0 for 2.
The third key, C, was for a deadbolt lock. This time, the machine happily got to work, and the duplicate produced looked like a 100% match.
When I got back to the site of the locks, I tried A, and sure enough, it wouldn’t work no matter how much cajoling or graphite lube I added. It was a bust, and it was plainly because the blank selected by the robot was a mismatch — interestingly a mismatch that it appears the machine caught correctly on key B, when it refused to proceed.
Key C worked very well in its deadbolt, bringing us to a 33% success rate for the entire job. Something in the decades of history of using hardware store employees for this task — employees with good eyesight, experience in the trade and attention to detail — tells me that there’s room for improvement here.
One aspect of the job was that the machine produced duplicates that had burrs — small pieces of cut metal that stuck to the duplicate. This is normal in key cutting but should be removed before handing to the customer. I could see through the window inside the machine that a final step did include a wire wheel, but the wheel did not spin, rendering the duplicate messy when delivered. I had to buff the duplicates myself at a nearby workshop to remove these burrs, otherwise I risked inserting pieces of metal debris into my locks.
Was my experience typical? Was this machine out of adjustment or maintenance, or is there a fundamental flaw in the software and/or hardware? I can’t say. I do know that I preferred a living, breathing hardware store employee, usually wearing a vest, skilled in the task. Rarely if ever did these folks turn in a 33% success rate, making what should be a single trip to the hardware store into at least three trips — as well as having to finish the buffing job myself.
I suggest the machine makers work harder on making the key-duplicating part of the robot work as flawlessly as the payment-accepting part of the robot.